Introduction to

Industrial Organization,
2nd Ed

Solutions to End-of-Chapter Exercises

Luı́s Cabral

This draft: March 2017

Please address questions or corrections to luis.cabral@nyu.edu

1 Introduction

1.1. Competition and performance.Empirical evidence from a sample of more than 600
UK firms indicates that, controlling for the quantity of inputs (that is, taking into account
the quantity of inputs), firm output is increasing in the number of competitors and
decreasing in market share and industry concentration.1 How do these results relate to the
ideas presented in this chapter?
Answer: In Section 1.2, I argued that one of the implications of market power is the
decline of productive efficiency. Controlling for input levels, the level of output is a
measure of productive efficiency. The number of competitors and the degree of
concentration are measures of the degree of competition (concentration is an inverse
indicator). The empirical evidence from UK firms is therefore consistent with the view
presented in the text.
2 Consumers

2.1. Fruit salad. Adam and Barbara are big fruit salad fans (and both agree that the
more the better). However, their tastes differ regarding the way the salad is made. For
Adam, for each apple you throw in, there should be one and only one banana (if you give
him more than one banana, he will throw it way). For Barbara, as long at it’s fruit, it
doesn’t matter; in other words, all that counts is the number of pieces of fruit.
(a) Show what Adam’s and Barbara’s indifference curves look like.
Answer: Figure 2.1 depicts Adam’s and Barbara’s indifference curves (left and right
panels, respectively).
(b) Are apples and bananas substitutes or complements?
Answer: For Adam, apples and bananas are perfect complements; for Barbara, perfect
substitutes.

2.2. Village microbrew. Village microbrew raised its price from $10 to $12 a case
(wholesale). As a result, sales dropped from 10,500 to 8,100 (in units). Based on your
estimate of the demand elasticity, what percent change in sales would you predict if price
were cut from $10 to $9? What demand level would this correspond to?

Figure 2.1
Indifference curves: Adam and Barbara
bananas bananas

3 U =1 U =2 3

2 ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ......
...
...
2 U =3
...
...
...
...
...
...

1 ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ........
.
...
...
...
...
1
... ..
... .
... ...
... ...
..
... .
...
..
.
.. ... apples apples
1 2 3 1 2 3

3
Answer: We can approximate it by the “change formula,”
  
∆q p 10, 500 − 8, 100 12
≈ = = −1.77
∆p q 10 − 12 8, 100
This is approximate, since we’re using discrete changes. If we assume that the elasticity of
demand is constant then we could get an exact solution by using the log formula:
∆ log q log 10500 − log 8100
= = = −1.42
∆ log p log 10 − log 12
Did revenue rise or fall? Since  < −1, the increase in prices led to an overall fall in
revenue. (If you want to make sure, then calculate the revenues before and after the price
change.) If the elasticity is constant, what is the demand at $9? If the elasticity is constant
then the log formula calculates the elasticity exactly and in addition we know that:
log 10500 − log q9
= −1.42
log 10 − log 9
where q9 is the demand when the price is $9 per case, so (after a little bit of algebraic
manipulation)  
10
q9 = exp log 10500 + 1.42 log = 12195
9
With constant demand elasticity, the percent variation method only gives an
approximation of the value of demand elasticity. Moreover, estimating demand for a
different price level will give a different value than the log formula. Specifically, the
demand estimate when price is $9 is given by
 
q9 = 10500 1 + (−1.77) × (−10%) = 12358

since the drop in price from 10 to 9 corresponds to a −10% variation

2.3. Demand elasticity. Based on the values in Table 3.2, provide an estimate of the
impact on sales revenues of a 10% increase in each product’s price.
Answer: Revenue is given by R = p × q. Differentiating, we get
dR = dp q + p dq
Dividing by R,
dR q p
= dp + dq
R R R
Since R = p q and  = dq /dp p/q, we have
dR q p
= dp + dq
R R R
q p
= dp + dq
pq pq
dp dp p dq
= +
p p q dp
dp dp
= + 
p p
dp

= 1+
p

4
Given that d p/p = 10%, we have

Product dR/R
Milk (1 − 0.5) × 10% = 5%
Cigarettes 5%
Beer 2%
Apples -3%
US luxury cars in US -9%
Foreign luxury cars in US -18%

2.4. Smartphones. The following pairs of price and quantity demanded for smartphones
have been observed: (100, 600); (105, 590); (110, 575); (115, 550); and (120, 510).
(a) Calculate the approximate elasticity of demand when price is
$105.
Answer: Using the percent variation method, an increase in price to 110 leads to an
estimate of elasticity of (575 − 590)/(110 − 105)105/590 = −.53. A decrease in price to
100 leads to an estimate of elasticity of (600 − 590)/(100 − 105)105/590 = −.36. We
conclude that the value of the demand elasticity is somewhere between −.36 and −.53.
Alternatively, we can approach this problem by using logs. The elasticity estimates would
then be (log(575) − log(590))/(log(110) − log(105)) = −.55 for a price increase and
(log(600) − log(590))/(log(100) − log(105)) = −.34 for a price decrease. (Notice that,
while the estimates at each point are quite different, the resulting range is similar.)
(b) Is the demand elasticity constant at all prices?
Answer: Generally speaking, the value of demand elasticity does not need to be constant
at all price levels. In this particular case, it can be shown that it is not. In fact, if the
demand elasticity is constant, then the estimate obtained by using the log formula gives
the exact value of elasticity. As shown in part (a), the estimates obtained with the logs
formula applied to two different pairs of points are different. It follows that the value of 
varies for different price levels. This is consistent with the rather wide range of estimates
we obtained in the previous question.
(c) How does the value of demand elasticity vary as price increases?
Answer: Using logs and computing the elasticity at each point based on a $5 price
increase, we obtain the following estimates, starting at $100 : −.35, −.55, −1.00, −1.77.
This pattern is not infrequent: as we increase price, demand elasticity increases (in
absolute value).
(d) If the monthly subscription fee for Internet access from a cell
phone falls from $10 to $2, what would you expect to happen to the
quantity of cell phones demanded at any given price? What effect
would this Internet access price change have on the mobile phones’
elasticity of demand?
Answer: We would expect the quantity of phones demanded to increase at all price levels

5
when the price of Internet access, a complementary product, decreases. However, it would
be difficult to determine what the change in demand elasticity would be at any given price
point.

2.5. Cars. Table 2.2 gives the “own” and cross-price elasticities for selected automobile
models.2 Specifically, each cell corresponds to the demand elasticity of the car model
listed in the row with respect to changes in the price of the car model listed in column.
(a) Why are the “own” elasticities so high?
Answer: These are models for which many substitute models are available. Thus, even if
the demand for cars is not very elastic, the demand for a particular model is.
(b) Are the Accord and Taurus complements or substitutes?
Answer: The cross-price elasticity of the Accord with respect to the price of the Taurus is
given by 0.1, a positive value. The two models are therefore substitutes. In fact, no two
models in this sample are complements.
(c) What are the Taurus’s closest competitors?
Answer: Looking at the Taurus row, we see that the cross-price elasticity is highest for
the Accord. In other words, a 1% change in the price of the Accord would have a greater
impact on the demand for the Taurus, than a 1% change in the price of any other model
(other than the Taurus).
(d) If GM lowers the price of its Chevy Cavalier, does it “cannibalize”
its Buick Century sales?
Answer: Yes. However, the cross price elasticity of the Century with respect to the price
of the Cavalier is fairly small. Therefore, a decrease in the price of the Cavalier will reduce
the demand for the Century by only a small amount.
(e) Why is the direct elasticity for the Mazda not lower than the
elasticity for more expensive models (as the rule of thumb would sug-
gest)?
Answer: As suggested by the qualitative analysis of demand elasticity, luxuries tend to
have higher elasticity than non-luxuries. However, another rule of thumb to keep in mind
is that the elasticity for a particular product is always higher than the elasticity for the
group of products it belongs to. As it happens, there are many more compact car models
than there are luxury car models. Therefore, even though the elasticity for luxury cars is
higher than the elasticity for compact cars, the elasticity for a particular luxury model
may not be much greater than the elasticity for a particular compact car.
(f) Suppose Honda sold 300k Accords in 2001. In 2002, the price
of the Accord decreased by 2%, whereas the price of the Taurus de-
creased by 3%. What is the likely change in Accord sales?
Answer: The percent change in demand is approximately given by
(−2%) × (−4.8) + (−3%) × (0.1) = 9.3%. We would expect an increase in Accord sales of
approximately 9.3%, or .093 × 300k = 27.9 k units.

6
 2.6. Netflix and Hulu. Suppose the demand for Netflix is given by
qN = a − bN pN + bB pB
where qN is the number of Netflix subscriptions, pN the price of a Netflix plan, and pH the
price of a Hulu plan.
(a) What is the price elasticity of Netflix subscriptions?
Answer: The price elasticity is
d qN pN pN
= = −bN
d pN qN qN
Note the demand is elastic for high values of pN /qN and inelastic for low values of pN /qN .
(b) Suppose a = 500, bN = 10, bH = 5, and pH = pN = 50. What
are N ’s elasticity and cross-price elasticity? Are products N and H
substitutes or complements?
Answer: The cross-elasticity of Netflix with respect to Hulu’s price is given by
d qN pH pH
NH = = bH
d pH qN qN
Substituting the values in the above expressions, we get  = −2, and NH = 1. The positive
sign of the cross-elasticity means the products are substitutes. (You might have guessed
it.)
(c) How much do consumers get in surplus at these prices?
Answer: Consumer surplus is the area under the demand curve but above the price. First
we compute the choke price, that is the highest value of willingness to pay. Since
qN = a − bN pN + bH pH
when a = 500, bN = 10, bH = 5, and pH = pN = 50 we get
qN = 500 − 10 pN + 5 × 50
If qN = 0, then pN = (500 + 250)/10 = 75. This means the choke price is given by 75.
Next we compute total output at the prevailing prices. This is given by
qN = 500 − 10 × 50 + 5 × 50 = 250
Since demand and cost are linear, consumer surplus is a triangle where the height is the
difference between the demand choke price (75) and price (50), whereas the base is output
(250). This implies
1
CS = × 250 × (75 − 50) = 3125
2

2.7. Lamborghini. The current US demand for the Lamborghini Gallardo SE is elastic;
specifically, it is estimated that demand elasticity is given by  = −3. The current price is
p = $120k. Annual sales at this price amount to q = 160 (number of cars).
(a) What do you estimate would be the impact of an increase in price
to $140k?

7
Answer: From the definition of elasticity, we have
q2 − q1 p2 − p1
≈
q1 p1
An increase to 140k represents a 16.6% shift in price. We should thus expect a change in
sales of −3 × 16.6 = −50%. The new level of sales is therefore

q2 = 160 × (1 − 50%) = 80

If instead we use logs, then we solve the equation

log q1 − log q2
=
log p1 − log p2
which in this case becomes (price in thousands of dollars)
log 160 − log q2
−3 =
log 120 − log 140
log q2 = log 160 + 3 (log 120 − log 140) = 4.613
Finally
q2 = exp(4.613) = 100.76
that is, between 100 and 101 cars.
Suppose the cross price elasticity of the demand for Lamborghinis with respect to the price
of the Maserati MC12 is LM = .05; and with respect to the price of gasoline, LG = −.1.
(b) What are the definitions of a substitute and of a complement?
Are Maserati MC12 and gasoline substitutes or complements with
respect to Lamborghinis? Can you think of other substitutes and
complements to the Lamborghini Gallardo SE?
Answer: Product A is a substitute with respect to B if the cross-price elasticity is
positive; and a complement if the cross-price elasticity is negative. Given these definitions,
the Lamborghini and the Maserati are substitute products, whereas the Lamborghini and
gasoline are complements. Additional substitutes might include other luxury models, e.g.,
the Porsche Carrera. Additional complements might include Lamborghini merchandise;
see http://www.lamborghini.co.uk/merchandising/index.php (Question: could this
conceivably be a substitute instead of a complement?)
(c) Suppose that, in addition to the price increase considered in (a),
there is also an increase in the price of the Maserati MC12 (from
$110k to $115k); and an increase in the price of gasoline (from $2 to
$2.8 per gallon). What do you estimate will be the new demand for
the Lamborghini Gallardo SE?
Answer: The price of the Maserati MC12 increases from $110k to $115k. That’s a 4.545%
increase in price. Since the cross-price elasticity is LM = .05, we get a percent increase in
demand of (.05) × 4.545% = .227%. Finally, new demand is

q2 = 160 × (1 + .227%) = 160.3632

8
(We thus have a very small impact, at most one extra car sold.) As to the price of
gasoline, we would have
 
2.8 − 2
q2 = 160 × 1 + (−.1) × = 153.6
2

If we want to consider the compound effect of all three changes, then we add the three
effects:
     !
140 − 120 115 − 110 2.8 − 2
q2 = 160× 1 + (−3) × + (.05) × + (−.1) × = 73.96
120 110 2

Let us now consider the solution with logarithms. From the definition of elasticity, we have

log q2 − log q1 =  (log p2 − log p1 )

This expression is valid both for a change in the own price (in which case  is the own
elasticity) and for the change in the price of a different product (in which case  is the
cross-price elasticity). And when there are several price changes, we simply add the
various elasticity-times-difference-in-log-price terms. Specifically, in this case we have

log q2 −log 160 = −3×(log 140−log 120)+(.05)×(log 115−log 110)−(.1)×(log 2.8−log 2.0)

which implies
log q2 = 4.581
or simply
q2 = exp(4.581) = 97.612
that is, between 97 and 98 cars.

2.8. Constant elasticity demand. Linear demand curves have constant slope, that is,
constant derivative dq /dp. Consider now a demand curve with constant elasticity.
(a) Show that such demand curve has the form q = α pβ (that is,
show that, if the demand curve has this form, then demand elasticity
is constant).
Answer: As shown in the text, demand elasticity may be written as
d log q
=
d log p
If the value of  is independent of p and q, then it must be that log q is liner in log p, that
is,
log q = a + b log p
Applying the exponential function to both sides of the equation, we get

q = ea pb

which is the desired expression, where β = b and α = ea .

9
 (b) Consider two points from the demand curve, (q1 , p1 ) and (q2 , p2 ).
Show that the expression ∆ log q/∆ log p gives the exact value of the
demand elasticity.
Answer: Since
log q = a + b log p
we have
log q1 = a + b log p1
log q2 = a + b log p2

Taking differences, we get

log q1 − log q2 ∆ log q
=b= =
log p1 − log p2 ∆ log p
3 Firms

3.1. DRAM factory. You own and operate a facility located in Taiwan that
manufactures 64-megabit dynamic random-access memory chips (DRAMs) for personal
computers (PCs). One year ago you acquired the land for this facility for $2 million, and
used $3 million of your own money to finance the plant and equipment needed for DRAM
manufacturing. Your facility has a maximum capacity of 10 million chips per year. Your
cost of funds is 10% per year for either borrowing and investing. You could sell the land,
plant and equipment today for $8 million; you estimate that the land, plant, and
equipment will gain 6% in value over the coming year. (Use a one-year planning horizon
for this problem.)
In addition to the cost of land, plant, and equipment, you incur various operating
expenses associated with DRAM production, such as energy, labor, raw materials, and
packaging. Experience shows that these costs are $4 per chip, regardless of the number of
chips produced during the year. In addition, producing DRAMs will cause you to incur
fixed costs of $500k per year for items such as security, legal, and utilities.
(a) What is your cost function, C(q), where q is the number of chips
produced during the year?
Answer: The $5 million you originally spent for the land, plant, and equipment is a sunk
expenditure and thus not an economic cost. However, there is a “user cost of capital”
associated with the land, plant and equipment, based on its current market value of $8
million and your cost of funds and the rate of depreciation or appreciation of the asset
over the planning horizon. Your (opportunity) cost of investing $8 million for one year is
$800k, but these assets will appreciate by $480k over the year, giving a (net) user cost of
capital of $320k. (The depreciation rate of capital is 6%.) This is a fixed cost of making
DRAM’s, to which we must add the other fixed costs of $500k to get a combined fixed
cost of $820k for the year. The variable costs are a constant $4 per chip, so the cost
function is C(Q) = 820k + 4 Q, in the range of 0 < Q < 1m. (One could also report that
C(0) = 0, by definition, and that C(Q) is infinite for Q > 1m, since your maximum
capacity is one million chips per year. Of course, in practice there would likely be a way to
push production beyond “rated capacity,” at some cost penalty, but that is beyond the
scope of this problem.)
Assume now that you can sell as many chips as you make at the going market price per

11
chip of p.
(b) What is the minimum price, p, at which you would find it prof-
itable to produce DRAMs during the coming year?
Answer: The average cost function is AC (Q) = 820k/Q + 4, again up to one million chips
per year. This declines with Q, so the minimum AC is achieved at full capacity
utilization. At one million chips per year, the fixed costs come to $0.82 per chip, so
average costs are $4.82 per chip. This is your minimum average cost, and thus the
minimum price at which is makes sense to stay open for the year.

3.2. mp34u. Music Ventures sells a very popular mp3 player, the mp34u. The firm
currently sells one million units for a price of $100 each. Marginal cost is estimated to be
constant at $40, whereas average cost (at the output level of one million units) is $90. The
firm estimates that its demand elasticity (at the current price level) is approximately -2.
Should the firm raise price, lower price, or leave price unchanged? Explain.
1
Answer: Optimal pricing implies m = || . In this problem, we have

p − MC 100 − 40
m= = = 0.6,
p 100

which is greater than 1/| − 2| = .5. This tells us that the price/cost margin is too high, so
a lower price would be optimal. Note that the margin depends on MC , not AC .

3.3. KindOfBlue jeans. Two years ago, KindOfBlue jeans were priced at $72 and
121,000 units were sold. Last year, the price was lowered to $68 and sales increased to
132,000.
(a) Estimate the value of the demand elasticity.
Answer: Using the percent variation method (and measuring q in thousands), we have
132−121
121
= 68−72 ≈ −1.64
72

Using the logarithm method, we have

ln 132 − ln 121
= ≈ −1.52
ln 68 − ln 72

(b) Based on your estimate of the demand elasticity, how many units
would you expect to be sold if price were lowered by an additional $1?
Answer: Let q be the new value of q. Continuing with the first method, I would estimate
q − 132 67 − 68
= −1.636 ×
132
 68
67 − 68
q = 132000 × 1 − 1.636 × ≈ 135, 177
68

12
Using the logarithm method,


ln(q) − ln(132) = −1.522 × ln(67) − ln(68)


q = exp ln(132) − 1.522 × ln(67) − ln(68) ≈ 135, 011

(c) In order to increase profits, should price be lowered below $68?

If your answer begins — as it should! — with “it depends,” indicate
as clearly as possible what additional information you would need and
how you would base your answer on such additional information.
Answer: It depends on what the value of marginal cost is. To be more precise, if marginal
cost is lower than marginal revenue, then we should further increase output, that is,
decrease price. Marginal revenue is given by
 
1
MR = p 1 +


Depending on which value of elasticity we use, we get MR ≈ 26.5 or MR ≈ 23.3 (when

p = $68). So, if marginal cost is lower than 23.3 we should further decrease price. If
marginal cost is greater than 26.5, we should instead increase price. Otherwise, it is not
clear.

3.4. EZjoint. After spending 10 years and $1.5 billion, you have finally gotten Food and
Drug Administration (FDA) approval to sell your new patented wonder drug, which
reduces the aches and pains associated with aging joints. You will market this drug under
the brand name of EZjoint. Market research indicates that the demand elasticity for
EZjoint is -1.25 (at all points on the demand curve). You estimate the marginal cost of
manufacturing and selling one more dose of EZjoint is $1.
(a) What is the profit-maximizing price per dose of EZjoint?
Answer: The optimal pricing rule (one version of it) states that
MC
p= 1
1+ 

where  is the price elasticity of demand. We thus have

1
p= 1 =5
1 − 1.25

Note that R&D expenditures are a sunk cost and thus do not enter into the pricing
decision.
(b) Would you expect the elasticity of demand you face for EZjoint
to rise or fall when your patent expires?
Answer: The level of demand for EZJoint must fall now that there are many very close
substitutes in the form of generic versions. Hopefully, your brand will still allow you to
command a premium price, but surely at any given price you will sell less as a result of
the presence of the generic competition.

13
 The elasticity of demand for EZjoint will very likely rise now that closer substitutes
are available. Customers will presumably be more price sensitive, which in turn induces
you to set a lower price. However, to the extent that the price elasticity is lower at higher
price levels of the demand curve; and to the extent that competitions from generics forces
the branded drug firm to specialize on the high-valuation segment; then it is possible that
the elasticity faced by the former monopolist now be less elastic.
Suppose that, after patent expiry, a generic version of EZjoint was introduced in the
market (under the chemical name clorophospartane). Reacting to entry, EZjoint decided
to increase price.
(c) Can this behavior be consistent with rational profit maximizing?
Answer: One possible explanation is that the market is divided into two segments, one
more price sensitive than the other. While a monopolist, EZjoint was setting a price that
balanced the goal of achieving both market segments. Once entry by a generic
manufacturer has taken place, EZjoint decided to “give up” on the price-sensitive segment
and target exclusively the segment with more inelastic demand — thus the price increase.

3.5. Las-O-Vision. Las-O-Vision is the sole producer of holographic TVs, 3DTVs. The
weekly demand for 3DTVs is D(p) = 10200 − 100 p. The cost of producing q 3DTVs per
week is q 2 /2 (note this implies that MC = q).
(a) What is Las-O-Vision’s total revenue schedule?
Answer: Total Revenue is given by p(q) q, that is, the revenue that Las-O-Vision receives
when it sells q units. To get p(q), we invert the demand function q = 10.2k − 100 p by
solving for p in terms of q, or p(q) = 102 − q/100. Substituting this into our total revenue
equation, we obtain TR (q) = (102 − q/100) q = 102 q − q 2 /100.
(b) What is Las-O-Vision’s marginal revenue schedule?
Answer: Marginal revenue is the derivative of Total Revenue with respect to q, so
MR (q) = 102 − q/50; or, since our demand equation is linear in q, we can obtain it by
recalling that the marginal revenue curve is twice as steep as the inverse demand curve
and starts at the same point on the vertical axis.
(c) What is the profit-maximizing number of 3DTVs for Las-O-
Vision to produce each day? What price does Las-O-Vision charge
per 3DTV? What is its daily profit?
Answer: The profit maximizing quantity, q ∗ , is that quantity at which marginal cost and
marginal revenue are equal. Setting MR (q) = MC , we have 102 − q ∗ /50 = q ∗ , or q ∗ = 100.
The profit maximizing price is that which generates q ∗ = 100 in sales or, substituting into
the inverse demand function calculated in (a), p(100) = 102 − (100/100) = 101. When
selling 100 units, Las-O-Vision generates total revenues equal to
TR (100) = 102 × 100 − 1002/100 = $10.1k. Its total cost is 1002 /2 = 5k. Therefore its
total profit when it sells 100 units is 10.1k − 5k = $5.1k.

14
Table 3.1
Monsanto’s Roundup: price, sales, revenue and profit
Year Price Quantity Revenue Profit
1995 45 13 599 400
1996 44 16 718 473
1997 40 20 823 517
1998 35 28 1003 578
1999 33 33 1092 592
2000 28 40 1105 509
2001 25 45 1139 464

Table 3.2
Monsanto’s Roundup: demand elasticity
Year Elasticity
1996 -7.13
1997 -2.59
1998 -2.51
1999 -2.11
2000 -1.07
2001 -1.32

3.6. Monsanto. With reference to the discussion in Box 3.1:

(a) How do you know that cutting the price of Roundup was a good
idea for Monsanto?
Answer: The key issue here is elasticity: Does a lower price produce greater revenue and
profit? The case suggests it does. In order to estimate whether profit increased when price
fell, you can guesstimate revenue (take price and quantity from the figures) and cost (the
international price is an upper bound). Almost any estimate will show profit going up.
Table 3.1 is based on a marginal cost of 15, which is probably on the high side (and thus
understates the increase in profit). A more specific, quantitative answer is given in the
next answer
(b) How might you estimate the elasticity of demand and the profit-
maximizing price for 1995. Do you think Monsanto set the right price?
Answer: The first issue to address is that the price drop is not a controlled experiment:
other things might have been changing at the same time. Did the market for herbicides
overall grow? We don’t know. Once you’ve made this point, you can think about
estimating the value of elasticity by applying the ratio of variations method for
year-on-year variations. Some estimates can be found in Table 3.2.

15
 If you take 2.5 as a ballpark estimate, the ratio of price to marginal cost should be
1.67, so with a cost of 15 the price should be about 25, that is, lower than what Monsanto
chose. There is an important qualification to this estimate of optimal price: One
possibility is that there is some form of price discrimination going on here. At the time
this case takes place, Monsanto was also selling Roundup Ready seeds, which is a
complementary product to Roundup. If seed buyers were willing to pay more than other
users of Roundup, Monsanto could target both markets by charging less for Roundup and
more for the seeds. Some observers suggest this is precisely what Monsanto did. More on
this in Chapter 6.
(c) If cutting price was a good idea, why didn’t Monsanto do it
earlier?
Answer: While we have data looking at the real effect of the price cut, Monsanto was
working off estimates, which may have suggested that what they were doing was optimal
given the information available. Even if Monsanto had an estimate of demand elasticity
similar to the ones we computed, there are several considerations to take into account.
First, it may have been the case that Monsanto’s executives had never thought about
changing prices (this possibility is not very appealing to an economist, who always
assumes that firms maximize profits). Second, we know that success breeds laziness: while
their product was successful and profits high Monsanto’s executives had little incentive to
work harder. Finally, one plausible explanation for waiting until the mid-1990s is risk:
experimenting with different prices might be costly. Early on during the Roundup patent
life, there is a lot at risk. When the patent is about the expire, the costs of making a
mistake are lower. In other words, “If it ain’t broke don’t fix it; and if it’s about to break,
try fixing it now.”

3.7. Windows. Microsoft is the dominant player in the market for desktop operating
systems. Suppose each copy of Windows is sold for $50. Suppose moreover that the
marginal cost of production and shipping is $5. What value of the demand elasticity is
consistent with this price? Does it make sense? What elements of the OS market may be
missing from the elasticity rule?
Answer: In order for the price of $50 to be consistent with the simple formula for optimal
pricing, the price elasticity of demand, , would need to be such that
50 − 5 1
=
50 −
which implies  ≈ −1.11. This means that, if the price of Windows were to go up by 10%,
more than 10% would stop buying Windows. But 10% of $50 is $5, which is a very small
fraction of the overall price tag a consumer pays for a computer. It seems unreasonable
that so many users would move away from Windows (to a Mac? Linux?) because of such
a small price increase.
To put it differently: if Microsoft were to apply the optimal price formula with the
true value of elasticity, it would probably set a higher price for Windows. Why then such
a “small” price for the Windows operating system? One possible answer is that, by selling
more units of Windows, Microsoft is able to sell more of other products (Office?), so that
there is some cross-subsidization going on. Another possibility is that, although the short

16
run elasticity is small, in the long run setting a higher price risks “tipping” the industry
towards a different operating system. More on this in Chapter 16.

3.8. Profit maximization. Explain why the assumption of profit maximization is or is

not reasonable?
Answer: The answer to this question may be found in Section 3.3. The main reason we
might think that the assumption of profit maximization is not reasonable is that firm
managers are frequently not the firm owners; and the goals of managers frequently differ
from those of the owners. However, it can be argued that the discipline imposed by
shareholders, the labor market, the product market and the capital market are sufficient
to enforce profit maximization. In particular, the threat of a takeover has been found to
have a significant effect on value maximization.

3.9. Car parts. Two parts in an automobile taillight are the plastic exterior cover and
the light bulb. Which of these parts is a car company more likely to manufacture
in-house? Why?
Answer: Light bulbs are generally a homogeneous good. External suppliers enjoy
economies of scale and specialization; and supply the entire industry. In contrast, the
plastic exterior cover must be custom-designed and manufactured for each make and
model. Because it requires more Relationship Specific Investment (RSI), it is more likely
to be made in-house.

3.10. Jet engines. There are three main suppliers of commercial jet engines, Pratt &
Whitney, General Electric, and Rolls-Royce. All three maintain extensive support staff at
major (and many minor) airports throughout the world. Why doesn’t one firm service
each airport? Why do all three feel they need to provide service and support operations
worldwide themselves? Why don’t they subcontract this work? Why don’t they leave it
entirely to the airlines?
Answer: Jet engines are marvelously idiosyncratic. The knowledge, tools and parts
needed to service one family (brand) of engines do not transfer fully across brands. One
firm does not typically service each airport because the economies of scale (across brands)
are small and the economies of specialization (within brand) are large. The only thing
worse for an airline than an AOG (an aircraft sitting on the ground with a broken engine)
is an aircraft flying with a broken engine or two. To ensure their reputation and revenues
and to avoid ex post hold up, airlines demand before purchasing an aircraft that engine
makers pre-commit capital to ensure that parts and service are available at major stations
world-wide. Because the skills to do this are RSIs, and because the engine owner’s
reputation is at stake, to sell engines and credibly commit to keeping them running, each
manufacturer must provide service and support at major stations.
Subcontracting would be difficult because of the RSI required (the subcontractor
would fear hold-up) and because a poor subcontractor would impose a negative
externality on the manufacturer. When the jet goes down, the manufacturer’s reputation
will suffer on a scale beyond any contractual penalty a subcontractor could likely be held
to, so the work is not usually subcontracted. In addition, the manufacturers benefit
directly from direct feedback within the firm on the performance of the engines they

17
produce. This information may flow more readily within the firm than across firms.
Some airlines with sufficient scale do perform their own routine engine maintenance
at their own maintenance bases. However, the airlines cannot efficiently do emergency
engine repairs away from an airline’s main bases. While there are enough GE engines
going through Karachi International Airport to justify an on-site GE technical support
staff, most airlines do not have enough flights through Karachi to justify the investment.
The economies of scale in non-routine work are site and engine specific, not generally
airline specific.

3.11. Smart car. The Smart car was created as a joint venture between Daimler-Benz
AG and Swatch Group AG. Although Micro Compact Car AG (the name of the joint
venture) was originally jointly owned, in November of 1998 Daimler-Benz AG took
complete control by buying Swatch’s share.3 The deal put an end to a very stressed
relationship between Daimler and Swatch. What does Section 3.4 suggest as to what the
sources of strain might have been?
Answer: Section 3.4 suggests that, when two parties invest in specific assets and contracts
are incomplete, the equilibrium solution is inefficient in every situation short of vertical
integration. It is likely that some of this happened in the “stressed relationship” between
Daimler and Swatch. Since none of the parties was in complete control (and ownership) of
the future developments in the joint venture, the incentives for each party to invest were
less than efficient.

3.12. Franchise retailing. Empirical evidence from franchise retailing suggests that,
even when stores have similar characteristics, the mother company resorts to a mix
between company-owned stores and franchised ones.4 How can this be justified?
Answer: Franchisers face a problem in judging the performance of their franchisees.
Keeping some retail locations in-house provides the parent company with a baseline of
more readily accessible and less biased information against which the performance of the
franchises can be measured. This information then helps to set standards in negotiating
and administering future franchise contracts. Franchising the majority of retail locations
limits the parent’s direct financial outlay and exposure. Franchisers might also have an
interest in direct control of locations that could have a particularly strong impact on its
brand or reputation.

3.13. Body Shop. The U.K. Body Shop franchise network consists of three types of
stores: franchised, company owned and partnership stores. All stores that are distant from
headquarters by more than 300 miles are franchised. More than half of the
company-owned stores are within 100 miles of headquarters.5 How can you explain these
facts?
Answer: Owning a store has the advantages of vertical integration discussed in Section
3.2. However, it also has the problem that it requires increased monitoring by the store
owner. We would expect the costs from monitoring to be lower the closer the store is to
headquarters. Consequently, we would expect vertical integration to be more likely when
the store is located closer to headquarters. The empirical evidence seems consistent with
this hypothesis.

18
 3.14. Intel. Explain why Intel has maintained, if not increased, its competitive
advantage with respect to rivals. Indicate the explanatory power of the different causes
considered in the text (impediments to imitation, causal ambiguity, strategy, history).
Answer: This is a complex question. In fact, as argued in this chapter, this is the question
in strategy. A good source for the particular case of Intel is the HBS case “Intel
Corporation: 1968–1997,” No. 9–797–137 (Rev. October 21, 1998).

3.15. Input shares. Consider a firm with production function (Cobb-Douglas)

q = ξ K α Lβ

Suppose that output price p and input prices r, w are given. Show that a profit
maximizing firm chooses inputs levels such that the ratio between labor costs and total
revenue w L/p q is equal to β.
Answer: Firm profit is equal to

p ξ K α Lβ −(r K + w L)

The first-order condition for maximization with respect to L is given by

β p ξ K α Lβ−1 − w = 0

Noting that
ξ K α Lβ−1 = q/L
the first-order condition may be re-written as

β p q/L − w = 0

or simply
wL
β=
pq

3.16. Cost minimizing input mix. Consider a firm with production function
(Cobb-Douglas)
1 1
q = K 2 L2
Suppose that one unit of capital costs r = 12.5, whereas one unit of labor costs w = 8.
(a) Determine the optimal input mix that leads to an output of q = 2.
Answer: The firm’s problem is
min r K + w L
K,L

subject to
1 1
q = K 2 L2
Solving the constraint for K I get
q2
K= (3.1)
L

19
Substituting this for K in the objective function, the problem becomes

q2
min r +wL (3.2)
L L
The first-order condition for this minimization problem is given by

q2
−r +w =0
L2
Solving for L, I get r
r
L=q (3.3)
w
Substituting 2, 12.5 and 8 for q, r and w I get L = 2.5. Substituting 2.5 for L in (3.1), I
finally get K = 1.6
(b) Determine the firm’s cost function, that is, the minimum cost
required to produce output q.
Answer: From (3.3) for L in (3.2) and simplifying I get
√

C = 2 rw q

Substituting 12.5 and 8 for r and w, respectively, we get C = 20 q.

3.17. Pricing with capacity constraints. Consider again the case of a monopoly facing
linear demand and constant marginal cost. Demand is

q = a − b p,

and marginal cost is MC = c. Suppose moreover that the monopolist has a limited
capacity of K. In other words, it must be q ≤ K. What is the optimal price?
Answer: The easiest way to solve this problem is to first derive the optimal output level
under the assumption that there are no capacity constraints; and then compare the
candidate output level to the capacity constraint. Solving the demand curve for p, we have
a 1
p= − q
b b
This leads to a profit function
 
a 1
π(q) = p q − c q = − q q−cq
b b

The first-order condition for profit maximization (ignoring capacity constraints) is given by
1 a 1
− q + − q − c = 0,
b b b
or simply
a−bc
q=
2

20
Now we have two possibilities. If a−b
2
c
≤ K, then q = a−b
2
c
is the optimal output level.
The optimal price is then given by
 
a 1 a−bc a+bc
p= − =
b b 2 2b

If however a−b
2
c
> K, then the optimal output level is q = K. The reason is that marginal
revenue is greater than marginal cost up to q = a−b c
2 , and so, if q <
a−b c
2 then we want q
as high as possible. Finally, in this case optimal price is given by
a 1 a−K
p= − K=
b b b

3.18. Optimal bidding. You are one of two companies bidding to try to win a large
construction project. Call your bid b. You estimate that your costs of actually performing
the work required will be $800k. You are risk neutral.a You will win if and only if your
bid is lower than that of the other bidder. You are not sure what bid your rival will
submit, but you estimate that the rival’s bid is uniformly distributed between $1m and
$2m.b What bid should you submit?
Answer: A risk-neutral bidder will use a bidding strategy that maximizes the expected
value of its bid b. This entails picking a bid value b that balances two offsetting effects:
changes in the value of winning due to changes in the bid (the larger your bid is the more
valuable the contract is) and changes in the chances of winning due to changes in b (the
larger your bid is the less likely you will win). Formally, the expected value of from
placing a bid b can be expressed

E[b] = (b − 800k) × P(b < br )

where br is the rival bidder’s bid and P(b < br ) is the probability that its bid, b, is less
than its rival’s bid, br . The first term in this equation is simply the payoff when a bidder
wins. The second term is its chances of winning (which requires that b < br ).
In this problem, the focal bidder believes that its rival’s bid can be anywhere between
$1m and $2m so
P(b < br ) = 1 − (b − 1m)/1m
for all bids between $1m and $2m. (Note that P(b < br ) = 1 for all bids less than $1m
since the focal bidder believes that the rival never bids below $1m; and P(b < br ) = 0 for
all bids greater than $2m since the focal bidder believes that the rival never bids above
$2m.) Substituting this expression into the expected value from placing a bid b we obtain
 
b − 1m
E[b] = (b − 800k) 1 −
1m
a . We say that an agent is risk neutral if he or she is indifferent between receiving 100 for sure and
receiving 0 or 200 with probability 50% each. More generally, a risk-neutral agent only cares about
the expected value of each outcome.
b . By “uniformly distributed between a and b” we mean that all values between a and b are equally
likely.

21
From this expression it is clear that the bidder’s payoff, conditional on winning the bid,
goes up with b; but that its chance of winning declines with b. Picking the optimal b
entails finding the maximum of E[b], which we can easily obtain by taking the derivative
of E[b], setting it equal to zero and solving for b. This bid will be the point at which the
two effects of changing b just offset each other. Working in millions as units, we have
∂

E[b] = 1 − (b − 1) − (b − .8) = 0
∂b
b − .8 = 1 − b + 1
b = 1.4m
An alternative approach to this problem is to construct a demand function from the
information we have about the market. You can then solve the problem in the same way
as you would with more straightforward problems in which you are given an explicit
demand function (i.e., set MR = MC and solve for optimal q, then solve for optimal b). To
see how this approach works, note that the bid the firm submits is just like a price. The
higher its bid, the lower its expected demand will be. In this case, demand falls as the
price goes up because the firm’s chance of winning is falling. Formally, expected demand,
q, at any level b is equal to
q = 1 − P(b < br ) = 1 − (b − 1m)/1m = 2 − b/1m
As this equation indicates, when the firm’s bid is equal to 1m, “demand” comes to 1 unit.
That is, the firm is sure to win the contract. As the bid (the price) increases, demand falls
to some fraction of a unit until at 2m “demand” is zero. Since the contract is a
winner-take-all item, the idea of fractional units is not really correct, but if there were,
say, N consumers instead of a single consumer, and the firm was bidding against other
firms for the business of each consumer, the aggregate demand function would then be
 
b bN
qN = N 2 − = 2N −
1m 1m
This is like a simple linear demand function.
To continue with this approach, we need to invert the demand function and solve for
b to get
b = 2m − 1m × q
The bidder’s total revenue is then
b q = (2m − 1m × q) q
Taking the derivative of this total revenue function, we find that the marginal revenue of
the firm is
2m − 2 × 1m × q
As we would expect, the marginal revenue curve has twice the slope of the inverse demand
curve. We can then set this marginal revenue equal to the marginal cost of 800,000 to get
optimal q:
800k = 2m − 2 × 1m × q
2 × 1m × q = 1.2m
q = 1.2/2 = .6

22
Substituting this value into our inverse demand function, we obtain the optimal bid of

b = 2m − .6 × 1m = 1.4m

3.19. Competitive pressure. Suppose that a firm’s profits are given by

π = α + φ(e) + , where α denotes (the inverse of) the intensity of product market
competition, e effort by the manager, and  a random shock. The function φ(e) is
increasing and concave, that is, dφ/de > 0 and d2 φ/de2 < 0.
In order for the firm to survive, profits must be greater than π. The manager’s payoff
is β > 0 if the firm survives and zero if it is liquidated, that is, if profits fall short of the
minimum target. The idea is that if the firm is liquidated, then the manager loses his or
her job and the rents associated with it.
Suppose that  is normally distributed with mean µ and variance σ 2 , and that µ > π.
Show that increased product market competition (lower α) induces greater effort by the
manager, that is, de/dα < 0.
Answer: The manager’s payoff is given by
 
P = β P α + φ(e) +  > π − e
  
= β 1 − F π − α − φ(e) −e

where P(x > y) is the probability that x > y and F () is the probability that  is less than
x (cumulative distribution function). Taking the derivative with respect to e, the
manager’s choice of effort level, we get

dP   dφ(e)
= β f π − α − φ(e) −1
de de

density function of . Since µ > π, we have µ > π − α − φ(e). Therefore

where f () is the
f π − α − φ(e) is in the increasing

portion of f . It follows that an increase in α leads to
a decrease in f π − α − φ(e) ; and this, in turn, implies a lower dP /de. Finally, a lower
dP /de implies a lower value of e. In words, a decrease in the degree of competition
(higher α) decreases the marginal benefit from managerial effort (dP /de), and ultimately
leads to a lower effort of managerial effort (e).
4 Competition

4.1. Vitamin C. Vitamin C is a generic vitamin that is produced by many companies:

brand names are not very important, entry is easy. A good friend — a world-renowned
orthopedic surgeon from New Jersey — tells you that he is about to publish in The New
England Journal of Medicine (a highly respected and widely quoted medical journal) a
study indicating that daily doses of 500 mg of vitamin C tends to improve the muscle tone
and increase the physical stamina of adults, with no adverse side effects. Though a very
good doctor, he is woefully ignorant about the basic workings of markets and wants to
know what is likely to happen, and why — in the short run and in the long run — to the
price of vitamin C, to the quantity sold, to the profits of the producers, and to the number
of firms that produce it. Summarize what you would tell him.
Answer: One would expect demand to increase as a result of the NEJM article. In the
short-run, supply is fixed. We would therefore observe a move along the supply curve,
with both price and output going up. The extent of the price hike would depend on the
steepness of the supply curve: the steeper the short-run supply curve is, the greater the
price increase.
In the long-run, one would expect the supply function to expand, as new producers
enter the market and existing producers expand their capacity. Assuming that demand is
kept at the same level, this would correspond to a movement along the demand curve,
with output going up and price going down.
To summarize: We would expect price to go up in the short-run, then back down in
the long-run, possibly to almost the same level as the initial level. As to output, we would
expect it to go up, with a greater increase in the long-run than in the short run.

4.2. Comparative statics: aspartame, oil. For each of the following, use a supply and
demand diagram to deduce the impact of the event on the stated market. Would you
expect the impact to be primarily on price or quantity? Feel free to mention issues that
you don’t think are captured by a traditional supply and demand analysis.
(a) Event: The FDA announces that aspartame may cause cancer.
Market: Saccharin. (Note: aspartame and saccharin are low-calorie
sweeteners.)
Answer: Saccharin and aspartame are substitutes. We would expect demand for saccharin
to increase (demand curve shifts up/right). This would lead to an increase in the price of

24
saccharin and the quantity produced. The supply is likely to be fairly elastic (flat) as
there are several producers. Thus we would expect the main impact to be on quantity.
The supply curve is expected to be flat particularly in the long-run. For example,
suppliers of other related products can convert their production capacity to saccharin if
the price were to remain at high levels for very long.
(b) Event: Oil price increases. Market: California electricity.
Answer: Oil is an input to some of the generating plants. This implies an upward shift in
marginal cost of some plants (those that use oil). If the plants are the marginal ones
(those determining the price) then price will rise. Since demand is inelastic, the primary
impact will be on price. Otherwise, there’s no impact: these plants simply make less
money. Some of you may have also noticed that oil and electricity are substitutes (in the
long run). This implies that you can also have a shift of the demand to the right. This
increases the price even further, and makes the overall effect on quantity ambiguous.

4.3. Comparative statics: price and quantity effects. Consider following events and
markets:

• OPEC reduces oil output [market: oil]

• Unusually rainy winter in New York City [market: umbrellas in NYC]
• Soccer Champions League final in Madrid [market: Madrid hotels]
• Unusually low catch of sole fish [market: sole fish]

Which of the four corresponds to the four cases considered in Figure 4.3?
Answer:

• Top left: umbrellas

• Top right: Madrid hotels
• Bottom left: sole fish
• Bottom right: oil

4.4. Kidney transplants. Suppose that in a given state — let’s call it state X — a few
recent kidney transplant malpractice suits have led to punitive damage awards of
unprecedented levels. What impact do you expect this to have in the market for kidney
transplant services in state X? To the extent that you can, and making the necessary
assumptions as you go along, indicate the expected effects on price and quantity; the
relative magnitude of these effects; and any possible differences between short-run and
long-run effects.
Answer: The possibility of expensive lawsuits will lead doctors to take on more insurance,
and premiums for such insurance will increase. This implies an upward shift in the supply
curve as each doctor now charges more per unit of service. Given a demand curve, this
immediately leads to an increase in price and a decrease in quantity. Depending on the
nature of demand, the demand elasticity may be greater or smaller in absolute value. As a
general rule, demand for medicines and medical services tends to be relatively inelastic.

25
However, while demand for kidney transplants is relatively inelastic, the demand for
kidney transplants in state X is probably fairly elastic: one can always travel to a
neighboring state. I would thus expect the main effect to be on quantity: state X will
likely become a state with very few kidney transplants.
In the longer run, two things will happen. First, to the extent that demand is
downward sloping, doctors are now receiving a lower net price per unit of medical services
(the rest is insurance premiums). In the long run, doctors will be more flexible regarding
their location and it is expected that some will move out of state. This implies a further
shift in the supply curve, this time to the left. Second, buyers are also likely to be more
flexible, which implies a shift in the demand curve to the left. Both movements imply a
decrease in quantity. The effect on price is unclear: the shift in supply leads to an increase
in price, whereas the shift in demand leads to a decrease in price.

4.5. Book publishing. The technology of book publishing is characterized by a high

fixed cost (typesetting the book) and a very low marginal cost (printing). Prices are set at
much higher levels than marginal cost. However, book publishing yields a normal rate of
return. Are these facts consistent with profit maximizing behavior by publishers? Which
model do you think describes this industry best?
Answer: The model of monopolistic competition is probably the best approximation to
describing this industry. The model of monopolistic competition shows that price-making,
profit-maximizing behavior is consistent with a zero-profit long-run equilibrium. The
strong scale economies in book publishing imply that the gap between price and marginal
cost is particularly high.

4.6. Laundry detergent. The market for laundry detergent is monopolistically

competitive. Each firm owns one brand, and each brand has effectively differentiated itself
so that it has some market power (i.e., faces a downward sloping demand curve). Still, no
brand earns economic profits, because entry causes the demand for each brand to shift in
until the seller can just break even. All firms have identical cost functions, which are
U-shaped.
Suppose that the government does a study on detergents and finds out they are all
alike. The public is notified of these findings and suddenly drops allegiance to any brand.
What happens to price when this product that was brand-differentiated becomes a
commodity? What happens to total sales? What happens to the number of firms in the
market?
Answer: Based on the information provided, it seems that the initial situation in this
market is like the long-run equilibrium of the monopolistic competition model; see Figure
4.7. The government’s announcement has turned a differentiated product into a
homogeneous one. In terms of the graph in Figure 6.3, this implies a flattening of the
demand curve faced by each firm and a new long-run equilibrium where d (now
horizontal) is tangent to the AC curve. At this new long-run equilibrium, price is given by
p0LR and each firm’s output is given by qLR
0 .

Clearly, the new equilibrium implies a lower price and a higher output per firm:
p0LR < pLR and qLR0 >q .
LR
Suppose that price were to drop from pLR to p0LR without changing the degree of

26
product differentiation or the number of firms. This would imply an output per firm equal
0 , where q 0
to qSR 0 . If we take into account the
is greater than qLR but lower than qLR
SR
disappearance of product differentiation (and continue with the same number of firms),
then the output per firm would be less than qSR 0 . Whatever the exact value is, each firm

would be losing money (p0LR < AC). Therefore, in the post-announcement long-run
equilibrium, some firms will need to exit the market.
Finally, it is not clear what will happen to total output. On the one hand, each firm’s
output goes up. On the other hand, the number of firms goes down. Which effect
dominates depends on how consumers value product differentiation and how the demand
curve shifts as a result of the government announcement.

4.7. T-shirt printing. The custom T-shirt printing business has many competitors, so
that the perfect competition model may be considered a good approximation. Currently
the market demand curve is given by Q = 120 − 1.5 p, whereas the market supply is given
by Q = −20 + 2 p.
(a) Determine the market equilibrium
Answer: Equilibrium price is given by the equality
120 − 1.5 p = −20 + 2 p
which implies p = 40. Substituting into the demand curve (or the supply curve, it doesn’t
matter), we get Q = 60.
Suppose there is a T-shirt craze that increases demand by 10% (that is, for each price,
demand is now 10% greater than it was before the price increase).
(b) Determine the new demand curve.
Answer: The new demand curve is obtained by multiplying the initial one by 1.1=1+10%:
Q = 1.1 ×(120 − 1.5 p) = 132 − 1.65 p

(c) Determine the change in equilibrium quantity.

Answer: Equilibrium price is given by the equality
132 − 1.65 p = −20 + 2 p
which implies p = 41.64. Substituting for p in the demand curve we get Q = 63.29. Note
that the initial value of Q was 60. We thus have an increase of 3.29/60, or 5.48%
(d) If your answer to the previous question is different from 10%,
explain the difference in values.
Answer: The shift of the demand curve is partly reflected in an increase in Q and partly
in an increase in p. If the supply curve were flat, then the increase in equilibrium Q would
be exactly 10%. Since the supply curve is upward sloping, the increase is less than 10%
Now go back to the initial demand curve and suppose there is an increase in the cost of
blank T-shirts, an essential input into the business of selling custom T-shirts. Specifically,
for each unit by each supplier, the production cost goes up by 10%.
(e) Determine the new supply curve.

27
Answer: Since we have a proportional vertical increase in the supply curve (that is, when
the supply curve is expressed in the inverse form with price units as a function of quantity
units), we must first find the inverse supply curve, then multiply it by 1.1, then invert it
back again to obtain the (direct) supply curve. From Q = −20 + 2 p we get
p = 10 + .5 Q
The new inverse supply curve is thus given by


p = 1.1 × 10 + .5 Q = 11 + .55 Q
which can be inverted into
Q = −20 + 1.82 p

(f) Determine the change in equilibrium price.

Answer: The new equilibrium price is given by the equality
120 − 1.5 p = −20 + 1.82 p
which implies p = 42.17. This corresponds to an increase of 2.17/40 = 5.42%.
(g) If your answer to the previous question is different from 10%,
explain the difference in values.
Answer: The answer is similar to that of question (d). If the demand curve were perfectly
inelastic, then the shift in the supply curve would imply a change in equilibrium price of
exactly 10%. Since the demand curve is downward sloping, the upward shift in the supply
curve implies an increase in prices as well as a decrease in equilibrium quantity. For this
reason, the increase in equilibrium price is lower than the shift in the supply curve.

4.8. Sales tax. Consider an industry with market demand Q = 550 − 20 p and market
supply Q = 100 + 10 p. Determine the equilibrium price and quantity. Suppose the
government imposes a tax of $6 per unit to be paid by consumers. What is the impact on
equilibrium price and quantity? What if the sales tax is paid by the seller instead of the
buyer?
Answer: We derive equilibrium p and Q by solving the supply and demand system of
equations. Since both are written in terms of Q, this is a relatively easy task:
550 − 20 p = 100 + 10 p
or simply
550 − 100
p= = 15
20 + 10
Substituting for p in the supply equation, we get
Q = 100 + 10 × 15 = 250
Suppose each consumer must pay a $6 tax. This means each consumer is willing to pay 6
dollars less for each unit then before. Consumer gross willingness to pay is given by the
inverse demand curve. From Q = 550 − 20 p, we get
550 1 1
p= − Q = 27.5 − Q
20 20 20

28
It follows that, with the sales tax, willingness to pay is now given by
 
1 1
p = 27.5 − Q − 6 = 21.5 − Q
20 20
or, to put in the same form as initially,

Q = 430 − 20 p

The new equilibrium is determined by the intersection of the demand and the new supply
curve:
430 − 20 p = 100 + 10 p
or simply
430 − 100
p= = 11
20 + 10
Substituting for p in the supply equation, we get

Q = 100 + 10 × 11 = 210

Notice that the price effectively paid by consumers is 11 + 6 = 17. This represents an
increase of 2 with respect to the initial price, which is considerably less than 6. Why?
because some of the tax’s burden is taken by sellers, who now receive a price of 11, a drop
of 4. (Note that, as expected, 2 + 4 = 6, the value of the tax.)
Suppose instead that sellers must pay the $6 tax. This means that if a seller was
willing to sell for a price p, it is now willing to sell for p + 6, so that the net price is the
same p as before. What is then the new supply function? It helps to solve it in terms of p
as a function of Q:
100 1
p=− + Q
10 10
The new supply curve is therefore given by
 
1 1
p = −10 + Q + 6 = −4 + Q
10 10
or, to put in the same form as initially,

Q = 40 + 10 p

The new equilibrium is determined by the intersection of the demand and the new supply
curve:
550 − 20 p = 40 + 10 p
or simply
550 − 40
p= = 17
20 + 10
Substituting for p in the supply equation, we get

Q = 40 + 10 × 17 = 210

As we compare this equilibrium to the case when the tax is paid by the buyer, we realize
that: (a) total output is the same; (b) the price effectively paid by the buyer is the same;

29
(c) the price effectively received by the seller is the same. We have just stumbled into an
important result from the economic theory of taxation: the burden of a sales tax does not
depend on who actually pays the tax!
Extra credit: What then determines the relative burden of a sales tax that falls on
buyers and on sellers?

4.9. Sales tax with steeper demand. Consider again Exercise 4.8. Suppose that
demand is instead given by Q = 280 − 2 p.
(a) Show that the equilibrium levels of p and q are the same as in
the initial equilibrium of Exercise 4.8.
Answer: Equilibrium price is determined by

280 − 2 p = 100 + 10 p

or simply
280 − 100
p= = 15
2 + 10
Substituting for p in the supply equation, we get

Q = 100 + 10 × 15 = 250

the same values as in the initial equilibrium in Exercise 4.8.

(b) Determine the impact of a $6 sales tax in terms of the price
effectively paid by buyers and sellers.
Answer: As shown in Exercise 4.8, it does not matter whether the tax is paid by seller or
buyer. Suppose that sellers must pay the $6 tax. As shown in Exercise 4.8, the new
supply function is given by
Q = 40 + 10 p
The new equilibrium is determined by the intersection of the demand and the new supply
curve:
280 − 2 p = 40 + 10 p
or simply
280 − 40
p= = 20
2 + 10
Substituting for p in the supply equation, we get

Q = 40 + 10 × 20 = 240

(c) Compare the results in (b) to those in Exercise 4.8. Explain the
economic intuition.
Answer: In Exercise 4.8, a $6 tax lead to an equilibrium price of 17. As a result, the $6
tax is borne by consumers ($2) and firms ($4). By contrast, in the present exercise price is
given by 20. As a result, the burden of the $6 tax falls primarily on consumers, who pay
$5, whereas firms pay only $1.

30
 The difference between Exercise 4.8 and the present exercise is that demand is now
steeper. A steeper demand means that consumers are less sensitive to price changes,
which in turn implies that sellers are able to pass through a greater portion of the tax.
More generally, the lower the demand elasticity, the greater the incidence of a sales tax on
consumers.

4.10. Car prices in Europe. Sales taxes on car purchases in Europe vary from 0% to
more than 200%.6 The UK is one of the countries with lowest taxes, whereas Denmark is
one of the countries with highest taxes.
(a) In which countries do you expect consumer prices to be the high-
est?
Answer: The countries with the highest sales tax.
(b) In which countries do you expect pre-tax consumer prices to be
the highest?
Answer: The countries with the lowest sales tax.
By law, if a consumer buys a car in country x and then registers the car in country y, the
consumer receives a refund from the tax paid in country x and then pays the
corresponding tax in country y.
(c) What is the optimal car buying strategy for a European who
does not mind to purchase abroad?
Answer: Purchase the car in the country with the highest sales tax.

4.11. Electricity supply. Consider an electricity market where there are three suppliers,
each with constant marginal cost (a reasonable approximation in electricity generation).
Firm 1 has a capacity of 200 and MC = 5. Firm 2 has a capacity of 100 and MC = 8.
Firm 3 has a capacity of 100 and MC = 10. Suppose that suppliers act as price takers.
(a) Determine the industry supply curve.
Answer: See Figure 4.2. Each firm supplies zero if price is below its marginal cost; if price
is greater than marginal cost, then the firm supplies all the way to capacity. It follows that
the supply curve is zero for p up to 5, 200 for p in the [5, 8) interval, 300 (= 200 + 100) for
p in the [8, 10) interval; and 400 (= 200 + 100 + 100) for p greater than 10.
(b) Suppose that market demand is given by Q = 540 − 20 p. Deter-
mine the market equilibrium. Is this equilibrium a long-run equilib-
rium?
Answer: At p = 10, demand is Q = 540 − 200 = 340. Then Firm 1 produces q = 200,
Firm 2 produces q = 100, and Firm 3 produces q = 40. Since the supply curve is such that
price equals 10 for any output level between 300 and 400, we confirm that market
equilibrium is given by p = 10 and Q = 340. If the technology available to potential
entrants is like firm 3’s, then the answer is yes, this is a long-run equilibrium: an
additional entrant would not be able to make positive profits. Notice that, in this
equilibrium, while firm 3 earns zero profit (it is the marginal firm), both firms 1 and 2
earn positive profits (rents from a better cost function).

31
Figure 4.2
Electricity supply
p

D2 D1

10 ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...
...
...
...
...
8 ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ..
.
...
...
...
... ...
... ...
... ...
... ...
... ...
5 ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ...... ..
.
.
..
.
...
...
...
...
.. ... ...
.
.. ... ...
.
.. ... ...
.
.. ... ...
.
.. ... ...
.
.. ... ...
.
.. ... ...
.
.. ... ...
.
..
...
...
..
... q
0
0 100 200 300 400

(c) Suppose that (i) demand falls to D(p) = 400 − 20 p; (ii) Firm 3
reduces its MC to 8; (iii) Firm 2 reduces its MC to 7. What happens
to equilibrium profits in each case?
Answer: See Figure 4.2. (i) If demand falls, then so do profits. (ii) If Firm 3 reduces its
MC to 8, the price drops to 8, and so do profits. (iii) If Firm 2 reduces its MC to 7, then
Firm 2’s profits increase, whereas Firm 1 and Firm 3’s profits remain constant.

4.12. Average and marginal cost. Show that, in a long-run equilibrium with free entry
and equal access to the best available technologies, the comparison of price to the
minimum of average cost or the comparison of price to marginal cost are equivalent tests
of allocative efficiency. In other words, price is greater than the minimum of average costs
if and only if price is greater than marginal cost.
Show, by example, that the same is not true in general (hint: consider a monopolist
with constant average and marginal cost c).
Answer: We first show the following fact: marginal cost is greater than average cost if
and only if average cost is increasing. To see this, notice that Average Cost is given by
the ratio Cost / Output. Taking the derivative with respect to Output q, we get
dC
d AC d C dq q−C 1

= = = MC − AC
dq dq q q2 q
which shows the fact.
In the long-run equilibrium of an industry with equal access, each firm will be
producing at a point in the left-hand portion of its Average Cost curve. Given the above
fact, it follows that marginal cost is lower than or equal to average cost. Since there is free
entry, price is equal to average cost. Specifically, either price is equal to the minimum of
average cost and equal to marginal cost; or price is greater than the minimum of average
cost and greater than marginal cost.

32
 The same is not true, for example, in a short-run equilibrium. Consider the case of
perfect competition and suppose that price is greater than the minimum of average cost.
Since firms are price takers, price is equal to marginal cost. So, price is greater than the
minimum of average cost, whereas price is equal to marginal cost.
5 Market Failure

5.1. Front yards. Front yards, if well tended, generate positive externalities for a
house’s neighbors. Do you think this is an important externality? Is the market solution
inefficient?
Answer: The externality is likely to be important: one of the characteristics that make a
neighborhood attractive is precisely how well tended the front yards are. If each household
were to choose gardening effort solely based on his or her benefits, it’s likely that the
equilibrium level of gardening effort would be lower than optimal. However, it’s also likely
that neighborhood social pressure induces residents to put a little more effort than they
would like: if they don’t, they risk being shunned by their neighbors. If the social norm
regarding acceptable gardening levels is at the socially optimal level, then it is possible
that the market solution be efficient.

5.2. AT&T. The long-run demand elasticity of AT&T in the period 1988–1991 was
estimated to be around -10.7 Assuming the estimate is correct, what does this imply in
terms of AT&T’s market power at the time?
Answer: A demand elasticity of 10 implies that AT&T’s demand is very elastic. In fact,
the author of the study that produced this estimate computes the welfare loss due to
AT&T’s market power to be less than 1% of sales volume.

5.3. Monopoly power. “The degree of monopoly power is limited by the elasticity of
demand.” Comment.
Answer: Optimal monopoly pricing leads to the following relation between the price-cost
margin and demand elasticity: (p − MC )/p = 1/||, where p is price, MC marginal cost,
and e demand elasticity. It follows that the greater the value of  the lower the value of
(p − MC ) and the lower monopoly profits. A monopolist facing a very elastic demand
curve makes profits at the level of a competitive firm.

5.4. Windows. Is the Windows operating system an essential facility? What about the
Intel Pentium microprocessor? To what extent does the discussion in Section 5.6 on
essential facilities (vertical integration, access pricing) apply to the above examples?
Answer: By means of preface: this is a very controversial question; not all economists

34
agree on a single answer. Having said that, both Microsoft (the producer of the Windows
operating system) and Intel (the producer of the Intel Pentium microprocessor) provide
computer makers with essential components, without which the machines could not
function. Nevertheless, strictly speaking, we cannot say that their output represents an
essential facility. The discussion in Section 5.6 applies to monopolists. The crucial
difference from the examples presented in the section is the fact that Microsoft and Intel
are not monopolists: computer makers always have the option of switching to another
provider of components.
However, the widespread use of the Windows operating system, and the fact that
Windows is only supplied by Microsoft, implies that the latter’s position is much closer to
the one of a monopolist than is Intel’s. Even though Intel’s chip design is very close to
being an industry standard, Intel is not the only company supplying microprocessors with
that design. Hence, the Windows operating system is closer to what is called an essential
facility than Intel’s Pentium processor.

5.5. Carbon tax. Consider an industry with demand q = 1 − p and supply q = p.

Suppose that each unit of output implies one unit of CO2 added to the atmosphere and a
marginal social cost of e, where e is the total level of emmissions.
(a) What is the level of CO2 emission at the market equilibrium?
Answer: Equating supply and demand, we get 1 − p = p, or simply p = 12 . Substituting in
the supply curve (or the demand curve) we get q = 21 . Since one unit of output
corresponds to one unit of emissions, we conclude that the emissions level is 12 .
(b) What is the socially optimal level of CO2 emissions?
Answer: Since e = q, The total social marginal cost is given by q + q = 2 q, where the first
term corresponds to production cost and the second to cost from emissions. The inverse
demand is given by p = 1 − q. Equating the two, we get

2q = 1 − q

or simply q ∗ = 13 .
(c) Determine the Pigou tax that achieves the social optimum.
Answer: Equilibrium output (supply = demand) is given by

q+t=1−q

Since the goal is q = 31 , it follows that t∗ = 1 − 2 q ∗ = 13 .

5.6. Common facility. Different divisions within a firm frequently compete for a
common resource. Suppose that divisions 1 and 2 of a given firm share a common facility
F . Let yi be the service level used by division i (i = 1, 2). Division i’s gross benefit in
terms of improved divisional earnings is given by yi − 0.25 yi2 − 0.1 (y1 + y2 ).
(a) What are the equilibrium levels of yi if the various divisions act
separately?

35
Answer: Each division maximizes

yi − 0.25 yi2 − 0.1 (y1 + y2 )

Let us consider division 1. Since it cannot control the other division, the optimal solution
corresponds to maximizing

y1 − 0.25 y12 − 0.1 y1 = 0.9 y1 − 0.25 y12

The solution is y1 = 1.8. By symmetry, the same is true for the other division, so y2 = 1.8.
(b) What are the optimal levels of yi from an overall firm point of
view?
Answer: The firm maximizes total benefit, that is

y1 − 0.25 y12 − 0.1 (y1 + y2 ) + y2 − 0.25 y22 − 0.1 (y1 + y2 ) =

= 0.8 (y1 + y2 ) − 0.25 (y12 + y22 )

Maximizing with respect to y1 we get y1 = 1.6, and similarly y2 = 1.6.

(c) Explain the difference between the results in (a) and (b).
Answer: The equilibrium value is greater than the optimal value. The reason is that there
is a negative externality: when division 1 uses the facility more intensely, it does not take
into account that this costs division 2 lower divisional earnings. In other words, part of
division 1’s gain is not a gain for a firm as a whole.
(d) How can equilibrium and optimality be reconciled?
Answer: One possibility is to create a system of transfer pricing whereby each division
pays for the use of the common facility F . If the fee is set at .1 (the cost imposed on the
other facility), then the equilibrium solution will be optimal.
6 Price Discrimination

6.1. Perfect price discrimination. Consider a monopolist with demand D = 120 − 2 p

and marginal cost MC = 40. Determine profit, consumer surplus, and social welfare in the
following two cases: (a) single-price monopolist; (b) perfect price discrimination.

Solution: As we have seen before, in these problems it is useful to work with the inverse
demand curve. In the present case, this is given by p = 60 − .5 q. Marginal revenue is
given by MR = 60 − q. Equating to marginal cost and solving with respect to q, we get
60 − q = 40, or q = 20. This implies p = 60 − 20/2 = 50. Monopoly profits are
(50 − 40) × 20 = 200, whereas consumer surplus is 12 (60 − 50) × 20 = 100. It follows that
social surplus is equal to 300.
Consider now the case of a perfect discriminating monopolist. Price for the qth unit
is now given p = 60 − .5 q, so long as the resulting price is greater than marginal cost. If
follows that the monopolist’s optimal output is given by solving p = MC , or 60 − .5 q = 40,
which yields q = 40. Under perfect discrimination, consumer surplus is zero (each
consumer pays a price equal to its willingness to pay, or else it doesn’t buy). Seller’s profit
is given by the integral of the difference between price and marginal cost. Since demand is
linear, this is the area of the triangle formed by the demand curve and the MC curve, from
0 to optimal q: 12 (60 − 40) × 40 = 400.

6.2. The Economist. First-time subscribers to the Economist pay a lower rate than
repeat subscribers. Is this price discrimination? Of what type?
Answer: This is an example of price discrimination by indicators (also known as
third-degree price discrimination). The market is segmented into new subscribers and
repeat subscribers. New subscribers, know the product less well and are thus likely to be
more price sensitive. Moreover, the fact that they have not subscribed in the past
indicates that they are likely to be willing to pay less than current subscribers. It is
therefore optimal to set a lower price for new subscribers.

6.3. Cement. Cement in Belgium is sold at a uniform delivered price throughout the
country, that is, the same price is set for each customer, including transportation costs,
regardless of where the customer is located. The same practice is also found in the sale of
plasterboard in the United Kingdom.8 Are these cases of price discrimination?

37
Answer: Yes, these are cases of price discrimination. Consider the total price being paid
by each customer, P , as being composed of the price actually charged and the
transportation cost; P = pi + ti . Since locations are different, transportation costs are
different, thus, each consumer is charged a price pi that depends on his or her location.
This is a clear example of geographic price discrimination, one instance of discrimination
by indicators.

6.4. Fulton fish market. A study of the New York fish market (when it was the Fulton
fish market) suggests that the average price paid for whiting by Asian buyers is
significantly lower than the price paid by White buyers.9 What type of price
discrimination does this correspond to, if any? What additional information would you
need in order to answer the question?
Answer: This appears to be a case of price discrimination by indicators (also known as
third-degree price discrimination), whereby a group of buyers (a market segment) pays a
different price than another group. Theory predicts that in a non-competitive market
(monopoly, oligopoly) buyers with higher price elasticity should be charged a lower price;
as a result, we can conclude that Asian buyers have higher price elasticity than white
buyers.
In order to have a more accurate picture, however, more information is needed.
Different prices could could simply result from quantity discounts and the possible fact
that different quantities are bought by the different groups. If that were the case, we
would have price discrimination by self-selection, not by indicators. In other words, it may
be that Asian buyers pay a lower price not because they are Asian but because they
purchase larger quantities. Similarly, the time of purchase (e.g., before 5am or after 5am)
could be correlated with race, so that it is not race that determines the price difference.
A similar reasoning applies to the type of establishment the buyer represents (store,
fry shop, etc.). Finally, it could also be the case that different groups use different types of
payment type (cash or credit), so that different prices reflect different costs.
For a more complete discussion, see the cited reference.

6.5. Coupons. Supermarkets frequently issue coupons that entitle consumers to a

discount in selected products. Is this a promotional strategy, or simply a form of price
discrimination? Empirical evidence suggests that paper towels are significantly more
expensive in markets offering coupons than in markets without coupons.10 Is this
consistent with your interpretation?
Answer: This may be interpreted as a case of price discrimination by self selection. By
offering coupons (hence a lower price), supermarkets can serve the buyers with a higher
price elasticity at a different price. In order for this strategy to improve revenues with
respect to single price, supermarkets should then set a higher regular price. Hence,
empirical evidence is consistent with the explanation that this is a form of price
discrimination.

6.6. GetGoing.com. In 2013, the travel booking site GetGoing was offering a “Pick
Two, Get One” deal: customers would select two flights — to different cities — and make
a purchase before finding out where they would be heading. GetGoing would then select

38
one of the destinations. Explain how this can be a profitable strategy for GetGoing.
Answer: The idea is that business travelers are less flexible: they have specific
appointments at specific locations. They would never accept a deal that would take them
to Istanbul or Beijing. Leisure travelers, by contrast, are more flexible: they want to go to
a fun destination, but it does not have to be exactly a particular destination.
Given this, the “Pick Two, Get One” provides a good vehicle for price discrimination.
It is effectively a form of damaged good: at no savings for itself, the seller makes the
service strictly worse (nobody likes the uncertainty about their final destination). The
point is that the utility loss from product damaging is much greater for high-valuation
buyers (business travelers) than for low-valuation travelers.

6.7. Coca-Cola. In 1999, Coca-Cola announced that it was developing a “smart”

vending machine. Such machines are able to change prices according to the outside
temperature.11 Suppose, for the purposes of this problem, that the temperature can be
either “High” or “Low.” On days of “High” temperature, demand is given by
Q = 280 − 2 p, where Q is number of cans of Coke sold during the day and p is the price
per can measured in cents. On days of “Low” temperature, demand is only Q = 160 − 2 p.
There is an equal number days with “High” and “Low” temperature. The marginal cost of
a can of Coke is 20 cents.
(a) Suppose that Coca-Cola indeed installs a “smart” vending ma-
chine, and thus is able to charge different prices for Coke on “Hot”
and “Cold” days. What price should Coca-Cola charge on a “Hot”
day? What price should Coca-Cola charge on a “Cold” day?
Answer: On a Hot day, Q = 280 − 2 p, or p = 140 − Q/2. Marginal revenue is
MR = 140 − Q. Equating to marginal cost (20) and solving, we get Q∗ = 120 and p∗ = 80.
On a Cold day, Q = 160 − 2 p, or p = 80 − Q/2. Marginal revenue is MR = 80 − Q.
Equating to marginal cost (20) and solving, we get Q∗ = 60 and p∗ = 50.
(b) Alternatively, suppose that Coca-Cola continues to use its normal
vending machines, which must be programmed with a fixed price,
independent of the weather. Assuming that Coca-Cola is risk neutral,
what is the optimal price for a can of Coke?
Answer: Observe from part (a) that even on a Hot day the optimal price is no greater
than 80 cents. So, we can restrict our attention to prices of 80 cents or less. In this price
range, the expected demand is given by Q = 12 (280 − 2p) + 12 (160 − 2 p) = 220 − 2 p.c
Solving for p gives p = 110 − Q/2. The marginal revenue associated with this expected
demand curve is given by MR = 110 − Q. Equating this marginal revenue to marginal
cost, we get Q∗ = 90 and p∗ = 65.
(c) What are Coca-Cola’s profits under constant and weather-
variable prices? How much would Coca-Cola be willing to pay to
enable its vending machine to vary prices with the weather, i.e., to
have a “smart” vending machine?
c . If p > 80, then we must consider the possibility of zero consumption by at least one of the market
segments.

39
Answer: Under price discrimination, from part (a), profits on a Hot day are
(80 − 20) × 120 = $72, and profits on a Cold day are (50 − 20) × 60 = $18. Expected
profits per day are therefore ($72 + $18)/2 = $45. Under uniform pricing, expected profits
per day are (65 − 20) × 90 = $40.50. It follows that Coca-Cola should be willing to pay up
to an extra $4.50 per day for a “smart” vending machine.
However, one might add that there are other important considerations. The fact that
Coca-Cola did not pursue the idea of a “smart” vending machine is probably less due to
the economics of it as it is due to the consumer backlash that the idea created.

6.8. Sal’s satellite. Sal’s satellite company broadcasts TV to subscribers in LA and

NY. The demand functions are given by
1
QNY = 50 − 3 PNY
2
QLA = 80 − 3 PLA

where Q is in thousands of subscriptions per year and P is the subscription price per year.
The cost of providing Q units of service is given by

TC = 1, 000 + 30 Q

where Q = QNY + QLA .

(a) What are the profit-maximizing prices and quantities for the NY
and LA markets?
Answer: Note that, from the cost function, we get MC = 30. It follows that PNY = 90,
QNY = 20, PLA = 75 and QLA = 30.
(b) As a consequence of a new satellite that the Pentagon developed,
subscribers in LA are now able to get the NY broadcast and vice versa,
so Sal can charge only a single price. What price should he charge?
Answer: Note that demand in NY is positive if and only if price is lower than 150;
whereas demand in LA is positive if and only if price is lower than 120. It follows that
total demand is given by

 QNY + QLA = 130 − P
 if P < 120
Q= 1
QNY = 50 − 3 P if 120 < P < 150

0 if P > 150


Suppose that P < 120, so that the first formula applies. Then the inverse demand curve is
given by P = 130 − Q. Optimal price is P = (130 + 30)/2 = 80, which confirms the
hypothesis that P < 120. Quantity is given by Q = 130 − P = 50.
(c) In which situation is Sal better off? In terms of consumers’ sur-
plus, which situation do people in LA prefer? What about people in
NY? Why?
Answer: Sal is better off without the new satellite (that is, when he can price
discriminate). People in NY prefer the new satellite, since they pay a lower price. People
in LA preferred the situation before the introduction of the satellite, because the price

40
increased for them. As often is the case, price discrimination makes some people better
off, some people worse off.

6.9. Stadium pricing. Stanford Stadium has a capacity of 50k and is used for exactly
seven football games a year. Three of these are OK games, with a demand for tickets
given by D = 150k − 3 p per game, where p is ticket price. (For simplicity, assume there is
only one type of ticket.) Three of the season games are not so important, the demand
being D = 90k − 3 p per game. Finally, one of the games is really big, the demand being
D = 240k − 3 p. The costs of operating the Stadium are essentially independent of the
number of tickets sold.
(a) Determine the optimal ticket price for each game, assuming the
objective of profit maximization.
Answer: Demand for OK games is given by D = 150 − 3 p, where number of tickets is
measured in thousands. Inverse demand is p = 50 − Q/3. Marginal revenue is
MR = 50 − 23 Q. Marginal cost is zero, since costs do not depend on the number of tickets
sold. Equating marginal cost to marginal revenue, we get Q = 75. This is greater than
capacity. Therefore, the optimal solution is simply to set price such that demand equals
capacity: 150 − 3 p = 50, which implies p = $33.3
Demand for not-so-important games is given by D = 90 − 3 p. Inverse demand is
p = 30 − Q/3. Marginal revenue is MR = 30 − 32 Q. Equating marginal revenue to marginal
cost, we get Q = 45. Substituting back in the inverse demand curve we get p = $15.
Since demand for the Big Game is greater than for the OK games, it will surely be
the case that MR = MC implies a demand level greater than capacity. The optimal price is
therefore determined by equating demand to capacity: 240 − 3 p = 50, or simply p = $63.3
Given that the Stadium is frequently full, the idea of expanding the Stadium has arisen.d
A preliminary study suggests that the cost of capacity expansion would be $100 per seat
per year.
(b) Would you recommend that Stanford go ahead with the project
of capacity expansion?
Answer: The marginal gross profit of an additional seat is the sum of the difference
between marginal revenue and marginal cost for all games where capacity was a
constraint. For OK games, marginal revenue is given by MR = 50 − 23 50 = 16.7. For the
Big Game, MR = 80 − 32 50 = 46.7. Adding these up (three times the first plus the second)
we get $96.7. Since this is less than the marginal cost of capacity expansion, it is not
worth it to pursue the project. (Note that during OK games there is excess capacity, so
these do not benefit from the project.)

6.10. Spoken Word. Your software company has just completed the first version of
SpokenWord, a voice-activated word processor. As marketing manager, you have to decide
on the pricing of the new software. You commissioned a study to determine the potential
demand for SpokenWord. From this study, you know that there are essentially two market
segments of equal size, professionals and students (one million each). Professionals would
be willing to pay up to $400 and students up to $100 for the full version of the software. A
d . Ignore the fact that Stanford Stadium used to hold 90,000 seats and was thought to be too big.

41
substantially scaled-down version of the software would be worth $50 to students and
worthless to professionals. It is equally costly to sell any version. In fact, other than the
initial development costs, production costs are zero. Although you know there are two
market segments, you cannot directly identify a consumer as belonging to a specific
market segment.
(a) What are the optimal prices for each version of the software?
Answer: It is optimal to price the full version at 400 and the scaled-down version at 50.
Total profits are 450.
Suppose that, instead of the scaled-down version, the firm sells an intermediate version
that is valued at $200 by professionals and $75 by students.
(b) What are the optimal prices for each version of the software? Is
the firm better off by selling the intermediate version instead of the
scaled-down version?
Answer: One first possibility would be to price the intermediate version at 75 and the full
version at 400. However, this would lead professionals to choose the intermediate version
since the difference between willingness to pay and price is greater for the intermediate
version. In order to induce professionals to buy the full version, the full version’s price
would need to be 75 + (400 − 200) = 275, where the value in parentheses is the
professionals’ difference in willingness to pay between the two versions. (This is the same
as saying professionals need to get a minimum surplus of 200-75=125.) This would lead to
a total profit of 275 + 75 = 350, which is lower than initially. Still another possibility
would be to price the full version at 400 and the intermediate version at
400 − (400 − 200) = 200. In this case, professionals would buy the full version but students
would not buy the intermediate version. Profits would then be 400: better than 350 but
still less than the 450 the firm would get with the truly scaled-down version.

6.11. SoS. SoS (Sounds of Silence, Inc) prepares to launch a revolutionary system of
bluetooth-enabled noise-cancellation headphones. It is estimated that about 800,000
consumers would be willing to pay $450 for the headphones; an additional 1,500,000
consumers would be willing to pay $250 for the headphones. Though SoS knows this
marketing information, it cannot identify a consumer as belonging to one group or the
other.
SoS is considering the launch of a stripped-down version of the headphones (the
stripped-down version uses wires instead of bluetooth). The 800,000 high-valuation
consumers would only be willing to pay $325 for the stripped-down version. The
remaining 1,500,000 consumers don’t particularly care about bluetooth vs. wire
connections; they are willing to pay the same $250 for either version.
Both the bluetooth version and the stripped-down version cost the same to produce:
$100 per unit.
(a) Determine the optimal pricing policy assuming that SoS only
sells bluetooth-enabled headphones.
Answer: There are two candidate price levels: 450 and 250. At 450, profit is given by
(450 − 100) × 800, 000 = $280m. At 250, profit is given by

42
(250 − 100) × (800, 000 + 1, 500, 000) = $345m. It follows the optimal price is $250.
(b) Determine the optimal pricing policy assuming that SoS offers
the two versions.
Answer: The new alternative to consider is offering the stripped-down version at a lower
price and the full version at a higher price. The relevant constraint is that high-end users
have no incentive to choose the stripped down version. This implies that

450 − pH ≥ 325 − pL

Since low-valuation buyers will not have an incentive to buy the more expensive version (a
fact that we can confirm later on), the best we can do is to charge their valuation: 250. It
follows from the above inequality that the highest price we can charge the full version is
pH = 450 − 325 + 250 = 375. Under this strategy, total profit is given by

π = (375 − 100) × 800, 000 + (250 − 100) × 1, 500, 000 = $445m

Since this is better than $345m, the optimal solution determined in (a), versioning is the
optimal strategy.
(c) Suppose that SoS finds out that the estimate regarding the num-
ber of low-valuation users is overly optimistic. In fact, there are only
300,000 consumers who would be willing to pay $250. How would you
change your answer to (a) and (b)?
Answer: Given a different number of low-valuation consumers, we must recompute both
the values in (a) and the values in (b). As to (a), we still have $280m from targeting
high-end users only. A price of $250 now leads to a profit of
(250 − 100) × (800, 000 + 300, 000) = $165m. It follows that, if the seller were to sell one
version only, then the optimal price is now $450.
Consider now the case of versioning. The no-deviation constraint is still the same and
so pH would still be equal to $375. Total profit is now given by

π = (375 − 100) × 800, 000 + (250 − 100) × 300, 000 = $265m

This is less than $280m. It follows that the overall optimal policy is now to forget about
versioning and simply target high-end users with a price of $450.

6.12. RawDeal. RawDeal is the new sushi bar in the neighborhood. Their estimated
marginal cost is 10 cents per sushi unit. RawDeal estimates that each consumer has a
demand for sushi given by q = 20 − 10 p, where q is number of sushi units and p is price in
dollars per unit.
(a) Determine the optimal price per sushi unit.
Answer: Inverse demand is given by

p = 2 − 0.1 q

If demand is linear and given by p = a − b Q, then optimal price is given by

a+c
p=
2

43
It follows that, in the present case,
2 + .1
p= = 1.05
2
(that is, one dollar and 5 cents).
(b) RawDeal is considering switching to an all-you-can-eat-sushi pol-
icy. Determine the optimal price per customer. How does profit com-
pare to pricing per unit?
Answer: All you can eat means the price per unit is zero. It follows that each consumer
will eat 20 units of sushi. The consumer surplus from 20 units at zero price is given by
1
CS = 2 × 20 = 20
2
It follows that revenue per customer is $20. Since each customer will eat 20 units, cost per
customer will be 20 × .1 = $2. It follows that profit per customer is given by $18. By
contrast, charging 1.05 per piece leads to a demand of

q = 20 − 10 p = 20 − 10 × 1.05 = 9.5

and a profit of
π = (1.05 − .10) × 9.5 = .95 × 9.5 = 9.025
It follows that switching to all you can eat nearly doubles RawDeal’s profit.
(c) Discuss other advantages and disadvantages of each pricing op-
tion.
Answer: Implementation costs play an important role here. One of the advantages of
all-you-can eat is that you don’t need to monitor how much each patron eats: all you need
to do is to charge the entrance fee. One potential problem is that all patrons are not
equal: if you get stuck with people who eat a lot, then the deal may be better for the
customers than for the seller. In fact, one problem when you have all-you-can eat
restaurants competing against “conventional” ones is that consumers naturally self select,
with the big eaters disproportionately visiting the all-you-can-eat venues more often.
(d) Ignoring implementation costs, what is the optimal two-part tar-
iff for sushi (i.e., a fee at the door plus a price per sushi piece).
Answer: The optimal two-part tariff is to set price equal to marginal cost and charge a
fixed fee equal to the consumer surplus at that price. If unit price is 10 cents, then each
consumer will order
q = 20 − 10 p = 19
units of sushi. This implies that consumer surplus is given by the area of the triangle
between the demand curve and price level, that is,
1
× (2 − .1) × 19 = 18.05
2
Profit per consumer is then given by

π = F + pq − cq

44
where F is the fixed fee, p is price per unit, c is cost per unit, and q quantity consumed.
However, since we set p = c, we simply get π = F . In other words, under the optimal
two-part tariff profit per consumer is given by 18.05. As expected, this is better than all
you can eat (18 per consumer), though by very little (less than 1%). In fact, given the
transactions costs of charging on a per unit basis, all-you-can eat may well be the overall
optimal solution.

6.13. Pricing with limited capacity. Consider the model of a monopolist with two
markets presented earlier in the chapter. Suppose that the seller has a limited capacity
and low marginal cost up to capacity. An example of this would be an airline with two
types of passengers or a football stadium with two types of attendees.
Derive the conditions for optimal pricing. How do they relate to the case when there
are no capacity constraints?
Answer: Let K denote capacity and p1 (q1 ), p2 (q2 ) the inverse demand functions. The
monopolist’s problem becomes:
max q1 p1 (q1 ) + q2 p2 (q2 ) − c (q1 + q2 )
q1 ,q2

subject to
q1 + q2 ≤ K
Suppose that the constraint q1 + q2 ≤ K is not binding. Then the elasticity rule applies
and we have
MR 1 = MR 2 = c
Suppose that the constraint q1 + q2 ≤ K is binding. Then the cost term in the objective
function is simply c K, a constant; and the constraint can be solve to yield q2 = K − q1 .
Since cost is constant, the objective is to maximize the revenue in market 1 plus the
revenue in market 2. Since q2 = K − q1 , each time we increase q1 by one unit we must
decrease q2 by 1 unit as well, and vice-versa. This implies that the derivative of the
objective function with respect to q1 is given by
MR 1 + (−1) MR 2
Since this must equal zero (first-order condition) we again obtain MR 1 = MR 2 .
The same result can be obtained intuitively. Suppose that the seller is capacity
constrained. Is the current set of prices optimal? One alternative is to take one unit from
one market and sell in the other market, changing prices accordingly. Would the seller
want to do this? By taking one unit away from Market 1, the seller would lose MR 1 . By
selling it in Market 2, the seller would get MR 2 . Optimality then requires that
MR 1 = MR 2 .
The difference with respect to the case when capacity is not a constraint is that, if
before MR 1 = MR 2 = c, now MR 1 = MR 2 takes on a higher value, whichever value solves
q1 + q2 = K.

For math aficionados. From a mathematical point of view, the best way to solve a
maximization problem with constraints is to write down the Lagrangean for this problem.
In this case we have


L = q1 p1 (q1 ) + q2 p2 (q2 ) − c (q1 + q2 ) + λ K − q1 − q2

45
The first-order conditions are:
MR 1 = MC + λ
MR 2 = MC + λ

Depending on whether capacity constraints are binding or not, we will have λ positive or
zero. Whichever is the case, the above equations show that optimality implies that
marginal revenue be equated across markets. Notice that, if demand elasticity differs
across markets, then this implies different prices for the different markets.

6.14. BlackInk. Printing Solutions, the maker of the printer BlackInk, faces an
important product design dilemma: deciding the speed of its popular laser printer. There
are two market segments: Professionals are willing to pay up to $800 (a − .5) for the
printer, where a is printer speed. Students, in turn, are willing to pay up to $100 a.
Maximum printer speed corresponds to a = 1, whereas a = 0 corresponds to a worthless
printer. There are one million professionals and one million students. It is equally costly
to produce a printer with any level of a. In fact, other than the initial development costs,
production costs are zero.
How many versions of the BlackInk should Printing Solutions sell? Which versions?
What are the optimal prices of each version?
Answer: Suppose first that the firm sells one version only. If that is the case, then it
might as well choose a = 1. There are then two candidates for optimal price: $400 and
$100. Profits are given by $400m in the first case and $200m in the second case (recall
that there are one million professionals and one million students). It follows that
a = 1, p = 400 is the optimal solution (conditional on there being only one version).
Since there are only two types of consumers, it will not be necessary to offer more
than two different versions. Since it is equally costly to produce any version and
willingness to pay is increasing in a, it follows that one of the versions should have
maximum speed (a = 1).
If there is to be self-selection between two different versions, it will be the case that
professionals choose the faster printer and students the slower one. We thus have four
possible constraints: incentive and participation; for type H and for type L (professionals
and students, respectively). I next argue that, if the value of a is optimally chosen, then
all constraints except the low type incentive constraint must be binding:
1. The type H incentive constraint must be binding. If that were not the case, then we
could increase the value of a a little bit without violating the incentive constraint.
This would imply a higher value for the L type, which would then allow me to
charge a higher price.

2. The type L participation constraint must be binding. Were that not the case, I
could increase the price of the slower printer and increase profit without violating
any constraint.

3. The type H participation constraint must be binding. If that were not the case, then
I could decrease the slow printer’s speed a bit, decrease the price of the slower
printer by the decline in valuation to type L buyers, and then increase the price of
the fast printer such that the type H incentive constraint remains valid. At the

46
 margin, this would increase seller’s profit, thus contradicting the possibility that the
type H participation constraint is not binding.
Given the above binding constraints, it must be that p1 (the price of the full version)
equals $400 (type H participation constraint), whereas pa (the price of the slow version,
with speed a) equals 100 a (type L participation constraint). Finally, the type H incentive
constraint implies that type H gets zero surplus from buying a slow printer:

800 (a − .5) − pa = 0

which implies a = 4/7, which in turn implies and pa = 100 × (4/7) ≈ 57.14.
Profits under one version are $400m. Under two versions, profits are $457.14m, an
increase of $57.14m. Basically, the increase corresponds to student sales.
Note for aficionados: the fact that three of the four constraints are binding need not
always be the case (see the baby iMac example in the main text). Suppose for example
that students have no use for printers that are slower than a = .5 but otherwise are not
interested in speed (that is, they are willing to pay 50 for any printer with a ≥ .5). Then
it is optimal to offer a printer with a = .5 and none of the incentive constraints are
binding. Suppose instead that students have no use for printers that are slower than
a = .6 but otherwise are not interested in speed (that is, they are willing to pay 60 for any
printer with a ≥ .6). Then it is optimal to offer a printer with a = .6 and the type H
participation constraint is not binding, for the optimal p1 is then 380 = 400 − 80 + 60.

6.15. Multiple two-part tariffs. Consider the model of non-linear pricing introduced in
Section 6.2. Suppose there are two types of consumers, in equal number: type 1 have
demand D1 (p) = 1 − p, and type 2 have demand D2 (p) = 2 (1 − p). Marginal cost is zero.
(a) Show that if the seller is precluded from using non-linear pricing,
then the optimal price is p = 12 and profit (per consumer) 38 .
Answer: Total demand from a consumer of Type 1 and a consumer of Type 2 is given by
D(p) = D1 (p) + D2 (p) = 1 − p + 2 (1 − p) = 3 (1 − p). The monopolist’s problem is:

max 3 p (1 − p)
p

The solution to this problem is given by the first order condition, 1 − 2 p = 0, so that we
get p = 21 and the profit is 34 . Social welfare is given by the sum of the firm’s profit and
the consumer surplus and is equal to: Wa = 3 p (1 − p) + (1 − p)2 = 1.
(b) Show that if the seller must set a single two-part tariff, then the
9
optimal values are f = 32 and p = 41 , for a profit of 16
9
.
Answer: In this case the monopolist’s demand is the same. However, the monopolist now
can also charge a fixed fee, f , from both consumers. The problem becomes:
max 3 p (1 − p) + 2 f
p
(1 − p)2
s.t. ≥f
2
where the constraint comes from the fact that the consumer of Type 1 must have a
positive surplus, otherwise it will not buy. Once the constraint for the Type 1 consumer is

47
satisfied, the constraint for Type 2 is also satisfied; we can therefore ignore it. The
monopolist is better off when it extracts as much surplus as possible from consumers.
Thus, its optimal policy requires that the fixed fee be equal to the Type 1 consumer
surplus, that is, the constraint should be binding. The monopolist’s problem becomes:

max 3 p (1 − p) + (1 − p)2
p

and the solution is given by the first order condition, 3 − 6 p − 2 + 2 p = 0, so that we get
p = 41 , f = 32
9
and the profit is 89 . Welfare is given by
(1−p)2 45
Wb = 3 p (1 − p) + (1 − p)2 + 0 + 2 = 32 > Wa .
(c) Show that if the seller can set multiple two-part tariffs, then the
optimal values are f1 = 18 , p1 = 12 , f2 = 87 , p2 = 0, for a profit of 58 .
Answer: In this case the monopolist’s problem is more complex:

max p1 (1 − p1 ) + f1 + 2 p2 (1 − p2 ) + f2
p1 ,p2
s.t. CS 1 (p1 ) ≥ f1 (PC1)
CS 2 (p2 ) ≥ f2 (PC2)
CS 1 (p2 ) − f2 ≤ CS 1 (p1 ) − f1 (IC1)
CS 2 (p1 ) − f1 ≤ CS 2 (p2 ) − f2 (IC2)

where the participation constraints assure that the consumer will prefer to consume and
the incentive compatibility constraints assure that each plan is chosen by the targeted
type of consumers, that is, Type 1 consumers will prefer plan 1 to plan 2 while Type 2
consumers will prefer plan 2 to plan 1.
One can show that PC1 and IC2 are binding, while PC2 and IC1 are not. Suppose
that PC1 and IC2 are satisfied. We have:

CS 2 (p2 ) − f2 ≥ CS 2 (p1 ) − f1 ≥ CS 2 (p1 ) − CS 1 (p1 ) ≥ 0

where the last inequality comes from the fact that, at any price, the surplus of the Type 2
consumers is higher, since they consume more. Therefore, PC2 is automatically satisfied.
PC2 will not be binding unless consumers of Type 1 are not served. To see this, suppose
PC2 is binding. From IC2 and PC1 we get

CS 2 (p1 ) ≤ f1 ≤ CS 1 (p1 )

which is obviously impossible. In contrast, PC1 must be binding: if PC1 and PC2 would
not bind the monopolist could increase its profits by increasing both f1 and f2 with the
same small amount without violating the ICs. If IC2 is not binding the monopolist could
increase f2 with a small amount and keep all other constraints satisfied, while increasing
her profits.
2
Therefore, we have f1 = CS 1 (p1 ) = (1−p
2
1)
and

(1 − p1 )2
f2 = CS 2 (p2 ) − CS 2 (p1 ) + f1 = (1 − p2 )2 −
2

48
The monopolist’s problem becomes:

max p1 (1 − p1 ) + 2 p2 (1 − p2 ) + (1 − p2 )2
p1 ,p2

The first order conditions are: 1 − 2 p1 = 0 and 2 − 4 p2 − 2 + 2 p2 = 0, and the solutions

are: p1 = 12 , f1 = 81 , p2 = 0, f2 = 78 , and profit is equal to 45 . Social welfare in turn is given
by
(1 − p1 )2 11
Wc = p1 (1 − p1 ) + + 2 p2 (1 − p2 ) + (1 − p2 )2 = < Wb
2 8

(d) Show that, like profits, total surplus increases from (a) to (b)
and from (b) to (c).
Answer: The proof is already contained in the previous points.

6.16. Sales. Many retail stores set lower-than-usual prices during a fraction of the time
(sale). One interpretation of this practice is that it allows for price discrimination between
patient and impatient buyers.
Suppose that each buyer wants to purchase one unit per period. Each period is
divided into two subperiods, the first and the second part of the period. Suppose there are
two types of buyers, i = 1, 2. Each type of buyer is subdivided according to the part of the
period they would ideally like to make their purchase. One half the buyers would prefer to
purchase during the first part of the period, one half during the second part. A buyer of
type i is willing to pay v i for a purchase during his or her preferred part of the period; and
v i for a purchase at another time.
Buyers of type 1, which constitute a fraction α of the population, are high-valuation,
impatient buyers; that is, v h is very high and v h very low. High valuation implies that v h
is very high; impatience implies that v h is very low: buyers of type 1 are not willing to
buy at any time other than their preferred time. Buyers of type 2, by contrast, are very
patient: v l ≈ v l . Assume that α is relatively low; specifically, α < v l /v h . To summarize:
v h > v l ≈ v l > α v h > v h ≈ 0.
(a) Show that, under a constant-price strategy, the seller optimally
sets p = v l .
Answer: If p > v 1 , then there is no sale. If v 2 < p < v 1 , then the only purchasers are the
impatient, high-valuation buyers, and the seller’s profit is π = α p, with maximum value
α v 1 . If p < v 2 , then all buyers make a purchase and the seller’s profit is π = p, with
maximum value v 2 . Since α v 1 < v 2 , it is clear that the best constant-price strategy is to
set p = v 2 .
(b) Determine firm profits when it sets prices p = v h and p = v l
in the first and second parts of the period, respectively. Show that
profits are greater under the “sales” strategy.
Answer: Under this strategy the seller’s profit is
1 1 1

π= α v 1 + (1 − α) v 2 + α v 2 = v 2 + α v 1 − v 2 > v 2
2 2 2
where the last inequality is based on the fact that v 2 ≈ v 2 .

49
 6.17. Optimal bidding strategy. Consider a first-price auction with two bidders.
Suppose Bidder 1 believes that Bidder 2’s bid is some number between 0 and 12 , with all
numbers equally likely. Show that Bidder 1’s optimal bid is given by b1 = v1 /2.
Answer: By bidding b1 , Bidder 1’s expected profit is given by

π1 = (v1 − b1 ) P(b1 > b2 )

where vi and bi is bidder i’s value and bid, respectively.

The higher b1 , the lower the net gain from winning the auction, v1 − b1 ; but the
higher the probability of winning the auction, P(b1 > b2 ). Specifically, if b1 = 0, then
P(b1 > b2 ) = 0; whereas, if b1 = 21 , then P(b1 > b2 ) = 1. More generally, for b1 ∈ [0, 12 ],

P(b1 > b2 ) = 2 b1

It follows that
π1 = (v1 − b1 ) 2 b1
Taking the derivative with respect to b1 and equating to zero, we get the first-order
condition for profit maximization (see Section 3.2):

2 ( − b1 + v1 − b1 ) = 0

which leads to the desired expression.

This exercise begs the question of what should we expect Bidder 1’s belief about
Bidder 2’s bid to be. This gets into the realm of game theory, which I introduce in
Chapter 7. See specifically Exercise 7.11.
7 Games

7.1. Dominant and dominated strategies. What are the assumptions regarding player
rationality implicit in solving a game by elimination of dominated strategies? Contrast
this with the case of dominant strategies.
Answer: See the discussion on pages 145 and following.

7.2. The movie release game. Consider the example at the beginning of the chapter.
Suppose that there are only two blockbusters jockeying for position: Warner Bros.’s Harry
Porter and Fox’s Narnia. Suppose that blockbusters released in November share a total of
$500 million in ticket revenues, whereas blockbusters released in December share a total of
$800 million.
(a) Formulate the game played by Warner Bros. and Fox.
Answer: The game in normal form is as follows, where payoffs are in $ million:
Warner Bros.
November December
250 800
November
250 500
Fox
500 400
December
800 400

(b) Determine the game’s Nash equilibrium(a).

Answer: There are two Nash equilibrium in this game: (N,D) and (D,N). See also the
discussion on page 147.

7.3. Ericsson v Nokia. Suppose that Ericsson and Nokia are the two primary
competitors in the market for 4G handsets. Each firm must decide between two possible
price levels: $100 and $90. Production cost is $40 per handset. Firm demand is as follows:
if both firms price at 100, then Nokia sells 500 and Ericsson 800; if both firms price at 90,
then sales are 800 and 900, respectively; if Nokia prices at 100 and Ericsson at 90, then
Nokia’s sales drop to 400, whereas Ericsson’s increase to 1100; finally, if Nokia prices at 90

51
and Ericsson at 100 then Nokia sells 900 and Ericsson 700.
(a) Suppose firms choose prices simultaneously. Describe the game
and solve it.
Answer: First, it may help to write the demand curve as a matrix. (Notice this is not the
game firms are playing.)
Ericsson
100 90
800 1100
100
500 400
Nokia
700 900
90
900 800

Now, based on price, marginal cost and demand, we can write the payoff corresponding to
each strategy pair. This is now the normal form game played by firms
Ericsson
100 90
48 55
100
30 24
Nokia
42 45
90
45 40

Pricing at 90 is a dominant strategy for Nokia and Ericsson alike. The Nash equilibrium is
therefore given by (90,90). Equilibrium profits are given by (40,45).
(b) Suppose that Ericsson has a limited capacity of 800k units per
quarter. Moreover, all of the demand unfulfilled by Ericsson is trans-
ferred to Nokia. How would the analysis change?
Answer: The new demand matrix is given by
Ericsson
100 90
800 800
100
500 700
Nokia
700 800
90
900 900

The new game is given by

52
 Ericsson
100 90
48 40
100
30 42
Nokia
42 40
90
45 45

It is now a dominant strategy for Ericsson to price at $100. It is still a dominant strategy
for Nokia to price at $90.
(c) Suppose you work for Nokia. Your Chief Intelligence Officer
(CIO) is unsure whether Ericsson is capacity constrained or not. How
much would you value this piece of info?
Answer: Nokia has a dominant strategy: price at 90. Therefore, it has no value for the
information of whether Ericsson is or is not capacity constrained (as far as the present
game is concerned).

7.4. ET. In the movie E.T., a trail of Reese’s Pieces, one of Hershey’s chocolate brands,
is used to lure the little alien out of the woods. As a result of the publicity created by this
scene, sales of Reese’s Pieces trebled, allowing Hershey to catch up with rival Mars.
Universal Studio’s original plan was to use a trail of Mars’ M&Ms, but Mars turned down
the offer. The makers of E.T. then turned to Hershey, who accepted the deal.
Suppose that the publicity generated by having M&Ms included in the movie would
increase Mars’ profits by $800,000 and decrease Hershey’s by $100,000. Suppose moreover
that Hershey’s increase in market share costs Mars a loss of $500,000. Finally, let b be the
benefit for Hershey’s from having its brand be the chosen one (while the deal would cost
$1 million).
Describe the above events as a game in extensive form. Determine the equilibrium as
a function of b. If the equilibrium differs from the actual events, how do you think they
can be reconciled?
Answer: The game’s extensive form is the following (payoffs in millions of dollars):

accept
.....
..... -.5, b - 1
.........................................................................................
.....
...
......
....
.....
.....
reject .......
.....

.....
....
.....
....
. ....
. ....
. ....
. ....
. ....
. ....
. ....
. ....
. ....
. ....
. ....
. ....
. ....
. ....
. ....
. ....
.
H...
..
.....
.....
.....
..... .....
......... .....
.....
..
..... .....
.....
...
..
.. .
...... reject
.....
.....
.......
M .....
.....
0, 0
..........................................................................................
.....
.....
.....
.....
.....
.....accept
.....
...........................................................................................
-.2, -.1

If b > 1, then Hershey is better off by accepting Universal’s offer, were it ever asked to
make that choice; in which case Mars is better off by accepting Universal’s offer. If b < 1,
then Hershey is better off by rejecting Universal’s offer, were it ever asked to make that
choice; in which case Mars is better off by rejecting Universal’s offer.

53
 7.5. ET (continuation). Return to Exercise 7.4. Suppose now that Mars does not know
the value of b, believing that either b =$1,200,000 or b =$700,000, each with probability
50%. Unlike Mars, Hershey knows the value of b. Draw the tree for this new game and
determine its equilibrium.
Answer: The game’s extensive form is now given by the following (payoffs in thousands of
dollars):

b = 1200 buy
-500
......
......
.................................................................... -500, 200
(50%) ..... ......
......
............................................................................
−500 × 50% + 0 × 50% = ...
... H ......

= −250 ......
...
......
......
......not buy
.
..
..
...................................................................
0, 0
not buy . ....
.
.

... N
.........................................................................
...
..
...
...
...
... b = 700 buy
..
.
.... ...
...
...
......
......
...................................................................... -500, -300
.......
.
. ...
...
.
.
(50%) . .........
......
M...
... H
........................................................................
......
......
...
...
...
......
...... not buy
...
...
...
0
.....................................................................
0, 0
...
...
...
...
...
... buy
..................................................................
-200, -100

where M and H refers to Mars and Hershey, whereas N refers to the player Nature. The
values next to the H nodes correspond to M ’s expected payoff if we ever get to that node.
Nature is not a strategic player: it simply chooses different branches according to
predetermined probabilities. In the present case, Nature flips a fair coin and chooses
b =$1,200,000 or b =$700,000 with equal probability. This implies that, from M ’s point of
view, the expected value given that we are in the N node is given by

−500 × 50% + 0 × 50% = −250

It follows that M ’s optimal choice is to accept Universal’s offer. To summarize, the

equilibrium strategies are given by

• Mars: accept Universal’s offer

• Hershey: accept Universal’s offer if b is high, reject otherwise.

7.6. Hernan Cortéz. In a message to the king of Spain upon arriving in Mexico,
Spanish navigator and explorer Hernan Cortéz reports that, “under the pretext that [our]
ships were not navigable, I had them sunk; thus all hope of leaving was lost and I could
act more securely.” Discuss the strategic value of this action knowing the Spanish
colonists were faced with potential resistance from the Mexican natives.
Answer: By eliminating the option of turning back, Hernan Cortéz established a credible
commitment regarding his future actions, that is, to fight the Mexican natives should they
attack. Had Cortéz not made this move, natives could have found it better to attack,
knowing that instead of bearing losses the Spaniards would prefer to withdraw.

7.7. HDTV standards. Consider the following game depicting the process of standard
setting in high-definition television (HDTV).12 The US and Japan must simultaneously

54
decide whether to invest a high or a low value into HDTV research. If both countries
choose a low effort then payoffs are (4,3) for US and Japan, respectively; if the US chooses
a low level and Japan a high level, then payoffs are (2,4); if, by contrast, the US chooses a
high level and Japan a low one, then payoffs are (3,2). Finally, if both countries choose a
high level, then payoff are (1,1).
(a) Are there any dominant strategies in this game? What is the
Nash equilibrium of the game? What are the rationality assumptions
implicit in this equilibrium?
Answer: The game in matrix form looks like the following:
Effort by Japan
Low High
3 4
Low
4 2
Effort by US
2 1
High
3 1

It is a dominant strategy for the US to choose Low. Given that the US chooses Low,
Japan’s best response is to choose High. (Low, High) is thus the only Nash equilibrium of
the game.
(b) Suppose now the US has the option of committing to a strategy
ahead of Japan’s decision. How would you model this new situation?
What are the Nash equilibria of this new game?
Answer: The most natural way to model this situation is by writing an extensive form
game as follows:

L ..................................................................... 4, 3
.....
.....
.....
.....
L ....
..........................................................................
Japan
.. ..
... .....
.. .....
.... .....
.....
...
...
.
... .....
H
.....
....................................................................
2, 4
..
..
...
.
..
...
.....
US ...
...
...
...
...
... L ..................................................................... 3, 2
... .....
... .....
... .....
... .....
... H
............................................................................
Japan .....
.....

.....
.....
.....
.....
H
.....
.....
...................................................................
1, 1

Japan’s optimal strategy is to choose H if the US choses L and to choose L if the US

chooses H. Anticipating that strategy, the US optimal strategy is to choose H. The
equilibrium is therefore (H,L).
(c)
Comparing the answers to (a) and (b), what can you say about
the value of commitment for the US?
Answer: In the simultaneous move game, US and Japan choose (L,H), respectively, which
gives the US a payoff of 2. In the sequential move game (with the US moving first), US

55
and Japan choose (H,L), respectively, which gives the US a payoff of 3. It follows that the
value of commitment for the US is 3 − 2 = 1.
(d) “When pre-commitment has a strategic value, the player that
makes that commitment ends up ‘regretting’ its actions, in the sense
that, given the rivals’ choices, it could achieve a higher payoff by
choosing a different action.” In light of your answer to (b), how would
you comment on this statement?
Answer: In the sequential choice game (with the US moving first), Japan ends up
choosing L. Given that Japan chooses L, the payoff for the US would be higher if it
chooses L instead of H. In this sense, there is ex-post regret. However, the sole reason for
Japan choosing L is precisely the fact the US commits to H. Were such commitment not
credible, that is, were the US able to change its choice easily, then Japan should anticipate
that change and accordingly choose H. In this sense, the US should not regret having
committed to H in the first place.

7.8. Finitely repeated game. Consider a one-shot game with two equilibria and
suppose this game is repeated twice. Explain in words why there may be equilibria in the
two-period game which are different from the equilibria of the one-shot game.
Answer: When the game is repeated twice the strategy space for each player becomes
more complex. Each player’s strategy specifies the action to be taken in period 1 as well
as the action to be taken in period 2 as a function of the outcome in period 1. The
possibility of linking period 2’s actions to past actions allows for equilibrium outcomes
that would not be attainable in the corresponding one-shot game (for example, the use of
a ’punishment’ action in period 2 if one of the players deviates from the designated period
1 payoff-maximizing action).

7.9. American Express’s spinoff of Shearson. In 1993, American Express sold Shearson
to Primerica (now part of Citigroup). At the time, the Wall Street Journal wrote that
Among the sticking points in acquiring Shearson’s brokerage operations would
be the firm’s litigation costs. More than most brokerage firms, Shearson has
been socked with big legal claims by investors who say they were mistreated,
though the firm has made strides in cleaning up its backlog of investor cases.
In 1992’s fourth quarter alone, Shearson took reserves of $90 million before
taxes for “additional legal provisions.”13
When the deal was completed, Primerica bought most of Shearson’s assets but left the
legal liabilities with American Express. Why do you think the deal was structured this
way? Was it fair to American Express?
Answer: One reason why the deal was structured this way is that it avoided the adverse
selection effect of a seller with superior information. At first, this may seem unfair to
American Express, but in the end it may be better for both sides.

7.10. Sale of business. Suppose that a firm owns a business unit that it wants to sell.
Potential buyers know that the seller values the unit at either $100m, $110m, $120m,
. . . $190m, each value equally likely. The seller knows the precise value, but the buyer only

56
Table 7.3
Sale of business
Probability Exp. value Expected
Price of sale if accepted profit
100 10 110 1
110 20 115 1
120 30 120 0
130 40 125 -2
140 50 130 -5
150 60 135 -9
160 70 140 -14
170 80 145 -20
180 90 150 -27
190 100 155 -35

knows the distribution. The buyer expects to gain from synergies with its existing
businesses, so that its value is equal to seller’s value plus $10m. (In other words, there are
gains from trade.) Finally, the buyer must make a take-it-or-leave-it offer at some price p.
How much should the buyer offer?
Answer: We can write down Table 7.3, which summarizes, for each offer that the buyer
makes, the probability that the offer gets accepted, the expected value (to the buyer)
conditional on having the offer accepted, and the overall expected profit from any given
offer. From this we see that the buyer should thus offer either $100m or $110m.
Suppose the buyer offers p = 100 (in $m). Then, in most cases the offer is rejected.
Specifically, 90% of the times the offer is rejected. Offering more would imply a higher
probability of sale, but the expected value of the unit would increase by less than the price
paid. The intuition for this result is the force of adverse selection: the seller will only sell
the unit if its value is relatively low.

7.11. First-price auction. Consider the following auction game. There are two bidders
who simultaneously submit bids bi for a given object. Bidder i values the object at vi ; it
knows its own value but not the other bidder’s value. It is common knowledge that
valuations vi are uniformly drawn from the unit interval, that is, vi ∼ U [0, 1].
(a) Suppose that Bidder 1 expects Bidder 2’s bid to be uniformly
distributed between 0 and 12 . What is Bidder 1’s optimal bid function
(that is, bid as a function of valuation v1 )?
Answer: Suppose Bidder 1 believes that Bidder 2’s bid, b2 , is some number between 0 and
1
2 , with all numbers equally likely (that is, Bidder 2’s bid is uniformly distributed in the [0
and 12 ] interval. By bidding b1 , Bidder 1’s expected profit is given by

π1 = (v1 − b1 ) P(b1 > b2 )

57
The higher b1 , the lower then net gain from winning the auction, v1 − b1 ; but the higher
the probability of winning the auction, P(b1 > b2 ). Specifically, if b1 = 0, then
P(b1 > b2 ) = 0; whereas, if b1 = 21 , then P(b1 > b2 ) = 1. More generally, for b1 ∈ [0, 12 ],

P(b1 > b2 ) = 2 b1

It follows that
π1 = (v1 − b1 ) 2 b1
Taking the derivative with respect to b1 and equating to zero, we get the first-order
condition for profit maximization (see Section 3.2):

2 ( − b1 + v1 − b1 ) = 0

or simply
v1
b1 = (7.4)
2

(b) If Bidder 2 expects Bidder 1 to follow the strategy derived in

part (a), what is Bidder 2’s belief about Bidder 1’s bid levels?
Answer: Since Bidder 2 knows that v1 is uniformly distributed in [0,1], (7.4) implies that
b1 is uniformly distributed in the [0, 12 ] interval.
(c) Determine the bidding game Nash equilibrium (assuming there
is only one).
Answer: From part (a), we know that the bidding function (7.4) is optimal given the
belief that the other bidder’s bid is uniformly distributed in [0, 12 ]. From part (b), we know
that, if bidders bid according to (7.4), then, from the other bidder’s perspective, bids are
uniformly distributed in the [0, 21 ] interval. Together this implies that strategies and
beliefs form a Nash equilibrium.

7.12. Ad games. Two firms must simultaneously choose their advertising budget; their
options are H or L. Payoffs are as follows: if both choose H, then each gets 5; if both
choose L, then each gets 4; if firm 1 chooses H and firm 2 chooses L, then firm 1 gets 8
and firm 2 gets 1; conversely, if firm 2 chooses H and firm 1 chooses L, then firm 2 gets 8
and firm 1 gets 1.
(a) Determine the Nash equilibria of the one-shot game.
Answer: H is a dominant strategy, so the unique Nash equilibrium is (H, H).
(b) Suppose the game is indefinitely repeated and that the relevant
discount factor is δ = .8. Determine the optimal symmetric equilib-
rium.
Answer: The condition that (L, L) is an equilibrium is that
5 4
≥8+δ
1−δ 1−δ

which is equivalent to δ ≥ 43 . Since δ = .8, it follows that (L, L) is indeed an equilibrium.

58
 (c) (challenge question) Now suppose that, for the first 10 periods,
firm payoffs are twice the values represented in the above table. What
is the optimal symmetric equilibrium?
Answer: From the analysis in the previous answer, we conclude that, after t = 10, (L, L)
is an equilibrium. Consider the situation at t = 10. Current payoffs are doubled. It follows
that the no-deviation constraint is
5 4
10 + δ ≥ 16 + δ
1−δ 1−δ

which implies δ ≥ 76 ≈ .86. If follows that (L, L) is not an equilibrium. By induction and a
fortiori, we also conclude that (L, L) is not an equilibrium for any earlier t. It follows the
best symmetric equilibrium is for firms to choose H during the first 10 periods and L
thereafter.

7.13. Finitely repeated game. Suppose that the game depicted in Figure 7.1 is
repeated T times, where T is known. Show that the only subgame perfect equilibrium is
for players to choose (B, R) in every period.
Answer: Suppose we are in period T , the last period of the finitely-repeated game.
Subgame perfection implies that we look for a Nash equilibrium of this subgame. As we
saw earlier, there exists a unique Nash equilibrium of this one-shot game: (B, R).
Now consider the subgame starting in period T − 1. This is effectively a two-period
game. Players correctly anticipate that, regardless of what happens in period T − 1,
(B, R) will be played in period T . For this reason, they should treat choices in period
T − 1 as if they were playing a one-shot game: nothing in the past or in the future
depends on the outcome of what takes place in period T − 1. Since there exists a unique
equilibrium in the one-shot game, players choose (B, R) in period T − 1.
By induction, we conclude that, in a subgame perfect Nash equilibrium, players must
choose (B, R) in every period.

7.14. Centipede. Consider the game in Figure 7.3.14 Show, by backward induction,
that rational players choose d at every node of the game, yielding a payoff of 2 for Player
1 and zero for Player 2. Is this equilibrium reasonable? What are the rationality
assumptions implicit in it?
Answer: Starting from the right-most node, we observe that Player 2’s strategy, if that
node is reached, is to play d, in which case its gets 101, whereas Player 1 gets 99. This
implies that, in the second to last node, Player 1 is better off choosing d. In fact, by
choosing r, Player 1 expects to get 99 (see sentence above) instead of 100 from d. And so
forth. We conclude that the unique sub-game perfect Nash equilibrium is for each player
to play d whenever it is called upon to make a move. The outcome of this equilibrium is
Player 1 getting 2 and Player 2 getting 0.
Obviously, one might question whether this result is reasonable or not. Here, the
implicit assumption is that each player is rational, believes that the other player is
rational, believes that the other player believes that the first player is rational, and so
forth.

59
 To see how important this assumption is, suppose that Player 1 chooses r in the first
period. Since this is not according to the equilibrium, Player 2 may not conjecture that
Player 1 is not rational. But then choosing d may no longer be in Player 2’s best interest.
But then choosing r may be, after all, a rational strategy by Player 1 in the first place.

7.15. Advertising levels. Consider an industry where price competition is not very
important: all of the action is on advertising budgets. Specifically, total value S (in
dollars) gets splits between two competitors according to their advertising shares. If a1 is
firm 1’s advertising investment (in dollars), then its profit is given by
a1
S − a1
a1 + a2
(The same applies for firm 2). Both a1 and a2 must be non-negative. If both firms invest
zero in advertising, then they split the market.
(a) Determine the symmetric Nash equilibrium of the game whereby
firms choose ai independently and simultaneously.
Answer: Firm i’s profit is given by
ai
πi = S − ai
ai + aj

where i 6= j. The first order condition for profit maximization with respect to ai is given by

(ai + aj ) − ai
S−1=0
(ai + aj )2

In a symmetric equilibrium, we have a1 = a2 = b

a. Thus

(b
a+b a) − ba
2
S−1=0
(b
a+ba)

or simply
1
a=
b S
4
Each player’s payoff is then given by
1 1 1
π
b= S− S= S
2 4 4
For aficionados: Note that in deriving the above solution I “cut some corners” by
assuming the solution is symmetric. I next follow a more complete line of reasoning. From
the first-order condition, we can derive firm i’s best response mapping. From the
first-order condition we get
aj S = (ai + aj )2
or simply p
ai = aj S − aj
Solving the system, and imposing that ai ≥ 0, we get ai = aj , as assumed in the earlier
derivation.

60
Figure 7.3
The centipede game. In the payoff vectors, the top number is Player 1’s payoff, the bottom
one Player 2’s.
" #
................ r ................. r ................. r ................. r ................. r ................ r ................. r ................. r 100
... ............................... ............................... ............................... ............................... ... ............................... ...............................
..... 1
..... ......
. ... 2
..... ......
. ... 1
..... ......
. ... 2
..... ......
. ... 1
..... ......
...................
. ... .....................
... 2
................
.. ... 1
..... ......
. ... 2
..... ......
............................
.
.........
...
.........
...
.........
...
.........
...
.........
... .
....
....
.
....
..
.. .........
... 100
.... .... ... ... ... ... ... ...
... ... ... ....
d ...
... d ...
... d .... d ...
...
d ...
...
d ...
... d ...
... d ...
... ... ... . ... ... ...
.
" # " # " # " # " # " # " #" #
2 1 4 3 6 97 100 99
0 3 2 5 4 99 98 101

(b) Determine the jointly optimal level of advertising, that is, the
level a∗ that maximizes joint profits.
Answer:
π1 + π2 = S − a1 − a2
It follows that a1 = a2 = a∗ = 0 maximizes joint profits.
(c) Given that firm 2 sets a2 = a∗ , determine firm 1’s optimal ad-
vertising level.
Answer: Given a2 = 0, any positive a1 gives firm 1 100% of the market. Since advertising
is costly, firm 1’s best response is to set an arbitrarily small but strictly positive value of
a1 (similarly to price undercutting under Bertrand competition).
(d) Suppose that firms compete indefinitely in each period t =
1, 2, ..., and that the discount factor is given by δ ∈ [0, 1]. Deter-
mine the lowest value of δ such that, by playing grim strategies, firms
can sustain an agreement to set a∗ in each period.
Answer: By choosing a∗ each period, each firm gets S/2. The optimal deviation yields
approximately S. Finally, the static Nash equilibrium yields S/4 for each firm. The
condition for a grim strategy equilibrium whereby firms set a = 0 in each period is then
given by
1 1 δ 1
S≥S+ S
1−δ 2 1−δ 4
or simply
2
δ≥
3
8 Oligopoly

8.1. Bertrand in the real world. The Bertrand model of price competition suggests
that, under a given set of conditions, firms make zero economic profits even if there are
only two firms. However, there are many instances of industries with a small number of
competitors where firms appear to earn more than zero economic profits.
Give an example of an industry dominated by a couple of firms where profits are
significant. Explain why the predictions of the Bertrand model are not borne out.
Answer: Examples include: P&G and Unilever, Pepsi and Coke, Christie’s and Sotheby’s,
Dell and Lenovo, breakfast cereals, air travel (?), vodka. Model conditions that may fail
include :

• Infinite supply ability (implicit in constant marginal cost assumption). In fact,

capacity limitations may dull the incentives to undercut. This may apply to airlines
in routes when capacity is lower than demand.
• Perfect substitutes. In practice, firms may create brands to distinguish themselves
and thus soften price competition. This is probably the most important point in the
P&G/Unilever, Pepsi/Coke, vodka, breakfast cereal examples.
• Identical marginal costs. Frequently, one of the firms is more efficient than the other.
This may apply to Dell in the desktop computer market and possibly to Southwest
Airlines in air travel.
• One-time competition. Repeated interaction opens the scope for implicit or explicit
collusion. Christie’s and Sotheby’s may be a good example.

8.2. Bertrand and zero profits. According to Bertrand’s theory, price competition
drives firms’ profits down to zero even if there are only two competitors in the market.
Why don’t we observe this in practice very often?
Answer: See the discussion starting on page 171. Some of the ways out of the so-called
“Bertrand trap” include product differentiation, dynamic competition, asymmetric costs,
and capacity constraints.

8.3. E-commerce. E-commerce represents an increasing fraction of economic

transactions in many different industries. Does e-commerce create a Bertrand trap? What

62
is special about e-commerce (and, more generally, the new economy) that makes the
Bertrand trap a dangerous trap? How can e-commerce firms avoid the trap?
Answer: In many ways, e-commerce makes sellers look more alike, that is, reduces the
degree of product differentiation. In this sense, it makes firms more prone to fall in the
Bertrand trap. Moreover, unlike brick-and-mortar stores, capacity constraints tend to be
less important at online stores — because, for example, they can pool the inventories at
various physical locations — which further increases the chance of a Bertrand trap.
Having said that, there are ways in which an online company can differentiate itself.
For example, Amazon.com, by means of a successful branding strategy and a superior
search engine, has managed to carve out a considerable market share of relatively loyal
consumers.

8.4. Price discrimination. A monopolist is generally better off by price-discriminating.

What about a duopolist? Consider the special case when Market A is Firm 1’s “strong”
market, whereas market B is firm 2’s “strong” market.
Answer: There may be situations where duopolists are strictly worse off by their ability to
price discriminate. Suppose that Firm 1 has a strong following in market A, whereas Firm
2 has a strong following in market B. If firms must set the same price in both markets,
they have good reasons not to price very aggressively: each firm sets a high price and
enjoys a large market share in its “home” market. In fact, undercutting the rival in its
home market would imply a large cost in terms of home market margin for a relatively low
gain. If however the firms can set different prices in each market, then there is less of a
reason for Firm 1 not to go after Firm 2’s home market more aggressively. This may end
up in a prisoner’s dilemma type of outcome: firms are worse off by pricing aggressively
even though firms would be better off if both priced less aggressively.15

8.5. Cournot model assumptions. Three criticisms are frequently raised against the use
of the Cournot oligopoly model: (i) firms normally choose prices, not quantities; (ii) firms
don’t normally take their decisions simultaneously; (iii) firms are frequently ignorant of
their rivals’ costs; in fact, they do not use the notion of Nash equilibrium when making
their strategic decisions. How would you respond to these criticisms? (Hint: in addition to
this chapter, you may want to refer to Chapter 7.)
Answer:

1. As explained at the end of Section 8.1, if firms are capacity constrained, then price
competition “looks like” like quantity competition.
2. If there are significant information lags, then sequential decisions “look like”
simultaneous decisions. See Chapter 7.
3. At the end of Section 8.2, we discussed an argument for the relevance of Nash
equilibrium which only requires each firm to know its own profit function.

8.6. Cournot vs. Bertrand. Which model (Cournot, Bertrand) would you think
provides a better approximation to each of the following industries: oil refining, internet
access, insurance. Why?

63
Answer: Capacity constraints seem relatively more important in oil refining and relatively
less important in insurance. Given the discussion in Section 8.3, one would be inclined to
select the Cournot model for oil refining and the Bertrand model for insurance. Internet
access is an intermediate case between the previous two examples.

8.7. ByeByeCold. You are currently the sole seller of ByeByeCold, a revolutionary drug
that almost instantly eliminates cold symptoms. Although the production cost is only
$.10 per dose, you sell ByeByeCold for $1.39 per dose, for a total profit of $900m a year.
You are currently considering licensing ByeByeCold to a second producer. Neither you nor
your competitor have any significant capacity constraints. One of your managers
suggested, since the firm would be sharing the market with a competitor, it would be
appropriate to charge a flat fee that covers half the current profits plus a generous margin;
the value of $700m was suggested. An alternative proposal would be to set a royalty fee of
$.50 per dose. What is your opinion?
Answer: If you license ByeByeCold for a flat fee, you will be competing with another firm
selling the same product and with a similar marginal cost. Except for the possibility of
collusion, this would imply approximately zero profits for both firms. It follows that half
the current profits would not be sufficient to compensate for the profit loss from licensing.
In fact, there exists no licensing contract that would be profitable for both parties.
Why do firms ever license, then? One possibility is that the second firm is more
efficient in production so that there are gains from bringing it on board. Also, there may
be reasons why production by a second firm increases the size of the market. If these were
true, then it is possible that a profitable licensing arrangement can be made.

8.8. French generics manufacturer. Consider the last problem in Section 8.4. Suppose
that a retailing campaign costing e 80m is expected to increase demand by 40%. Suppose
also that the current rupee/euro exchange rate is 50 INR/e. Should the French firm go
ahead with the campaign? One macroeconomics expert tells you that “it is likely that the
rupee will appreciate in the near future.” How would this influence your decision?
Answer: As we saw earlier, the French company’s profit when the exchange rate is 50 is
π1 = e 250 m. The advertising campaign would increase demand by 40%. Since price
remains constant, it would also increase profits by 40%, that is by an amount equal to
40% × 250 m = e 100 m. This is more than cost ($80m), so the French company should
go ahead with the campaign. The possibility of the rupee appreciating only makes things
better: it raises the Indian firm’s cost in e. which means the French firm can charge a
higher price. Not only does this increase the French firm’s profit, it also increases the
increase in profit resulting from the retailing campaign.

8.9. Karmania automobiles. There are two auto producers in Karmania, F1 and F2.
The cars they produce are essentially identical. The market inverse demand curve is given
by p = a − b Q, where p is price (in thousands of dollars); Q market output (in thousands
of units); and a and b are parameters. It is estimated that a = 25 and b = .1. Both F1 and
F2 have a marginal cost of 10 thousand dollars per car.
Competition in the Karmania auto market works as follows. At the beginning of each
year, both firms simultaneously and independently decide how many cars to produce.

64
Then the market price adjusts so that supply equals demand.
(a) Determine F1’s best response mapping.
Answer: F1’s profit is given by


π1 = p q1 − c q1 = 25 − .1 (q1 + q2 ) q1 − 10 q1

Taking the derivative with respect to q1 and equating to zero, we get

25 − .1 (q1 + q2 ) − .1 q1 − 10 = 0

which leads to
1
q1 = 75 − q2
2
which gives F1’s best response mapping.
(b) Determine the equilibrium of the game played between F1 and
F2.
Answer: Since F1 and F2 have the same marginal cost, in equilibrium q1 = q2 = q.
Substituting in the best-response mapping derived above, we get
1
q = 75 − q
2
or simply
2
q= 75 = 50
3
Regarding price, we have
p = 25 − .1 ×(50 + 50) = 15

(c) Suppose that an increase in incomes shifts demand to p = 28 −

0.1 Q. What do you expect will happen to price and the number of
cars sold?
Answer: The new first-order condition for F1 is given by

28 − .1 (q1 + q2 ) − .1 q1 − 10 = 0

leading to
1
q1 = 90 − q2
2
1 1


Again, q1 = q2 = q implies q = 90 − 2 q, 1 + 2 q = 90, or simply

2
q= 90 = 60
3
Finally,
p = 28 − .1 ×(60 + 60) = 16
We conclude that the increase in demand leads to an increase in price from 15 to 16 and
in output from 50 to 60.

65
 8.10. Ethanol. In the ethanol industry, each firm chooses what output to produce and
price is determined by aggregate output. Market demand is given by Q = 1500 − 2 p,
where Q is in million tons and p in $/ton. There are two producers and their marginal
costs are constant and given by c1 = 340, c2 = 420 (both in $/ton).
(a) Determine equilibrium price, output and market shares.
Answer: Given the nature of industry competition, the Cournot model provides a good
approximation for firm behavior. Firm 1’s profit function is given by

π1 = (a − b Q) q1 − c1 q1

Keeping in mind that Q = q1 + q2 (and thus dQ/dq1 = 1), the first-order condition for
profit maximization becomes
a − b Q − b q1 − c1 = 0 (8.5)
Likewise, for Firm 2 we have
a − b Q − b q2 − c2 = 0
Adding up these two equations gives

2 a − 2 b Q − b q2 − b q1 − c1 − c2 = 0

Since Q = q1 + q2 , this simplifies to

2 a − c1 − c2
Q= (8.6)
3b
Going back to (8.5), we get
a − c1
q1 = −Q
b
Substituting (8.6) for Q, we get
2 a − c1 2 a − c1 − c2 3 a − 3 c1 − 2 a + c1 + c2 a + c2 − 2 c1
q1 = − = =
b 3b 3b 3b
More generally,
a + cj − 2 ci
qi =
3b
Since Q = 1500 − 2 p, p = 750 − Q/2, and thus we have a = 750, b = .5, c1 = 340 and
c2 = 420. It follows that

q1 = 327 million tons

q2 = 167 million tons
s1 = 66%
s2 = 34%
p = 503 $/ton
π1 = 53.36 $billion
π2 = 13.89 $billion

66
Firm 2 is currently considering two possible strategies: (a) a public opinion campaign that
would cost $1.15 billion and shift the demand curve to Q = 1520 − 2 p; (b) a capital
investment of $4.9 billion that would reduce marginal cost c2 to 400 $/ton.
(b) Are investments (a) and (b) worthwhile in isolation? Are they
worthwhile if taken together? Justify your answer.
Answer: The shift in the demand curve corresponds to a new value of a, which is now
760. All other parameter values remain constant. Recomputing the equilibrium, we now
get π2 = 15.02, an increase of $1.13 billion. This does not compensate the $1.15 billion
investment. Recomputing profit with the lower marginal cost c2 = 400 (and the initial
value of a = 750), we get π2 = 18.69, an increase of $4.8 billion. This does not compensate
the $4.9 billion investment.
Now consider the joint investment strategy: at a cost $1.15 + $4.9 = $6.05 billion,
parameter a becomes 760, whereas c2 becomes 400. Recomputing the equilibrium, we get
π2 = 20, a gain of $6.11 billion — greater than the $6.05 billion investment.
We conclude that the whole is greater than the sum of the parts: the increase in
profits from expanding demand and reducing marginal costs is greater than the sum of the
partial increases. The idea is that, with a greater demand, the benefit from lower costs is
greater: a one dollar decrease in marginal cost applies to a larger quantity. Conversely, an
increase in demand is worth more the greater the margin the firm sells for; and a lower
marginal cost implies a higher margin.

8.11. Natural gas. Suppose there are only two natural gas producers in Kabralkstan.
In each period, firms determine how much natural gas to sell; market price is then
determined by total demand and total supply. Marginal cost is given by 77 for Firm 1 and
74 for Firm 2. Currently, Firms 1 and 2 are producing 170 and 200, respectively, whereas
market price is 94.
By making an important discovery in the process of hydraulic fracturing (or
“fracking”), Firm 2 managed to cut its marginal cost from 74 to 68.
(a) What impact do you expect Firm 2’s cost reduction to have on
its market share?
Answer: Suppose that demand is linear: p = a − b Q. Equilibrium p and Q are given by
a + c1 + c2
p=
3
2 a − c1 − c2
Q=
3b
From the first equation I get
a = 3 p − c1 − c2 = 131
From the second equation I get
2 a − c1 − c2
b= = .1
3Q
Firm i’s equilibrium output is given by
a + cj − 2 ci
qi =
3b

67
and so firm i’s market share is given by
a + cj − 2 ci
si =
2 a − c1 − c2
It follows that, initially, Firm 2’s market share is given by
131 + 77 − 2 × 74
s2 = ≈ 54%
2 × 131 − 77 − 74
With a lower cost, Firm 2’s market share becomes
131 + 77 − 2 × 68
s2 = ≈ 62%
2 × 131 − 77 − 68

Some studies suggest that Firm 2’s new production process may not be environmentally
sound.
(b) How much would Firm 1 be willing to pay in support of a cam-
paign to (successfully) prevent Firm 2 from using its new fracking
process?
Answer: Firm i’s equilibrium profit is given by
 2
1 a + cj − 2 ci
π
bi =
b 3

It follows that, if Firm 2’s cost is high, then Firm 1’s profit is given by
 2
1 131 + 74 − 2 × 77
π
b1 = = 2890
b 3

If Firm 2’s cost is low, then Firm 1’s profit is given by

 2
1 131 + 68 − 2 × 77
π
b1 =
b = 2250
b 3

It follows that Firm 1 would be willing to pay up to 2890 − 2250 = 640 to prevent Firm 2
from lowering its cost.

8.12. Wolframium. Suppose there are two producers of wolframium in the world.
Wolframium is a homogenous product. Producers set prices simultaneously and capacity
constraints are not binding at the current levels of world demand. Both producers have a
marginal cost of $900 per metric tonne. One producer is located in the US, the other one
in Mexico. Demand for wolframium is exclusively found in the US It is estimated that, at
p = $1000, world demand for woframium is 130 thousand metric tonnes per year, and that
demand elasticity is  = −.5.
(a) Suppose the government imposes an import tax of 20% on wol-
framium imports. What are equilibrium price and profits?

68
Answer: The US producer sets a price of $1080 = 900 (1 + 20%) (or infinitesimally
smaller) and receives a profit of D(1080) (1080 − 900) = 180 D(1080), where D(p) is the
demand curve. Given the information on demand and elasticity, we have
log D(1080) − log 130
−.5 =
log 1080 − log 1000
and so
log D(1080) = log 130 − .5 (log 1080 − log 1000) = 4.829
From this we estimate that D(1080) = exp(4.829) = 125.1. Finally, we conclude that the
US firm’s profit is given by 180 × 125.1 = 20736, or $20.7 bn. The Mexican firm sets a
price of 1080 and earns zero profit.
Alternatively, we can estimate demand at p = 1080 by using the percent variation
method:  
1080 − 1000
D(1080) = 130 1 − .5 = 124.8
1000

(b) Suppose a third producer enters the wolframium industry. It is

located in China and has a marginal cost of $600 per metric tonne.
What impact does this have on the equilibrium prices and profits?
Answer: The US producer now sets a price of $900. The Chinese firm sets a price of $750
(or infinitesimally smaller), so that price after import taxes is $900, beating the US firm.
The American firm makes zero profit, the Chinese company earns a profit of
D(900) (750 − 600) = 150 D(900), where D(p) is the demand curve.
An estimate of D(900) is given by
 
900 − 1000
D(900) = 130 1 − .5 = 136.5
1000
We conclude the Chinese producer earns a profit of 150 × 136.5, or $20.475 bn.

8.13. Shipbuilding. The world shipbuilding industry is dominated by three countries/

regions: Japan, Europe and China. Demand for ships is given by p = a − b Q, where b has
been estimated by industry participants to equal 0.42. Before 2006, the world quarterly
production of ships was 19 bulk carriers per quarter. The average price of a bulk carrier
was US$17.8 million. Country market shares were as follows: China 24%, Europe 8%,
Japan 68%.
(a) Assuming that the industry is well described by a Cournot
game played between countries, estimate each country’s production
marginal cost before 2006.
Answer: First we estimate the inverse demand intercept by setting

a = p + b Q = 17.8 + 0.42 × 19 = 25.78

Next we use the first order condition to get

 
X
ci = a − b Q 2 si + sj 
j6=i

69
which implies
cChina = 15.885
cEurope = 17.161
cJapan = 12.374

In 2006, China introduced a government plan to guide the development of its shipbuilding
industry. After 2006, the number of Chinese shipyards increased dramatically. The same
happened with China’s ship production rate: its market share jumped to 50%, while
Europe’s dropped to 5% and Japan’s to 45%. World Trade Organization (WTO)
agreements prohibit government industrial subsidies. Thus, complaints by WTO members
led to an investigation to find out if the Chinese government subsidized shipbuilding and if
so by how much.
(b) Suppose that a production subsidy of z implies a decrease in
China’s marginal cost from c to c − z. Use the pre- and post-2006
data to estimate z.
Answer: By repeating the same process as in the previous question, we get

cChina = 13.8

so that z = 15.9 − 13.8 = US$2.1 million, which in turn represents 13% of cost.
(c) Compute consumer surplus and profits by country. Who was
hurt and who gained from China’s production subsidies?16
Answer: With the initial cost values, we get
πChina = 6.52
πEurope = .73
πJapan = 52.35
S = 31.84
With the new cost values, we get
πChina = 21.55
πEurope = .02
πJapan = 42.78
S = 36.14
As expected, Chinese firms and consumers benefit from the subsidy, whereas Japanese and
European firms are harmed by it.

8.14. Strategic trade policy. Suppose a given country’s domestic market is supplied by
two firms competing a la Cournot: firm 1, a domestic firm, and firm 2, a foreign firm.
Demand is given by p = a − Q, where Q is total output, and marginal costs by c1 and c2 ,
where we assume ci < a (i = 1, 2). Suppose that the domestic government levies an
import tariff t to be paid by firm 2 for every unit sold in the domestic market.
(a) Determine the equilibrium values of qi for a given value of t.

70
Answer: Variable profit functions are given by

π1 = (a − q1 − q2 − c1 ) q1
π2 = (a − q1 − q2 − c2 − t) q2

From the first-order conditions, we get the best response mappings

q1∗ = 1
2 (a − c1 ) − 1
2 q2
q2∗ = 1
2 (a − c2 − t) − 1
2 q1

Solving the system of best responses, we get

1
qb1 = 3 (a + c2 + t − 2 c1 )
1
qb2 = 3 (a + c1 − 2 c2 − 2 t)

(b) Show that a small import tariff increases domestic welfare, where
the latter is defined as the sum of consumer surplus, the domestic
firm’s profit and import revenues.
Answer: Equilibrium consumer surplus is given by

Sb = 1
2 (a − pb)2

Equilibrium price can be obtained from the values of equilibrium quantities, qb:
1
pb = 1 − qb1 − qb2 = 3 (a + c1 + c2 + t)

It follows that
Sb = 1
18 (2 a − c1 − c2 − t)2
The domestic firm’s equilibrium profit is given by

π
b1 = 1
9 (a + c2 + t − 2 c1 )2

Import tariff revenues are simply t q2 , or t qb2 in equilibrium. Adding up the three terms,
we conclude that domestic welfare is given by

W
cd = Sb + π
b1 + t qb2
= 1
18 (2 a − c1 − c2 − t)2 + 1
9 (a + c2 + t − 2 c1 )2 + t
3 (a + c1 − 2 c2 − 2 t)

Taking the derivative with respect to t we get

dW
cd
1
= 3 (a − c2 ) − t
dt
which, at t = 0, is positive, given our assumption that ci < a.
(c) Show that, the more efficient the foreign firm, the greater the
increase in domestic welfare from an import tariff. Discuss.
Answer: From the previous answer, we see that, the lower c2 , the greater d W cd /dt. The
main reason is that, the lower c2 , the greater q2 is. This implies that the import tariff t

71
applies to a larger quantity. In the limit, if c2 = a, the foreign firm’s imports are equal to
zero, in which case the import tariff has no effect.
(d) Show that, if c1 is not very different from c2 , then a small import
tariff decreases world welfare, where the latter is defined as the sum of
consumer surplus, and the profits of the domestic and foreign firms.
Answer: The foreign firm’s equilibrium profit is given by

π
b2 = 1
9 (a + c1 − 2 c2 − 2 t)2

This implies that world welfare is given by

W
ct = Sb + π
b1 + π
b2
= 1
18 (2 a − c1 − c2 − t)2 + 1
9 (a + c2 + t − 2 c1 )2 + 1
9 (a + c1 − 2 c1 − 2 t)2

Taking the derivative with respect to t, and then setting c2 = c1 , we get

dW
ct
= − 49 (a − c1 ) + 11
9
dt
which, for small values of t, is negative, given our assumption that ci < a.
(e) In light of the above results, what can be an important role of
the World Trade Organization (WTO)?
Answer: The above results suggest that trade policy may have the nature of a prisoner’s
dilemma (cf Chapter 7): for each country individually, raising tariffs is a dominant
strategy, but the world as a whole is worse off when countries raise tariffs. For this reason,
a world organization that attempts to collectively reduce import tariffs may contribute to
an increase in world welfare. Long live the WTO!

8.15. Cournot with n asymmetric firms. Consider an industry with n output setting
firms, each with constantPmarginal cost ci and fixed cost Fi . Market demand is given by
p = a − b Q, where Q = ni=1 qi .
(a) Show that firm i’s best-response mapping is given by qi∗ (Q−i ) =
a−ci 1 P
2b − 2 Q−i , where Q−i ≡ j6=i qj .
Answer: Firm i’s profit function is given by

πi = (a − b Q) qi − ci qi − Fi

The first-order condition for profit maximization is given by

a − b Q − b qi − ci = 0 (8.7)
Pn
Since Q = i=1 qi , this can be re-written as
X
a−b qj − 2 b qi − ci = 0
j6=i

72
or simply
a − ci 1 X
qi∗ (qj ) = − qj
2b 2
j6=i

(b) Show that, in equilibrium,

 total output is given by
Pn

Q = n a − i=1 ci / b (n + 1) . (Hint: add up all n first-order
b
conditions for profit maximization.)
Answer: Adding up all equations (8.7), we get
n
X n
X
na − nbQ − b qi − ci = 0
i=1 i=1
Pn
Since Q = i=1 qi , this reduces to
n
X
n a − (n + 1) b Q − ci
i=1

or simply
n a − ni=1 ci
P
Q= (8.8)
b (n + 1)
 
(c) Show that equilibrium price is given by pb = a + ni=1 ci /(n + 1)
P

Answer: Substituting for Q in the demand curve, we get

p = a − bQ
n a − ni=1 ci
P
= a−
n+1
(n + 1) a − n a + ni=1 ci
P
=
Pn n + 1
a + i=1 ci
= (8.9)
n+1

(d) Show that, in equilibrium,

 firm i’s output level is given by
P

q̂i = a − n ci + j6=i cj / b (n + 1)

Answer: Substituting the expression for Q in (8.7) and solving with respect to qi , we get
1

qi = a − b Q − ci
b 
n a − ni=1 ci
P 
1
= a− − ci
b n+1
(n + 1) (a − ci ) − n a + ni=1 ci
P
=
b (n + 1)
P
a − n ci + j6=i cj
=
b (n + 1)

73
 (e) Show that, in equilibrium, firm i’s profit is given by
 2
π̂i = a − n ci + j6=i cj / (n + 1)2 / b − Fi
P

Answer: The profit function can be re-written as

πi = (a − b Q − ci ) qi − Fi

Moreover, the first-order condition can be rewritten as

a − b Q − ci = b qi

Hence, at the equilibrium point, we have

P 2
a − n ci + j6=i cj

1
π
bi = b qi2 − Fi = − Fi
b n+1

(f) Show that, in equilibrium, consumer surplus is given by

 Pn 2
2
CS = 21b n a − i=1 ci /(n + 1)

1
Answer: Consumer surplus is given by CS = 2 (a − p) Q. Substituting a − b Q for p
1

CS = a − (a − b Q) Q
2
b
= Q2
2
n a − ni=1 ci 2
 P 
1
=
2b n+1

8.16. Elasticity rule (reprise). Show that the elasticity rule derived in Chapter 3, that
is (p − MC )/p = −1/, holds under Cournot competition with linear demand and costs,
where MC is firm i’s marginal cost and  its demand elasticity (not the market elasticity).
Answer: From the definition of demand elasticity, I can write
1 dp qi
− =−
i dQ p
Therefore, the elasticity rule may be re-written as
dp
p − ci = − qi (8.10)
dQ
Let the inverse demand curve be given by p = a − b Q. From Exercise 8.8.15,
a + ni=1 ci
P
p=
n+1
which implies Pn
a − (n + 1) ci + i=1 ci
p − ci = (8.11)
n+1

74
Also from Exercise 8.15, P
a − n ci + j6=i cj
qi =
b (n + 1)
which I can re-write as Pn
a − (n + 1) ci + j=1 cj
qi = (8.12)
b (n + 1)
Substituting (8.11) for p − ci , (8.12) for qi , and b for − ddQ
p
in (8.10), we obtain an equality.
Bingo!

8.17. Efficiency loss under Cournot. Consider a market where two firms simultaneous
set quantities of a homogeneous product with demand given by Q = 37.5 − P/4. Each
firm has constant marginal cost equal to 30.
(a) Determine equilibrium output and price.
Answer: Duopolist i’s profit is given by


πi = qi p(Q) − C(qi ) = qi 150 − 4 (qi + qj ) − 30 qi

The first order condition for profit maximization is given by:

150 − 4 qi + qj − 4 qi − 30 = 0

By symmetry, we have qi = qj = q. Solving the above equation, we then get q = 10.

Moreover, p = 150 − 8 q = 70.
(b) Compute the efficiency loss as a percentage of the efficiency loss
under monopoly.
Answer: The monopoly profit function is given by

πm = Q p(Q) − C(Q) = Q (150 − 4 Q) − 30 Q

The first order condition for profit maximization is given by:

150 − 8 Q − 30 = 0

Solving with respect to Q we get Q = 15, and then p = 90.

Under perfect competition the prevailing price would be given by marginal cost:
p = 30; total quantity would be Q = 30 and welfare


p(0) − p Q
W = CS = = 1800
2
Under duopoly, total welfare is given by:


Wd = 2 π + CS = 2 q (p − c) + p(0) − p q = 1600

Under monopoly, total welfare is given by

p(0) − p Q
Wm = π + CS = (p − c) Q + = 1350
2

75
Finally, the duopoly efficiency loss as a percentage of the monopoly efficiency loss is given
by
1800 − 1600
EL = = 44.4%
1800 − 1350

8.18. Equilibrium price under Cournot. Show analytically that equilibrium price under
Cournot is greater than price under perfect competition but lower than monopoly price.
Answer: In a Cournot oligopoly, firm i’s profit is given by πi = qi P (Q) − C(qi ), where Q
is total output. The first-order condition for profit maximization is given by
dP
P (Q) + qi − MC = 0 (8.13)
dqi
The first-order condition for a monopolist is given by
dP
P (Q) + Q − MC = 0 (8.14)
dQ
Finally, under perfect competition we have

P (Q) − MC = 0

Notice that dP /dqi = dP /dQ < 0. Consider the case of oligopoly and suppose that price
is equal to monopoly price. Monopoly price is such that the (8.14) holds exactly. The only
difference between (8.13) and (8.14) is that the latter has Q instead of qi . Since Q > qi , it
follows that, for p equal to monopoly price, the left-hand side of (8.13) is positive and not
zero. Finally, if it is positive, each firm has an incentive to increase output, which results
in a lower price.
By a similar argument we can also show that price under Cournot competition is
greater than marginal cost.
See also Figure 8.6 for a graphical derivation of the same result.

8.19. Cournot with increasing marginal cost. Consider a duopoly for a homogenous
product with demand Q = 10 − P/2. Each firm’s cost function is given by
C = 10 + q(q + 1).
(a) Determine the values of the Cournot equilibrium.
Answer: Duopolist i’s profit is given by


πi = qi p(Q) − C(qi ) = qi 20 − 2 (qi + qj ) − 10 − qi (qi + 1)

The first order condition for profit maximization is given by:

20 − 2 (qi + qj ) − 2 qi − 2 qi − 1 = 0

The problem of duopolist j is symmetric, therefore we have qi = qj = 2.375 and p = 10.5.

(b) Re-compute the equilibrium values assuming that one of the firms
— say, firm 2 — has a cost function given by C = 10 + q(q + 1).

76
Answer: The first-order conditions are now given by:
20 − 2 (q1 + q2 ) − 2 q1 − 2 q1 − 1 = 0
20 − 2 (q2 + q1 ) − 2 q2 − 2 q2 − 2 = 0
Solving this system we get q1 = 2.4375, q2 = 2.1875

8.20. Cement. Two firms compete (a la Cournot) in the cement market. Demand for
cement is given by Q = 450 − 2 P . Firm 1’s marginal cost is constant at 50, firm 2’s at 40.
A technological innovation allows firms to reduce marginal cost by 6.
(a) How much would each firm be willing to pay for the innovation
if it were the only competitor to acquire it?
Answer: We saw in class that, in an asymmetric Cournot equilibrium,
1 a + c2 − 2 c1 2
 
π1 =
b 3
By the same token,
 2
1 a + c1 − 2 c2
π2 =
b 3
In the present case, we have c1 = 50, c2 = 40, a = 225, and b = 12 . (Recall that a and b are
the coefficients of the inverse demand curve, which in this case is given by p = 225 − Q/2.)
It follows that, in the initial equilibrium, π1 = 6050 and π2 = 8450.
A reduction in c1 by 6 would take firm 1’ cost down to 44. Recomputing the
equilibrium, we have π1 = 6962 and π2 = 7938. It follows that firm 1 would be willing to
pay 6962 − 6050 = 912 for the innovation.
A reduction in c2 by 6 would take firm 2’s cost down to 34. Recomputing the
equilibrium (with c1 set at its initial level), we have π1 = 5618 and π2 = 9522. It follows
that firm 2 would be willing to pay 9522 − 8450 = 1072 for the innovation. Notice that
firm 2 is willing to pay more for the new technology than firm 1.

Suppose the innovation costs 600. Consider a “metagame” where firms first
simultaneously decide whether to acquire the innovation and then compete a la Cournot
with whatever marginal cost results form the first stage.
(b) What is the equilibrium of the 2 × 2 game played by firms at the
technology choice stage?
Answer: First, we need to consider a fourth possibility not considered before: both firms
acquire the new technology, so that c1 = 44 and c2 = 34. In this case, profits are given by
π1 = 6498 and π2 = 8978. Considering all four possibilities, we construct the following
payoff matrix:
Firm 2
No Investment Investment
8450 8922
No Investment
6050 5618
Firm 1
7938 8378
Investment
6362 5898

77
Firm 1’s best response is to make the cost-reduction investment if and only if Firm 2 does
not make the investment. However, making the investment is Firm 2’s dominant strategy.
It follows that the unique Nash equilibrium is for Firm 2 to make the investment and for
Firm 1 not to make the investment.

8.21. Model calibration. Choose an industry for which you can find firm level data on
prices and market shares, and for which the Cournot model seems a good approximation.
(a) Making the necessary assumptions, estimate each firm’s marginal
cost and margin.
Answer: Consider for example the desktop computer industry. In 2010, global sales of
Windows-based desktop computers were 351 million units, whereas average price was
$605.17 The leading market shares were as follows: HP, 17.9%; Acer, 13.9%; Dell, 12%;
Lenovo, 10.9%; Asus, 5.4%; others, 39.3%. Suppose that inverse demand is given by
p = a − b Q; that each firm has constant marginal cost; and that, from previous studies, the
price elasticity of market demand is estimated to be -0.5 (at the market equilibrium level).
The price elasticity of market demand is given by
p
 = −b
Q
This implies
Q
b = −
p
Given the available data on Q and p, as well as the estimate of , we estimate b = 0.29.
Firm i’s profit function is given by

πi = p qi − ci qi

The first-order condition for firm i’s profit maximization is given by

p − b qi − ci = 0

which implies
ci = p − b si Q
since qi = si Q. Since margin is given by
p − ci
mi =
p
based on the available data for si , p and Q, as well the estimate of parameter b, we get the
following estimates of marginal costs and margins:

Firm Unit cost ($) Margin (%)

Hewlett-Packard 587 3.0
Acer 591 2.3
Dell 593 2.0
Lenovo 594 1.8
Asus 600 0.9

78
 (b) Use the estimated model to run the counterfactual whereby one
of the firm’s cost declines by 5%.
Answer: Suppose Dell’s cost declines by 5%. As I showed in the answer to Exercise 8.15,
the Cournot equilibrium price is given by

a + ni=1 ci
P
p=
n+1
Moreover, firm i’s equilibrium output is given by

a − (n + 1) ci + nj=1 cj
P P
a − n ci + j6=i cj
qi = =
b (n + 1) b (n + 1)

which in turn implies that firm i’s market share is

a − (n + 1) ci + nj=1 cj
P
si =
n a − nj=1 cj
P

In order to move any further, I need to estimate the parameter a as well as obtain the
value of n. Regarding a, recall that the inverse demand curve, p = a − b Q, implies that

a = p + bQ

Based on the available data as well as the estimate for b, I estimate that a = 707
(rounding to the dollar).
Regarding the value of n, I have the market shares for the leading 5 firms. This
leaves a total 39.3% share of the market unaccounted for. One possible assumption is that
all of the remaining firms are as big as the smallest firm for which I have data. This
implies an estimate of 39.3/5.4=7.27 firms. Rounding to the next highest integer I get 8,
or a total of 13 firms. (Note that the number of firm cannot be smaller than this. Why?)
We are now ready to proceed with the “comparative statics” calculations. A 5%
decrease in Dell’s cost implies that the new value of cD is 563.4. The estimated values of p
and sD before and after the cost decrease are as follows.

Data Initial cost Lower cost

Price 605 604.8 602.4
Dell’s market share 12.0 11.7 37.9
9 Collusion

9.1. Tacit collusion.Industries A and B can be characterized by a series of parameters:

n, the number of firms, is 8 in both industries; r, the annual interest rate, is 10% in both
industries; f , the frequency with which firms interact (number of times per year), is 1 in
industry A and 12 in industry B; g, the industry growth rate, is 10% in industry A and
−30% in industry B; finally, h, the likelihood that the industry will continue in existence
into the next period, is 80% in industry A and 100% in industry B.
In which of the two industries do you think tacit collusion is more likely to take
place? Briefly justify your answer.
Answer: The effective discount factor is given by
h (1 + g)
δ=
1 + r/f
Substituting the above values, we get δA = 0.8 and δB = 0.694. Based on these
calculations, and considering the number of firms is the same in both industries, we would
expect tacit collusion to be easier in industry A. Although interaction between firms is
more frequent in industry B, the fact that it is a declining industry ultimately renders
collusion very difficult: the promise of continuing collusion in the future is of little
importance, leading firms to have a greater incentive to cheat on a collusion agreement.

9.2. n firm oligopoly. Consider a price-setting oligopoly with n firms, all with constant
marginal cost c. Suppose market demand is given by D(p) and the discount factor is .8.
Determine the maximum number of firms such that there exists an equilibrium with
monopoly pricing.
Answer: The no-deviation constraint is given by
1 πM
≥ πM
n 1−δ
or simply
1
n≤ =5
1−δ

9.3. Repeated interaction. Explain why collusive pricing is difficult in one-period

competition and easier when firms interact over a number of periods.

80
Answer: Refer to Section 9.1. In one-period competition, each firm has a strong incentive
to deviate from the pre-agreed collusive price, since the gains from deviating are higher
than the losses. In terms of the example in Section 9.1, had the duopolists interacted in
only one period, the gain would be given by one half of monopoly profits, while the loss
from deviating would be 0. We would then be led to the usual Nash-Bertarand
equilibrium when both firms price at marginal cost.
If, however, firms interact over a number of periods, history, in the form of past
pricing behavior, becomes important. Deviation from the collusive price in one period can
be met with punishment (deviation) in future periods. Hence, the original defector must
weigh short-term gains against long-term losses, made possible exactly by multi-period
interaction.

9.4. Airbus and Boeing. Boeing and Airbus seem to cycle between periods of severe
price competition and pledges that they will not sink into another war. Based on the
analysis of Section 9.1, why do you think it is so difficult for aircraft manufacturers to
collude and avoid price wars?
Answer: Refer to Section 9.1. Aircraft manufacturers receive orders infrequently.
Moreover, the terms of each sale are seldom made public. For these reasons, it is very
difficult for them to collude. The incentive to cheat on a tacit or explicit agreement would
be very high because: (a) the short run is very important with respect to the long run
(low discount factor); (b) the probability that cheating would be detected is low.

9.5. Price wars. “Price wars imply losses for all of the firms involved. The empirical
observation of price wars is therefore a proof that firms do not behave rationally.” True or
false?
Answer: False. As Section 9.2 shows, price wars may be part of the equilibrium of a game
played between rational firms.

9.6. Airline fare wars I. Empirical evidence from the US airline industry suggests that
fare wars are more likely when carriers have excess capacity, caused by GDP growth
falling short of its predicted trend. Fare wars are also more likely during the Spring and
Summer quarters, when more discretionary travel takes place.18 Explain how these two
observations are consistent with the theories presented in Section 9.2.
Answer: Refer to Section 9.1 and 9.2. The first model in Section 9.2 (secret price cuts)
predicts that price wars start in periods of unexpected low demand. This is consistent
with the first observation above. However, the effect of unexpected low demand is also
consistent with a theory of price wars caused by financial distress (see the end of Section
9.2). The observation that fare wars take place during periods of higher demand is
consistent with the second model in Section 9.2 (demand fluctuations).

9.7. Airline fare wars II. A 1998 news article reported that

Delta Air Lines and American Airlines tried to raise leisure air fares 4% in
most domestic markets, but the move failed Monday when lone-holdout
Northwest Airlines refused to match the higher prices.

81
 The aborted price boost illustrates the impact Northwest’s woes already are
having on the industry. Months of labor unrest ... are prompting passengers to
book away from the fourth largest carrier.19

What does this say about the nature of price dynamics in the airline industry?
Answer: The observation seems consistent with the theory that financially distressed
firms, which effectively have a lower discount factor δ, render collusive agreements more
difficult to sustain.

9.8. Paper products. In the third quarter of 1999, most North American paper and
forest-products companies experienced an improvement in their results. The industry,
analysts said, was in a cyclical upswing: not only was demand increasing at a moderate
pace; more important, the industry practiced restraint in keeping low production levels,
thus providing support for higher prices.20
How do you interpret these events in light of the models presented in Section 9.2?
Answer: The analysis of Section 9.1 predicts that collusion is easier in growing industries
(the promise of future profits under collusion is worth more). This is consistent with the
fact that “restraint in keeping low production levels” took place during the “cyclical
upswing.”

9.9. Export cartels. In 1918, the US Congress passed a low allowing American firms to
form export cartels. Empirical evidence suggests that cartels were more likely to be
formed in industries where American exporters had a large market share, in
capital-intensive industries, in industries selling standardized goods, and in industries that
enjoyed strong export growth.21 Discuss.
Answer: The effect of export growth seems consistent with the analysis in Section 9.1.
The effect of standardization may correspond to the fact that it is easier to monitor
collusion with a standardized product (however, the effect of product differentiation on
collusion is a controversial issue). The effect of market share is consistent with the
analysis in Section 9.3 (concentration facilitates collusion).

9.10. Cartel leniency. Many antitrust authorities throughout the world have
implemented leniency programs targeted at busting secret cartels. These programs offer
immunity from prosecution to firms who blow the whistle on their co-cartel conspirators.
These programs have proven extremely successful: in the US, from 1993 to 2000 the total
amount of fines for anti-competitive behavior increased by twentyfold.
Show how the leniency programs first implemented by the US Department of Justice
and later replicated in many countries change the rules of the game played between firms
in a secret cartel.
Answer: Consider an industry with two firms who secretly have been colluding to fix
prices. Absent a leniency program, their dominant strategy is not to reveal any
information to the anti-cartel authorities. In fact, by doing so a firm would increase the
probability of being prosecuted and found guilty — and receive a total payoff that is lower
than its payoff from continuing to collude. The worst that can happen is for its rival to

82
reveal information to the cartel office, but then its ultimate payoff does not depend on its
action. Overall, mum is the word and the optimal strategy.
Now suppose that the leniency program is in effect. If my rival is going to come
forward, then I am better off beating them to the cartel office. If in addition there is some
probability that the cartel office will find out about my secret agreement from some other
source, then even if my rival were not to go to the cartel office I would be better off by
going to the cartel office to the extent that this will save me from punishment. All in all,
the game may turn into a prisoner’s dilemma: although both firms would prefer to
continue in the cartel agreement point, they both have as a dominant strategy to reveal
information to the cartel office.

9.11. Corporate leniency. A study of the European corporate leniency program shows
that the likelihood of a firm becoming the chief witness increases with its character as
repeat offender, the size of the expected basic fine, the number of countries active in one
group, as well as the size of the firm’s share in the cartelized market.22 Are these results
consistent with the discussion of these programs in the text?
Answer: Repeat offenders suffer bigger fines. This increases the incentive to be the
leading witness, thus explaining the first result. The same argument applies to the size of
the expected fine. The greater the number of countries involved, the greater the
“prisoner’s dilemma” effect: if I don’t report, someone else will, so I might as well report.

9.12. Ivy League. The endowments of the Ivy League universities have increased
significantly in recent decades. This wealth notwithstanding, for years the universities
managed to refrain from using financial incentives as a means to compete for students: the
manual of the council of Ivy League Presidents stated that the schools should “neutralize
the effect of financial aid so that a student may choose among Ivy Group institutions for
non-financial reasons.” In 1991, the Justice Department argued that this amounted to
price collusion and forced the agreement to end. However, no significant price competition
took place until 1998, when Princeton University started offering full scholarships for
students with incomes below $40,000. Stanford, MIT, Dartmouth and Cornell followed
suit. Allegedly, Harvard sent a letter to accepted 1998 applicants stating that “we expect
that some of our students will have particularly attractive offers from the institutions with
new aid programs, and those students should not assume that we will not respond.”e
How do you interpret these events in light of the theories discussed in this chapter?
Answer: If the Department of Justice was right in assuming the council manual’s clause
was an explicit form of price collusion, then what happened after 1991 is that collusion
ceased to be explicitly supported by the clause and turned into tacit collusion. In fact, the
analysis in Chapter 9 suggests that explicit, contractual arrangements are not necessary to
sustain a collusive agreement. The chapter also states that, under tacit collusion, each
firm balances the short-run benefits from deviation against the long-term cost of entering
into non-cooperative play. The fact that endowments have increased so much (especially
Princeton’s) may be what has tipped the balance in the direction of giving away full
scholarships.
e . The Economist, December 5, 1998.

83
 9.13. Spanish hotels. Based on data from the Spanish hotel industry, it was estimated
that the rate set by hotel i in market k is positively influenced by a variable that measures
the intensity of multimarket competition between hotel i and its competitors in market k:
the more markets m 6= k in which firm i and its competitors meet, the greater the measure
of multimarket contact. It was also observed that the measure of multimarket contact is
highly correlated with hotel chain size, that is, the larger hotel i’s chain, the greater the
measure of multimarket contact for firm i.23
Provide two interpretations for the positive coefficient of multimarket contact on
hotel rates, one based on collusion, one based on a different effect.
Answer: When interaction between oligopolists takes place over a number of periods, it is
easier to sustain collusion: long-term losses weigh more compared to short-term gains
from deviation. Multimarket contact adds another “dimension” to the balance between
gains and losses. A firm’s gain from deviating in one market may be punished by its
competitors in all the markets they meet, making the potential cost from deviation higher.
However, the optimal behavior of the deviating firm would call for deviation in all
markets. Thus, we have higher losses from deviating but also higher gains. As discussed in
Section 9.3, if everything is identical (firms, markets) then multimarket contact does not
increase the likelihood of collusion because the potential gains from deviation increase in
the same proportion as the losses. However, asymmetries between firms or markets can
make losses weigh more than gains, thus increasing the likelihood of collusion. This
justifies the positive correlation between multi-market contact and average rates.
There is, however, an alternative interpretation. Maybe rates are higher in hotels of
greater size. This could happen either because consumers attach a greater value to hotels
that have larger chains or because bigger hotel chains command greater (unilateral)
market power. Given the empirical correlation between hotel size and multi-market
contact this would also imply a correlation between multi-market contact and rates, even
if there is no implicit or explicit collusion between hotel chains.

9.14. Railroads. In 1986, the US Congress enacted a regulation (PL99-509) requiring

railroads to disclose contractual terms with grain shippers. Following the passing of the
legislation, rates increased on corridors with no direct competition from barge traffic,
while rates decreased on corridors with substantial direct competition.24 How do you
interpret these events?
Answer: One possible interpretation for these results is that, when there is no competition
to railroad shipping, there is potential for collusion among railroad operators, whereas the
opposite is true when there is direct competition from barge traffic. In this context,
increased information about railroad contracts has the effect of

1. improving collusion among railroad operators when the latter have no competition.
This is consistent with the idea that when price cuts are difficult to observe collusion
is more difficult to sustain.
2. increasing competition in markets where railroad operators compete with barge
operators. This is consistent with the idea that, in a competitive environment,
better information about prices increases demand elasticity (consumer are more
aware of price differences) and thus decreases margins.

84
 9.15. Multi-market contact. Consider the model of multi-market contact presented in
Section 9.3: Firm 1 has cost c in Market 1, while Firm 2 has a cost c. The situation is
reversed in Market 2. Demand is the same in both markets: consumers are willing to buy
q units for a price of up to pM (that is, for p ≤ pM , quantity demanded does not depend
on price). It is assumed that c < c < pM . In each period, firms set prices in both markets
simultaneously.
Determine the minimum value of the discount factor such that the optimal collusive
solution is stable.
Answer: As discussed in Section 9.3, the efficient collusive agreement is the following: In
each market, the firm with a cost advantage sets monopoly price, while the other sets a
higher price and sells 0.
If firms price according to equilibrium strategies, then each firm gets 0 in the
“foreign” market and q (pM − c) in the “domestic market” in each period. The net present
values of these profit streams is given by q (pM − c) /(1 − δ).
If a firm deviates from the equilibrium strategies and undercuts the rival in the rival’s
“domestic” market, then it gets a total profit of q (pM − c) + q (pM − c) in the first period.
Thereafter, it only makes a profit in the domestic market, specifically q (c − c) per period.
It follows that the crucial equilibrium condition is given by

q (pM − c) q (c − c)
≥ q (pM − c) + q (pM − c) + δ
1−δ 1−δ
Solving for δ, I get
1
δ≥
2

9.16. Price-matching guarantees.. In some industries firms offer price-matching

guarantees (also known as meet-the-competition clauses): if some rival offers a price lower
than firm i, then firm i is forced to offer the same price to all of its customers. Consider a
price-setting, homogenous-product oligopoly and suppose that all firms have constant
marginal cost c and offer the same price-matching guarantee. Specifically, firms
simultaneously set prices (as in the Bertrand model); and if some firm sets a lower price
than other firms, then all firms must match that price. Show that any price between cost
and monopoly price can be obtained as the play of a Nash equilibrium.
Answer: Suppose that all firms set p ∈ [c, pM ]. Given these prices, each firm makes
positive profits. If a firm unilaterally sets a higher price, then its profits drop to zero. If a
firm unilaterally sets a lower price, then all firms must match that price, and all firms
(including the price leader) earns lower profits. We conclude that pricing at p is a Nash
equilibrium.

9.17. Number of competitors. Consider an n firm homogeneous-good oligopoly with

constant marginal cost, the same for all firms. Let δ̄ be the minimum value of the discount
factor such that it is possible to sustain monopoly prices in a collusive agreement. Show
that δ̄ is decreasing in n. Interpret the result.
Answer: Let π M be total industry profits. Under the collusive agreement, each firm
receives π M /n. If one of the firms undercuts its rivals, then it gets approximately π M .

85
Finally, if firms revert to a (perpetual) price war each firm gets zero. It follows that the
condition such that it is an equilibrium for firms to price at the monopoly level is given by
1 πM
≥ πM
1−δ n
Solving with respect to δ we get
n−1
δ≥
n
It follows that collusion is stable if and only if δ > δ̄ ≡ n−1
n . (Note that the condition is
independent of the value of π M , so the same condition would apply for any level of
collusion.)
Taking the derivative of δ̄ with respect to n, we get

d δ̄ n − (n − 1)
= = 1/n2 > 0
dn n2
It follows that δ̄ is increasing in n. In words, the more firms there are, the more difficult it
is to sustain a collusive agreement. The idea is that the relative gain from cheating is
greater the greater the number of firms (the profit from cheating is always the same, but
the profit from collusion is lower the greater n is).

9.18. Two markets. Consider the model of multi-market contact presented in Section
9.3. Determine the minimum value of the discount factor such that the optimal collusive
solution is stable.
Answer: The setting of the problem consists of Firms 1 and 2, and Markets A and B.
Firm 1 has cost c in Market A, while Firm 2 has a cost of c. The situation is reversed in
Market B. Demand is the same in both markets. It is assumed that c < c < pM .
As discussed in section 9.3 the efficient collusive agreement is the following: In each
market, the firm with a cost advantage sets the monopoly price, while the other sets a
higher price and sells 0. Let us use the following notation: π M represents the monopoly
0
profit of the firm with cost advantage, π M = π(pM − , c̄) is the profit of the firm with high
marginal cost when it charges (slightly less than) the monopoly price and π C = π(c̄, c) is
the profit of the firm with low cost when it charges a price equal to the other firm’s costs.
In the efficient collusive agreement each firm gets:
1
π M + δ π M + δ 2 π M + ... = π M (1 + δ + δ 2 + ...) = πM
1−δ
If a firm decides to deviate, it will do so only in the market where it has a cost
disadvantage, since in the other market it already earns monopoly profits. Suppose that
the punishment for deviation is for both firms to engage in a price war so that the
prevailing price in each market is c. If Firm 2 deviates in Market A, then it gets π M C in
that market in the first period, plus 0 from then on; and π M + δ π C + δ 2 π C + ... in the
other market. The situation is symmetric. Therefore, the deviating firm’s total profits are
given by:
δ
π M C + π M + δ π C + δ 2 π C + ... = π M C + π M + πC
1−δ

86
The stability condition is the following:
1 δ
πM ≥ πM C + πM + πC
1−δ 1−δ
This gives the minimum value for the discount factor:

πM C
δ=
πM C + πM − πC

9.19. Secret price cuts. This exercise formalizes the model of secret price cups
presented in Section 9.2.25 Suppose that all consumers are willing to pay u for the
(homogeneous) product sold by two duopolists. In each period, demand can be high
(probability 1 − α) or low (probability α). When demand is high, h = 1 units can be sold
at price u (or any lower price). When demand is low, only l = 0 < h units can be sold.
The probability that demand is high or low in each period is independent of what it was
in the previous period. Moreover, firms are unable to observe the state of market demand;
all they can observe is whether their own demand is high or low. Finally, for simplicity,
assume that production costs are zero.
Consider the following equilibrium strategies. Firms start by setting p = u. If they
receive a positive demand (namely, 21 ), then they continue to set p = u, that is, they
remain in the “co-operative phase”. If however one of the firms (or both) receives zero
demand, then both firms enter into a “price war:” they set p = 0 during T periods and,
after this period, revert to p = u again (the co-operative phase).f Let V be the net present
value of a firm in equilibrium (starting in a period where collusion takes place).
(a) Show that
u 
V = (1 − α) +δV + α δ T +1 V
2

Answer: The first term on the right-hand side corresponds to the case when demand is
high (probability 1 − α), whereas the second term corresponds to the case of low demand
(probability α). If demand is high, then each firm receives current profits of u2 . Moreover,
beginning next period, their continuation expected payoff is V , for there is no reason to
start a price war. If, however, demand is low (probability α), then it is common
knowledge that at least one of the firms receives zero demand, and that a price war will
start in the next period. As a result, firms receive zero profits today (because demand is
zero) and zero profits in the next T periods (because they engage in a price war). After
these T periods, firms revert to the co-operative phase, so that their continuation
expected payoff from then on is V .
(b) Show that the best a firm can do by deviating is to get

V 0 = (1 − α) u + δ T +1 V

f . Notice that, if a firm receives zero demand, then it is common knowledge that a price war is going
to start, that is, it is common knowledge that one of the firms receives zero demand. In fact, either
demand is low, in which case both firms receive zero demand, or one of the firms deviates from p = u,
in which case the deviating firm knows that the rival receives zero demand.

87
Answer: In words, this equation tells us that, if demand turns out to be high (probability
1 − α), then setting a slightly lower price gives the deviator a current profit of u (as
opposed to u2 ). However, regardless of what the state of demand is today, firms will
certainly enter in a price war beginning in the next period. In fact, regardless of the state
of demand, the rival firm (the non-deviator) receives zero demand today, the condition
that triggers a price war. For this reason, expected future discounted payoff is simply
δ T +1 V .
(c) Show that the condition that the prescribed strategy constitutes
an equilibrium is given by

1 ≤ 2 (1 − α) δ + (2 α − 1) δ T +1

Answer: The condition is given by V ≥ V 0 . It simplifies into the condition in the text.
(d) What is then the optimal equilibrium?
Answer: If α < 12 , then the above condition is equivalent to T ≥ T , where T is a positive
number. What is then the optimal equilibrium? (Note that optimality here is understood
within the class of equilibria we are considering. It is possible to find collusive equilibria
that perform better than the ones considered here.) In other words, what is the optimal
value of T ? From the equation for V , we can see that the equilibrium discounted payoff is
decreasing in the value of T . This is intuitive, for the greater the value of T the longer the
price wars will be; and firms don’t like price wars. Therefore, the optimal value of T is the
lowest value such that the equilibrium is stable, that is, the lowest value such that the
condition V ≥ V 0 holds, that is, T = T .g

9.20. Demand fluctuations. This exercise formalizes the model of demand fluctuations
considered in Section 9.2.26 The new model is similar to the model in Exercise 9.19, with
the difference that we now assume that in each period, before setting prices, firms observe
the state of demand.h We also make the simplifying assumption that α = 12 , that is, the
high- and the low-demand states are equally likely.
(a) Show that, if the discount factor is sufficiently large, specifically,
if δ > 3 2h+l
h
, then there exists an equilibrium where firms set monopoly
price in every period (similarly to Section 9.1).
Answer: In this equilibrium, firms set p = u in every period, regardless of the state of
demand. In fact, if a firm decides to go along with the agreed-upon equilibrium strategies,
its expected discounted profits are
 
1 δ 1 1 1
ud + uh + ul
2 1−δ 2 2 2

where d = h or d = l depending on whether today’s demand is high or low, respectively.

Assuming that a deviation implies that firms switch to setting price equal to marginal
cost, the payoff from deviation is simply ud, where, again, d = h or d = l. The conditions
g . We are ignoring here the fact that T must be an integer.
h . Firms can also observe past decisions by rival firms.

88
that setting p = u is an equilibrium are therefore
 
1 δ 1 1 1
uh + uh + ul > uh
2 1−δ 2 2 2
 
1 δ 1 1 1
ul + uh + ul > ul
2 1−δ 2 2 2

or simply
2
δ > ≡δ (9.15)
3 + l/h
2
δ > ≡δ (9.16)
3 + h/l

Notice that, since h > l, (9.15) implies (9.16). This is intuitive: the temptation to cheat
on the agreement and set a slightly lower price is especially strong in periods of high
demand. The condition for stability of the full-collusion agreement is therefore (9.15).
(b) Suppose now that the discount factor δ is lower than, but close
to, the threshold derived in the previous answer. Show that, while
there exists no equilibrium where firms set monopoly price in every
period, there exists an equilibrium where firms set monopoly price
during periods of low demand and a lower price during period of high
demand.
Answer: If δ is lower than, but close to, δ, then full collusion cannot be an equilibrium.
However, suppose that firms decide to set price equal to ph < u during periods of high
demand and pl = u during periods of low demand. The condition for no-deviation during
periods of high demand is now given by
 
1 δ 1 1 1
ph h + ph h + u l ≥ ph h
2 1−δ 2 2 2

or simply
δl/h
ph ≤ u
2 − 3δ
Since firms would like to set prices as high as possible, we get

δl/h
ph = u (9.17)
2 − 3δ

Substituting δ for δ in (9.17), we get, as expected, p = u. Any value δ < δ yields ph < u.
Assuming that δ is close to δ, ph will in turn be close to u. This implies that the
constraint (9.16) is still satisfied. We thus conclude that setting p = ph in periods of high
demand and p = u in periods of low demand is indeed an equilibrium.
10 Market Structure

10.1. Market size and market structure. Explain in words why the number of firms in
a free-entry equilibrium may be less than proportional to market size.
Answer: The explanation lies in the fact that as the number of firms increases, so does
competition. As a result, prices will fall, reducing the margin, p − c. Therefore, variable
profit per unit of market size decreases, making the number of firms the market can
sustain increase less than proportionally to market size.

10.2. Single market. Suppose that two countries, initially in autarchy, decide to create
a single market. For simplicity, assume that, in both economies, there is only one product.
Demand for this product is given by Di = Si (a − pi ), (i = 1, 2), where Si is a measure of
country i’s size. Upon the creation of a single market, total demand is given by the
horizontal sum of the two initial demands.
Assuming there is free entry and that firms compete a la Cournot, determine the
equilibrium number of firms in autarchy and after the completion of the single market.
Interpret the results.
Answer: In autarchy we have pi = a − Di /Si . Assuming that the cost function takes the
form C(qik ) = F + c qik the equilibrium number of firms in market i is given by (10.2),
that is,
a−c 2
 
Π(ni ) = Si −F
ni + 1
The equilibrium number of entrants is given by (10.4), that is,
" r #
Si
bi = (a − c)
n −1
F

After the completion of the single market the size of the market increases and demand
becomes
D1+2 = D1 + D2 = (S1 + S2 ) (a − p)
We now have " r #
S1 + S2
b1+2 = (a − c)
n −1
F

90
Note that, if the values of n are large, then
r r
S1 S2
n b2 ≈ (a − c)
b1 + n + (a − c)
F F
a − c p p 
= √ S1 + S2
F
a − c p 
> √ S1 + S2
F
≈nb1+2

In words, n1+2 is lower than n1 + n2 : the free trade agreement implies that some firms will
exit. The explanation for this effect is the same as in Exercise 10.1.

10.3. California and Montana.. The number of imported automobiles in California is

four times higher than in Montana, in per capita terms. The population of Californian is
mainly urban, whereas the population of Montana is mainly rural. How do demographic
differences and the model presented in Section 10.1 explain the differences in consumption
patterns?27
Answer: One possible explanation is related to the answers to Exercises 10.1 and 10.2. As
Exercise 10.2 shows, when two markets get merged, the total number of firms increases
(though by less than simply adding the number of firms). A larger number of firms
implies a lower price. This in turn implies that quantity demanded is higher.
Comparing two states with different degrees of urbanization is akin — to a certain
extent — to analyzing the effects of creating a single market out of several markets: an
urbanized state is like a state with rural population where many small markets were
turned into a single, large market. In this sense, one would expect urban states to have
more competitive markets, thus with lower prices and a higher consumption on a per
capita basis.
In other words, theory predicts that smaller markets will have fewer firms and higher
margins. The fact that the population of Montana is mainly rural implies that the typical
market for a car dealer is smaller than in California.

10.4. Market size and market structure. In some industries, the number of firms
increases as market size increases. In other industries, the number of firms seems
remarkably stable despite changes in market size. Discuss.
Answer: Refer to the earlier discussion on the difference between endogenous and
exogenous entry costs.

10.5. Retail in Switzerland. Retail in Switzerland is mostly dominated by highly

profitable cartels. The Swiss authorities anticipate the gradual collapse of these cartels as
the country becomes better integrated with the rest of Europe. OECD, by contrast, hold
a more skeptical view, claiming that the collapse of cartels does not necessarily lead to
more competitive markets; rather, they add, cartel breakdowns are frequently associated
with an increase in concentration. Which prediction seems more reasonable? Are the two
views inconsistent?

91
Answer: Integration is likely to imply greater competition from foreign suppliers. Lower
margins will then imply that the Swiss market cannot hold the same number of firms as
currently. It is therefore possible that the two predictions hold true: that prices go down
and that the industry becomes more concentrated.

10.6. Market definition and market structure. Consider the following goods: cement,
mineral water, automobiles, retail banking. In each case, determine the relevant market
boundaries and present an estimate of the degree of concentration.
Answer: This is an open answer question.

10.7. Cost reduction and the Herfindahl and Lerner Indexes. Consider an industry
where demand has constant price elasticity and firms compete in output levels. In an
initial equilibrium, both firms have the same marginal cost, c. Then Firm 1, by investing
heavily in R&D, manages to reduce its marginal cost to c0 < c; a new equilibrium takes
place.
(a) What impact does the innovation have on the values of H and
L?
Answer: Initially, H = 12 . In the new equilibrium, we have a duopoly with different
marginal costs, which implies H > 12 . Specifically, let s be the larger firm’s market shares.
Then
H = s2 + (1 − s)2 = 1 − 2 s (1 − s)
This is increasing in s, attaining a maximum when s = 1 (monopoly) and a minimum
when s = 12 . Since L = H/(−) and  is constant, it follows that L increases as a result of
the innovation
(b) What impact does the innovation have on consumer welfare?
Answer: From our treatment of the Cournot model, we know that a decrease in one of the
firms’ cost leads to a lower price, thus higher consumer surplus.
(c) What do the previous answers have to say about L as a perfor-
mance measure
Answer: L may not be a good indicator of consumer welfare when the comparative statics
corresponds to changes in cost levels. By contrast, if cost levels remain constant and there
is only variation in the number of firms or the price elasticity of demand, then L provides
a good indicator of how consumer welfare varies.

10.8. Barriers to entry and welfare. “Barriers to entry may be welfare improving.”
What particular industry characteristics might make this statement valid?
Answer: Following the discussion in section 10.4 we said that free entry is decreasing
welfare when the business stealing effect dominates. For this to happen, as in the example
with retail banking, ”products” should be homogenous (product differentiation is
unimportant) and competition should be soft. In this case, paying a fee for setting up a
branch represents a barrier to entry and may act as an efficient means of blocking
excessive entry.

92
 10.9. Number of competitors and equilibrium profits. Derive Equation (10.2).
Answer: Let the inverse demand curve be given by

P = a − Q/S

S is a measure of market size. If the only difference between market A and market B is
that the latter is twice the size of the first, then the demand curve in market B is as in A
but with a value of S that is twice that of market A’s.
Each firm’s profit, in case it enters the market, is given by

Π = P qi − F − c qi = (a − Q/S − c) qi − F

The first-order condition for profit maximization is

a − Q/S − c − qi /S = 0

In a symmetric equilibrium, we have qi = q = Q/n, where n is the number of active firms.

It follows that
a − nq/S − c = q/S = 0
or
a−c
q= S
n+1
Equilibrium price is given by
p = a − nq/S
a−c
=a−n
n+1
Substituting in the profit function, we get
Π(n) = (p − c) q − F
 
a−c a−c
= a−n −c S −F
n+1 n+1
a−c a−c
= S −F
n+1 n+1
 2
a−c
=S −F
n+1

10.10. Market structure and market power under Cournot competition. Derive
equation (10.7).
Answer: Firm i’s profit is given by

πi = p qi − Ci

where p = P (Q), P (·) is the inverse demand curve, and Q is total output. The first-order
condition for profit maximization is given by
dp
qi + p − MC i = 0
dqi

93
 P
Since Q = i qi , ∂ p/ ∂ qi = ∂ p/ ∂ Q. We can thus rewrite the above equation as

dp
p − MC i = − qi
dQ
The industry Lerner index is defined as
n
X p − MC i
L≡ si
p
i=1

Since si ≡ qi /Q and p − MC i = − ddQ

p
qi , this can be re-written as

n dp n n
X qi − d Q qi dp Q X qi2 1 X H
L= =− 2
= − dQ p
s2 =
Q p dQ p Q −
i=1 i=1 dp Q i=1

10.11. Scale economies. Show that the coefficient of scale economies, AC /MC , is
greater than one if and only if average cost is decreasing.
Answer: Note that
 
C
d AC d q 1 dC 1 1

= = −C 2 = MC − AC
dq dq q dq q q

Therefore, if MC > AC then d AC d AC

d q > 0; and if MC < AC then d q < 0.

10.12. Technology and market structure. Consider an industry with market demand
Q = a − p and an infinite number of potential entrants with access to the same technology.
Initially, technology is given by C = F + c q. A new technology allows for a lower marginal
cost, c0 < c, at the expense of a higher fixed cost, F 0 > F .
(a) What can you say about the effect of the new technology on
equilibrium price?
Answer: For a given number of firms, a lower marginal cost (c0 < c) implies a lower
equilibrium price. However, to the extent that F 0 > F , it is also possible that the number
of firms in a free-entry equilibrium declines, which in turn implies a higher equilibrium
price. As a result, one cannot say in general whether equilibrium price increases or
decreases.
Specifically, suppose that firms compete a la Cournot and assume that the value of n
is sufficiently large that we can ignore the integer constraint (that is, the fact that the
number of firms must be an integer number). In equilibrium, the number of active firms is
given by r
1
n = (a − c) −1
F
For a given n, equilibrium price is given by
a + nc
p=
n+1

94
We can therefore define “iso-price” curves by the expression
q
a − c + c (a − c) F1 √
p= q =c+ F
(a − c) F1

(b) Suppose that a = 10, F = 2, F 0 = 3, c = 2, c0 = 1. Determine

equilibrium price under each of the two technologies.
Answer: Let n1 and n2 be the equilibrium values of n under the old and the new
technologies, respectively. In general,
" r #
S
b = (a − c)
n −1
F

It follows that
" r #
1
b1 = (10 − 2)
n −1 =4
2
" r #
1
b2 = (10 − 1)
n −1 =4
3

It follows that equilibrium price is given by

a + nc 10 + 4 × 2
p1 = = = 3.6
n+1 5
a + nc 10 + 4 × 1
p2 = = = 2.8
n+1 5

10.13. Alternative production technologies. Consider an industry with a homogeneous

product where firms set output (or capacity) levels and price is determined by total output
(or capacity). Suppose there is a large number of potential entrants and that each firm
can choose one of two possible technologies, with cost functions Ci = Fi + ci qi (i = 1, 2).
(a) Derive the conditions for a free-entry equilibrium.
Answer: From Exercise 8.8, we know that equilibrium profits in an asymmetric
Cournot-like oligopoly are given by
P 2
1 a − n ci + j6=i cj

π
bi = − Fi
b n+1

Suppose there are two groups of firms: n1 firms choose technology 1 and n2 firms choose
technology 2, where n = n1 + n2 . Then

1 a − n ci + (ni − 1) ci + nj cj 2
 
π
bi = − Fi
b n+1

where j 6= i.

95
 In a free-entry equilibrium, for type 1 and type 2 firms must make positive profits.
Moreover, a potential entrant would need to make negative profits were that entrant to
choose either of the two available technologies. We thus have four different constraints.

1 a − n ci + (ni − 1) ci + nj cj 2
 
− Fi ≥ 0
b n+1
1 a − (n + 1) ci + ni ci + nj cj 2
 
− Fi ≤ 0
b n+2

(b) Show, by means of a numerical example, that there can be more

than one equilibrium, with different numbers of large and small firms.
Answer: Suppose that a = 100, b = 1, F1 = 15, c1 = 0, F2 = 12, c2 = .5. The following pairs
(n1 , n2 ) satisfy the above equations: (5,22); (24,0); (3,24).

10.14. Herfindahl index bounds. Suppose you only know the value of the market shares
for the largest m firms in a given industry. While you do not possess sufficient information
to compute the Herfindahl index, you can find a lower and an upper bound for its values.
How?
Answer: A lower bound would result from an industry where, in addition to the top m
firms, there is a very large number of firms with a Pvery small market share. In the limit of
infinitesimal shares, the value of H would be H = m 2
i=1 si . An upper bound would result
from an industry where all the remaining firms havethe same market  share as the m-th
Pm 2 Pm
firm. The value of H would then be H = i=1 si + 1 − i=1 si sm . (Notice that the
 
remaining firms would be 1 − m
P
i=1 si /sm in number.
The above lower and upper bounds are frequently very close, so a fairly good
approximation is often possible.

10.15. Product differentiation and market structure. Consider the monopolistic

competition model, presented in Section 4.3. What is, according to this model, the
relation between the degree of product differentiation and market structure?
Answer: Refer to Chapter 4. The greater the degree of product differentiation, the steeper
the demand curve d faced by each firm. In the long run, price equal average cost.
Therefore, the steeper d is the lower each firm’s output is in the long run equilibrium. We
would therefore expect a more fragmented market structure when the degree of product
differentiation is higher.

10.16. Doctors and plumbers. Consider the structure of geographically isolated

markets in the US (small towns) in the following businesses: doctors, dentists, plumbers.
It can be shown that the minimum town size that justifies the entry of a second doctor is
approximately 3.96 times the required size for the first doctor to enter. For plumbers, the
number is 2.12. How can these numbers be interpreted?21
Answer: The higher number for doctors has two interpretations. The first one is that
competition between two doctors is very intense, so that it would take a much larger
market before the second doctor could recoup entry costs. The second interpretation is

96
that there are specific barriers to entry by a second doctor which are not present in the
case of a plumber.

10.17. Advertising costs. Consider the following model of entry into an

advertising-intensive industry. To simplify the analysis, and to concentrate on the effects
of advertising, suppose that there is no price competition. Specifically, the value of the
market, in total sales, is given by S. (One can think of a demand curve D(p) and an
exogenously given price, whereby S = p D(p).) S is therefore a measure of market size.
Each firm must decide whether or not to enter the industry. Entry cost is given by F .
If a firm decides to enter, then it must also choose how much to invest in advertising; let
ai be the amount chosen by firm i. Finally, firm i’s market share, si , is assumed to be
equal to its share of the industry total advertising effort:
ai ai
si = Pn =
j=1 aj A

where n is the number of firms in the industry and A ≡ ni=1 is total industry advertising.
P

(a) Show that each firm i’s optimal level of advertising solves S (A −
ai )/A2 − 1 = 0
Answer: The profit of each firm is given by

πi = si S − ai − F

Therefore, each firm is solves

ai
max S − ai − F
ai A
The first order condition is given by
A − ai
S−1=0
A2
P
Notice that A = ai . Therefore, when deriving the first-order condition we must take
into account the effect of changes in ai on the total A.
(b) Show that, in a symmetric equilibrium, a = S (n − 1)/n2 , where
a is each firm’s level of advertising.
Answer: In a symmetric equilibrium we have ai = a = A/n, or alternatively A = n a. The
above result then implies
na − a
S−1=0
(n a)2
Solving with respect to a we obtain the equation in the text.
(c) Show that equilibrium profit is given by π = S/n2 − F
Answer: As shown above, firm profit is given by

πi = si S − ai − F

Substituting the equilibrium values, we get

1 n−1 S
π= S− 2
S−F = 2 −F
n n n

97
 (d) Show that the equilibrium
p number of entrants is given by the
highest integer lower than S/F .
Answer: The equilibrium requires π(n) ≥ 0 and π(n + 1) ≤ 0, where π(n) denotes
equilibrium profits when there are n entrants. This implies that the equilibrium value of n
is the largest integer such that π(n) ≥ 0. Solving

S
−F =0
n2
with respect to n we get r
S
n=
F
Therefore, "r #
S
n
b=
F

(e) Interpret this result in light of the previous discussion on the

effects of endogenous entry costs.
Answer: In equilibrium, advertising expenditures increase with market size. This is an
instance of an endogenous entry cost. This implies that the number of firms increases less
than proportionately with respect to market size — even though price is (by assumption)
constant.

10.18. Entry and welfare. Consider a homogeneous product industry with inverse
demand function P (Q) where every firm has the same cost function: C(q). Suppose that
firms decide sequentially whether or not to enter the industry and that the number of
firms can be approximated by a continuous variable n. Show that, if (a) an increase in n
leads to a decrease in qn (the equilibrium output per firm when there are n firms); and (b)
equilibrium price is greater than marginal cost; then the equilibrium number of firms is
too high from a social welfare point of view. Hint: derive the condition such that a firm is
indifferent between entering and not entering; then show that, at that value of n, the
derivative of social welfare with respect to n is negative.
Answer: The question we are trying to address is the relation between the optimal
number of firms, n∗ , and the equilibrium number of firms, n b. The optimal number of firms
maximizes total welfare, which is given by
Z n qn
W (n) ≡ P (x) dx − n C(qn ) − n F (10.18)
0
where, as before, n is the number of firms and qn each firm’s output (given that there are
n active firms).
The effect of additional entry on welfare is given by
 ∂q  ∂ qn
n
W 0 (n) = P (n qn ) n + qn − C(qn ) − n C 0 (qn )
∂n ∂n

98
The equilibrium number of firms, n
b, is given by the zero-profit condition


P nb qnb qnb − C(qnb ) = 0

Substituting this in (10.18), we get

  ∂q
n
W 0 (b b qnb − C 0 (qnb )


n) = n
b P n
∂n
We conclude that, if margins are positive (P > C 0 ) and there is a business stealing effect
(∂ qn / ∂ n < 0), then, at the equilibrium level of entry, further entry would reduce social
welfare; and, conversely, less entry would increase welfare. In other words, there is
excessive entry in equilibrium.
Notice that, by using differential calculus, this result abstracts from the fact that n
must be an integer value. When this problem is taken into consideration, examples may
be found where there is insufficient entry in equilibrium, For example, suppose that
marginal cost is constant and that duopoly price is equal to marginal cost (Bertrand
competition). In this case, even if the entry cost is very small, the equilibrium number of
firms is just one. However, if the entry cost is indeed very small, then society would be
better off with a second competitor.
11 Horizontal Mergers

11.1. Mergers and output level. “The combined output of two merging firms decreases
as a result of the merger.” True or false?
Answer: If the merger implies little or no cost efficiencies (namely at the level of marginal
cost), we would expect the combined output of the merging firms to decline. If however
the merger reduces the marginal cost of the combined firm significantly, then it is possible
that the combined output increases as a result of the merger.

11.2. Mergers in the paper industry. One of the efficiencies created by mergers in the
paper industry results from reorganization of production. A machine is more efficient the
narrower the range of products it produces, among other reasons because the length of
each production run can be made longer.
The paper industry underwent a wave of mergers in the 1980s. Of the firms that
merged, about two thirds increased their market share as a result of the merger. Assuming
that (i) firms compete by setting production capacity and (ii) paper products are relatively
homogeneous across firms, explain how the previous paragraph explains the pattern of
changes in market shares. Which firms would you expect to increase their market share?28
Answer: According to the paragraph, there are increased cost efficiencies from mergers.
Applying the analysis from Section 11.1, it seems that for two thirds of the merging firms
the cost efficiencies were so big that the merging firms increased their output and market
share, while for the rest the efficiencies were not big enough, resulting in a decreased
market share.

11.3. BAe and GE. “The renewed prospect of a link-up between British Aerospace PLC
and the Marconi defense arm of General Electric Co. PLC of the U.K. has led to revived
talks between the top defense companies in Germany and France.”29 Discuss.
Answer: Refer to the discussion on merger waves in this chapter.

11.4. The HP-Compaq merger. In 2001, HP acquired Compaq. The merger had an
impact on two different markets: desktop PCs and servers. Pre-merger market shares in
the desktop PC market were as follows: Dell, 13; Compaq, 12; HP, 8; IBM, 6; Gateway, 4.
Pre-merger market shares in the servers market were as follows: IBM, 26; Compaq, 16;

100
HP, 14; Dell, 7.30
(a) Determine the value of HHI in each market before the merger.
Answer: Since we don’t have a complete set of market shares, the best we can do is to
determine an approximate value of the H indexes (in particular, a lower bound). We then
have

HD = 132 + 122 + 82 + 42 + 62 = 429

HS = 72 + 162 + 142 + 262 = 1177

(b) Assuming market shares of each firm remain constant, determine

the value of HHI after the merger.
Answer: We now have

HD = 132 + 202 + 42 + 62 = 621

HS = 72 + 302 + 262 = 1625

(c) Considering the values determined above and the DoJ merger
guidelines, was the Department of Justice right in allowing the merger
to take place?
Answer: In the US, the Department of Justice has guidelines based on the HHI. As a rule
of thumb, mergers that lead to an HHI of 1000 or less ordinarily lead to no further
scrutiny; those that raise the HHI by 100 or more and produce a post-merger HHI
between 1000 and 1800 raise significant questions; and those that raise the HHI by 50 or
more and produce a post-merger HHI above 1800 raise significant concern.
According to these guidelines, the merger should be OK’s as far as the desktop
market is concerned, and questions should be raised regarding the server market.
Questions were raised and eventually the merging parties agreed to divest some of their
assets in the server market as a precondition for the merger.

11.5. Merger and the Herfindahl index. Consider an industry with demand Q = a − p
where 3 identical firms that compete a la Cournot. Each firm’s cost function is given by
C = F + c q. Suppose two of the firms merge and that the merged firm’s cost function is
given by C = F 0 + c0 q, where F < F 0 < 2 F .
(a) Determine each firm’s market share before and after the merger
Answer: Before the merger, each firm’s market share is 1/3 (since they are symmetric).
After the merger, we have a duopoly with marginal costs c and c0 . The equilibrium output
level are given by

a + c0 − 2 c
q=
3
a + c − 2 c0
q0 =
3

101
It follows that market shares are given by

a + c0 − 2 c
s=
2 a − c − c0
a + c − 2 c0
s0 =
2 a − c − c0

(b) Suppose that a = 10 and c = 3. Determine the Herfindahl

index after the merger takes place when (i) c0 = 2 and (ii) c0 = 1.
Compare this to the post-merger Herfindahl index calculated based
on pre-merger market shares. Why do these values differ?
Answer: If c0 = 2, then
10 + 3 − 2 × 2 9
s0 = = = .6
2 × 10 − 3 − 2 15
The post-merger Herfindahl index is then
 2  2
0 9 6 117
H = + = = .52 (or 5,200)
15 15 225

If c0 = 1, then
10 + 3 − 2 × 1 11
s0 = = = .6875
2 × 10 − 3 − 1 16
The post-merger Herfindahl index is then
 2  2
0 11 5 146
H = + = = .5703 (or 5,703)
16 16 256

Based on the pre-merger market shares, the post-merger Herfindahl index is given by
 2 
1 1 2 5

0 1
H = + + = ≈ .56 (or 5,600)
3 3 3 9

We conclude that, if marginal cost efficiencies are not very significant (e.g., c0 = 2), then
the post-merger estimate of H based on pre-merger market shares is likely to
over-estimate the post-merger value of H. The reason is that by simply using pre-merger
market shares we are not accounting for the equilibrium adjustment following the merger,
namely the fact that the merging parties will produce a total output that is lower than the
combined output of the two pre-merging firms. By contrast, if marginal cost efficiencies
are very significant (e.g., c0 = 1), then the post-merger estimate of H based on pre-merger
market shares is likely to under-estimate the post-merger value of H. The reason is that
by simply using pre-merger market shares we are not accounting for the equilibrium
adjustment following the merger, namely the fact that the merging parties, by becoming
more efficient, will produce a total output that is higher than the combined output of the
two pre-merging firms.
In other words, there are two effects to consider regarding the merging firms’
post-merger output level: first, to the extent that there are fewer competitors, there is a
tendency to reduce output level; second, to the extent that marginal cost is lower, there is
a tendency to increase output level.

102
 11.6. Merger wave. Consider an industry where firms compete by setting output levels
(Cournot). Market demand is given by D = 150 − P , marginal cost is constant and equal
to 50, and fixed cost is 150 (the same for all firms).
(a) Show that profits per firm are given by 961, 475 and 250 as the
number of firms is equal to 2, 3 or 4.
Answer: In Section 11.1 we saw that equilibrium profit under Cournot competition is
given by
a − n ci + j6=i cj 2
 P 
π
bi = − Fi
n+1
Applying this to the present case yields the values 961, 475 and 250 as the number of
firms is equal to 2, 3 or 4.
Suppose that a merger leads to a new firm with the same fixed cost and the same
marginal cost.
(b) Suppose that initially there are four firms. Show that a merger
between Firms 1 and 2 is unprofitable.
Answer: The merged firm would earn a profit of 475. This is less than 250+250, the
combined profit of the two merging firms.
(c) Suppose that Firms 3 and 4 decide to merge, forming Firm 3&4.
Show that now a merger between Firms 1 and 2 is profitable.
Answer: If 1 and 2 merge after the 3&4 merger, then their profit is 961. This is more than
475+475, the combined profit as independent entities.

11.7. Entry by acquisition. A large fraction of industry entry corresponds to acquisition

of incumbent firms. For example, from a sample of 3,788 entry events, about 70% were
acquisitions.31 Econometric analysis suggests that entry by acquisition is more common in
more concentrated industries.32 Can you explain this observation?
Suggestion: Consider a Cournot oligopoly with n symmetric firms. Determine the
maximum that an entrant would be willing to pay for one of the incumbent firms.
Determine also the minimum that an incumbent would require from a buyer, knowing that
the alternative to selling the firm is for the entrant to create a new firm. Show that the
difference between the two values above is greater when the industry is more
concentrated.33
What other factors would you expect to influence the “build or buy” decision when
entering an industry?
Answer: Suppose that the inverse demand and cost functions are given by p = a − b Q and
C(q) = F + c q, respectively. By an appropriate change of units, I can make b = 1. I will
thus work with an inverse demand p = a − Q. Each firm maximizes

π = (a − c − Q) q − F

which leads to
a−c
q=
n+1

103
Equilibrium profits if there are n competitors are given by
 2
a−c
π(n) = −F
n+1

Suppose that a potential entrant, if it enters by creating a new firm, does so with the
same cost function as the incumbents. For a potential entrant, there are two choices: (a)
to buy an incumbent for a price x, which leads to net profits of π(n) − x; and (b) to create
a new firm, which leads to profits π(n + 1). The potential entrants prefers acquisition if
and only if π(n) − x > π(n + 1). In other words, the potential entrant is willing to pay at
most π(n) − π(n + 1) for an incumbent firm.
For an incumbent firm, either you sell out to a potential entrant, in which case you
get x, or you turn down the offer, in which case entry takes place and profits are π(n + 1).
It follows that an incumbent should ask for at least a payment of x = π(n + 1).
The condition that what the entrant is willing to pay is more than what the
incumbent requires to sell is therefore given by
 
∆ = π(n) − π(n + 1) − π(n + 1)
= π(n) − 2 π(n + 1)
(a − c)2 2 − n2 [1.0]


=F+
(n + 2)2 (n + 1)2

If n = 1 (the lowest possible value), ∆ is equal to F plus a positive value. If n ≥ 2, then

we have ∆ plus a negative value. This suggests that, if n = 1, then the acquisition path
looks more attractive. This seems broadly consistent with the empirical evidence that
acquisition is more likely when the number of incumbents is smaller (more concentrated
industries).
One other reason why acquisition may be a better strategy is time to build:
acquisition provides the entrant with quicker entry than the alternative of creating a new
firm.
12 Market Foreclosure

12.1. Store clustering. “Tourists traipsing along a half-mile stretch of 23rd Street in
New York pass five Starbucks outlets. In Tokyo, 7-Eleven boasts 15 stores within a similar
distance of Shinjuku station.”34 Does it make any sense for stores to be clustered in this
way?
Answer: Refer to the discussion in Section 12.1. If there were not potential competition,
it would make little sense to create such a high density of stores. However, if there are
potential entrants, then increasing the density of outlets in Manhattan or Tokyo may have
the benefit of discouraging entry.

12.2. LC and CS burgers. LC Burgers is currently the sole fast-food chain in Linear
city, a city that is one mile long and consists of one street, with one thousand consumers
distributed uniformly along the street. The price for the BigLC, the only product sold by
the LC Burger chain, is set nationally at $4, so that the local Linear city manager’s
decision is limited to choosing the number and location of its stores.
Each store costs $600,000 to open and lasts indefinitely. Each consumer buys one
burger per week at the current price of $4. However, no consumer will walk for more than
a quarter of a mile to buy a burger. Operating costs are $1 per burger. The interest rates
is 0.1% per week. The market conditions are unchanging, so present discounted profits can
be regarded as level perpetuities.
(a) Suppose that LC Burgers faces no competition and no threat
of entry. How many stores should LC Burgers open, and at what
locations?
Answer: If LC Burgers opens a single store, it can capture at most 500 consumers (the
ones living a quarter mile to the left and a quarter mile to the right of the store). In such
a case, the profits would be

(4 − 1) × 500
− 600, 000 = 900, 000
0.001
Since only half of the market is covered, LC Burgers can open another store, again hoping
to capture at most 500 customers. To cover the entire market, the optimal location is at 14
and 43 , case in which there is no overlap between the ”markets” of each store. By opening
two stores, LC’s total profit is given by 1,800,000.

105
CS Burgers is contemplating entering Linear city. CS Burgers’ costs and price are the
same as those of LC Burgers. Moreover, consumers regard the products at both chains as
equally good, so, if both brands are in town, each consumers buys from the closest store.
(b) At what locations should CS Burgers open stores, given that LC
Burgers has opened the locations found to be optimal in part (a)?
Answer: Note that a store is profitable if covers 250 customers. In fact, profit is then
given by
3 × 250
− 600, 000 = 150, 000
0.001
Therefore, CS Burgers can open stores in 4 locations, namely at a very small distance to
the left and right of each LC Burgers store. In this case CS Burgers will have the whole
market and will make profits of $600.000 = 4 × $150, 000.
(c) Recognizing the threat of entry by CS Burgers, at what locations
should LC Burgers open stores?
Answer: A store becomes profitable if it covers more than 200 customers. To see this,
solve the equation
3×x
− 600, 000 = 0
0.001
with respect to x. Given this, if the distance between to stores is less than 0.4 miles, no
firm locating in between is viable (locating in the middle or right next to a store brings at
most 200 customers). Therefore, in order to avoid entry LC Burgers can open 3 stores,
one located at 0.2, the other at 0.6 and the last at 1. In such a case CS Burgers cannot
find a profitable position to open a store.

12.3. German telecommunications. In less than one year after the deregulation of the
German telecommunications market at the start of 1998, domestic long-distance rates fell
by more than 70%. Deutsche Telekom, the former monopolist, accompanied some of these
rate drops by increases in monthly fees and local calls. MobilCom, one of the main
competitors, feared it might be unable to match the price reductions. Following the
announcement of a price reduction by Deutche Telekom at the end of 1998, shares of
MobilCom fell by 7%. Two other competitors, O.tel.o and Mannesmann Arcor, said they
would match the price cuts. VIAG Interkom, however, accused Telekom of
“competition-distorting behavior,” claiming the company is exploiting its (still remaining)
monopoly power in the local market to subsidize its long-distance business.35
Is this a case of predatory pricing? Present arguments in favor and against such
assertion.
Answer: One could indeed argue that this is a case of predatory pricing. If Deutsche
Telekon has monopoly in local markets it has strong financial resources and can afford to
lose money in the long distance market by pricing below marginal cost. However, since
there are 2 other competitors that matched Deutsche Telekom’s prices, one can argue that
there exists technology with marginal cost less than the low price charged. Evidently, the
other explanations can also be used, namely, low-cost signaling and reputation for
toughness.

106
 12.4. Exclusive contracts as a barrier to entry. Suppose that there is one buyer who is
willing to pay up to one unit of a given good for the price of 1. There is an incumbent
seller with production cost 12 (a value that is publicly known). Suppose there is a potential
competitor to the seller. Unlike the incumbent seller, the potential entrant’s cost is
unknown to buyer and incumbent. Both expect it to be a value uniformly distributed in
the [0, 1] interval. The potential entrant, by contrast, is able to observe its cost and make
an entry decision conditional on the observed value. If entry takes place, then the sellers
play a price-setting game (with homogeneous product), whereby the low-cost firm sets a
price just below the high-cost firm’s cost.
(a) Show that the incumbent and the consumer’s ex-ante expected
payoffs are both given by 14 .
Answer: If the entrant’s cost is greater than 21 , then no entry takes place and the
incumbent sets the monopoly price 1. If the entrant’s cost is less than 12 , then the entrant
enters and sets price equal to the incumbent’s cost, 21 . The buyer’s expected payoff is 12
times zero (no entry) plus 12 times (1 − 21 ) (entry takes place), that is, 14 . The incumbent,
in turn, receives an expected payoff of 12 (no entry) times (1 − 12 ), or 41 .

Now suppose that incumbent and consumer agree to an exclusive contract before the
potential entrant makes its entry decision. The contract stipulates that the consumer buys
the product from the incumbent (and only from the incumbent) at a price of 34 . Moreover,
if the consumer were to buy from the entrant then it would pay the incumbent a penalty
of 21 for breach of contract.
(b) Show that both consumer and incumbent are better off with this
exclusive contract.
Answer: Notice that, despite the contract, entry still takes place when the entrant is very
efficient. In fact, if the entrant’s cost is less than 14 , then the entrant enters and sets price
equal to 14 — or just a little less than that. Given this price, the buyer switches seller: it
now pays 14 (price) plus 21 (penalty from breach of contract), a total of 34 (the price
required by the incumbent). Following this reasoning, we conclude that entry occurs with
probability 14 , the probability that the entrant’s cost is lower than 14 . The buyer’s expected
utility is 14 times (1 − 14 − 12 ) (entry takes place) plus 34 times (1 − 34 ) (entry does not take
place), or simply 41 (the same value as before). The incumbent, in turn, now gets 14 times
1 3 3 1
2 (entry takes place and the entrant receives the damage payment) plus 4 times 4 − 2
5 1
(entry does not take place), or 16 , which is strictly greater than 4 , the previous payoff.
Not surprisingly, the loser in this process is the entrant. Without contracts, the
entrant sells at a price 12 , whereas with the above contract the maximum the entrant can
charge is 14 . Notice that, if the entrant’s cost lies in the interval [ 41 , 12 ], then entry does not
take place even though the entrant is more efficient than the incumbent.
Note that the buyer is indifferent between no contract and the above exclusive
contract. However, to the extent that the incumbent seller is strictly better off with an
exclusive contract, there exist contracts (with a slightly lower price) such that both
incumbent and buyer are strictly better off.
13 Vertical Relations

13.1. McDonald’s. Empirical evidence suggests that McDonald’s restaurants that are
wholly owned by the parent company charge lower prices than do independent franchise
ones. How can this difference be explained?
Answer: This is an example of double marginalization. If firms are vertically integrated
(as is the case with franchise-owned McDonalds restaurants), then the retailer price is the
monopoly price for the vertical structure. If the firms are not vertically integrated, then
retailer’s profit maximization leads to a second monopoly margin which takes as marginal
cost the wholesale price. If the wholesale price is equal to the marginal cost of the
upstream firm, then the two retail prices are the same. However, in such a case the
upstream firm makes zero profits. We would thus expect the wholesale price to be greater
than marginal cost. It follows that the retail price for independent retailers is higher than
for franchise-owned retailers.

13.2. Resale price maintenance. The following industries are known to practice or have
practiced resale price maintenance: fashion clothing, consumer electronics, fine fragrances.
In each case, indicate the probable motivation for RPM and the likely welfare
consequences.
Answer: In the case of consumer electronics, there is an important positive externality
from investing in sales effort: Retailers can free-ride on the investment efforts made by
competing retailers, since one consumer can benefit from the point-of-sale services
provided by a retailer (who invested in sales effort) and shop at a lower-price retailer (who
did not invest). The result of this externality is that no retailer invests and the demand
for the good is lower. An RPM policy induces investment in sales effort (which increases
demand); instead of competing in price (which is now the minimum price required by the
manufacturer), the retailers compete in investments in sales effort to attract customers.
The final beneficiaries of this policy are, obviously, the retailers and the manufacturer.
Nevertheless, consumers also benefit from better services at the point of sale.
In the fashion clothing and fine fragrance industries the degree of externality is likely
to be much lower. Still, the incentives to invest may not be high enough, since the
retailer’s benefit from investing in sales effort depends on the margin it receives.
Specifically, if the margin is low, then the retailer will invest a small amount. Using an
RPM policy, the manufacturer can create a larger margin for the retailer, thus inducing

108
the optimal level of investment.

13.3. Vermont Castings. Vermont Castings is a manufacturer of wood-burning stoves,

a somewhat complex product. One of Vermont Castings’s dealers once complained about
the terms of the relations between the manufacturer and dealers, stating that “the worst
disappointment is spending a great deal of time with a customer only to lose him to
Applewood [a competing retailer] because of price.” Specifically, the dealer lamented “the
loss of 3 sales of V.C. stoves . . . to people whom we educated and spent long hours
with.”36
How do you think this problem can be resolved? How would you defend your solution
in an antitrust/competition policy court?
Answer: Obviously, this is a case when one retailer makes an investment in sales effort
while the other free-rides and gets the customer by charging a lower price. As we saw in
Section 13.2, one possible solution to this problem is to use an RPM policy. In this way,
the price would be “fixed” at the minimum level, while the retailers would compete in
sales efforts to attract customers.
One would expect that the minimum price set in the RPM policy would be high,
hence, an antitrust court would not agree with this policy. However, Vermont Casting
may argue that, absent the RPM policy, the retailers will have no incentives to further
invest in sales effort and to provide services to customers, making them (the customers)
worse off. The price may be lower but the services may be poor. On the contrary, by
using an RPM policy, the price may indeed increase, but the customer will now benefit
from proper services provided at the point of sale. Obviously there is a trade-off between
using and not using an RPM policy, with the crucial issue being the level at which the
minimum price should be set.

13.4. Exclusive territories in the European Union. Should the European Union outlaw
the practice of exclusive territories in car dealerships? Why or why not?
Answer: Exclusive territories represent an instance of vertical restraints that helps in
resolving the inter-retailer externality represented by underinvestment in sales effort.
Thus, if awarded an exclusive territory, a car dealer has all the incentives to invest in
advertising, educating customers, etc., while absent this policy it would, most likely,
under-provide these services. The issue is to quantify the positive and negative effects of
such a policy, that is, to observe how prices and service levels are set in areas where this
policy is in use compared to areas where it is not.

13.5. Exclusive dealing in beer sales. Beer producers are wont to impose an exclusive
dealing clause on retailers. Discuss the efficiency and market power effects of this practice.
Answer: Exclusive dealing has the effect of foreclosing upstream competition, that is,
competition between manufacturers. This is likely to increase market power and reduce
welfare. One possible defense of exclusive dealing is that there may be important
investments to be made by the manufacturer at the retail store, so that, if there is
competition between manufacturers, an externality may arise and lead to sub-optimal
investments. In the case of car dealerships, such externality might arise in the context of
dealer training to be done by the manufacturer. In the case of beer, however, it is unlikely

109
there are significant manufacturer externalities.

13.6. Cable TV. Evidence from the US cable TV industry suggests that the likelihood
that a given network is carried by a distributor is greater when the network is owned by
the distributor (see Box 13.1). Consider the analysis of double marginalization in Figure
13.1. Suppose that there the distributor must pay a cost F in order to carry a given
network (this could correspond to the opportunity cost of adding a channel to the
distributor’s lineup, for example the profit lost from not being able to carry a different
channel). Show that there exist values F such that the distributor will carry the network
if and only if it owns the network.
Answer: Refer to Figure 13.1. Suppose that the value of F is greater than area A + B but
lower than A + C. (Such values exist since C > B.) If Firms M and R are merged into
M &R, then their total profit is given by A + C. This value is greater than F , thus it
compensates for carry the content. However, if the firms are vertically separated, then
their combined profit is A + B. Even if each firm were to contribute to cover the fixed
cost, they would not willing to pay enough to get to F .

13.7. Wintel. Consider the following highly simplified picture of the personal computer
industry. A large number of price-taking firms assemble computer systems; call them
“computer OEMs” (Original Equipment Manufacturers). Each of these firms must buy
three inputs for each computer system that it sells: (1) a variety of components that are
themselves supplied competitively and collectively cost the computer OEM $500 per
computer; (2) the Windows operating system, available only from Microsoft, at a price
pM , to be discussed below; and (3) a Pentium microprocessor, available only from Intel, at
a price pI , also to be discussed below. Since each computer system requires precisely one
operating system and one microprocessor, the marginal cost of a computer to an OEM is
500 + pM + pI . Assume that competition among OEMs drives the price of a computer
system down to marginal cost, so we have p = 500 + pM + pI , where p is the price of a
computer system. The demand for computer systems is given by
Q = 100, 000, 000 − 50, 000 p. Microsoft is the sole supplier of the Windows operating
system for personal computers. The marginal cost to Microsoft of providing Windows for
one more computer is zero. Intel is the sole supplier of the Pentium microprocessors for
personal computers. The marginal cost to Intel of a Pentium microprocessor for one more
computer system is $300.
(a) Suppose that Microsoft and Intel simultaneously and indepen-
dently set the prices for Windows and Pentium chips, pM and pI .
What are the Nash equilibrium prices, pbM and pbI ?
Answer: First consider Microsoft’s best response to any given price pI by Intel. Using the
underlying demand for computers, the demand for Windows is given by
Q = 100, 000, 000 − 50, 000(500 + pM + pI ). For a given value of pI , the demand for
Windows is Q = 75, 000, 000 − 50, 000pI − 50, 000pM . The corresponding marginal revenue
for Microsoft is MR M = 1500 − pI − Q/25, 000. Setting this equal to Microsoft’s marginal
∗ = 37, 500, 000 − 25, 000p , and the corresponding optimal price of
cost of zero gives qM I
∗
pM = 750 − pI /2. Next, repeat these steps to consider Intel’s best response to any given
price pM by Microsoft. The only difference is that Intel has a marginal cost of $300.

110
These calculations imply that MR I = 1500 − pM − Q/25, 000. Setting this equal to Intel’s
marginal cost of $300 gives with the corresponding optimal price of p∗I = 900 − pM /2.
Finally, solve these two equations together to get the Nash Equilibrium prices, which are
p∗M = $400 and p∗I = $700. Note that the resulting price of a computer is $1600, so total
computer sales are 20 million.
Now suppose that Microsoft and Intel sit down to negotiate an agreement to sell Windows
and Pentium chips as a package to computer OEMs for a package price of pMI .
(b) What package price would maximize Microsoft’s and Intel’s com-
bined profits? By how much would an agreement between Microsoft
and Intel boost their combined profits?
Answer: This is a basic monopoly pricing problem for Microsoft and Intel collectively. If
they set a package price of pMI , the price of a computer system will be 500 + pMI . The
number of computers sold will be Q = 100, 000, 000 − 50, 000(500 + pM I ). The marginal
revenue corresponding to this demand curve is MR MI = 1500 − Q/25, 000. Setting this
equal to the (combined) marginal cost of $300 gives a quantity of Q∗MI = 30, 000, 000 and
a corresponding package price of p∗MI = $900. At this price, the contribution to
Microsoft’s and Intel’s combined profits is $600 per computer times 30 million machines,
or $18 billion. In comparison, the Nash Equilibrium in part (a) involved a contribution of
$800 per computer times 20 million machines, or $16 billion. Cutting a deal is worth $2
billion to Microsoft and Intel together.
(c) Would final consumers benefit from such an agreement between
Microsoft and Intel, or would they be harmed? What about computer
OEMs? Relate your answer to your calculations in parts (a) and (b),
and explain the economic principles underlying your answer.
Answer: Since Windows and Pentium are complements, Microsoft’s profits are decreasing
in the price of Pentium chips, and Intel’s profit is decreasing in the price set by Microsoft.
This implies that the two companies together would benefit from lower prices than they
would set separately. Indeed, comparing parts (a) and (b) we see a lower price in part (b)
than in part (a). Final consumers thus benefit from the cooperation between Microsoft
and Intel. OEMs are indifferent, because their profits are driven to zero by competition,
whatever the prices of components. (In practice, OEMs would benefit in the short run
from the lower input prices, and OEMs able to differentiate themselves with their own
brand names would benefit for a longer period of time.) The underlying principle is that
cooperation among suppliers of complements tends to benefit consumers, just as
cooperation among suppliers of substitutes (i.e., collusion) harms consumers. This is
closely related to the theory of “double marginalization” that we discussed in this chapter;
the only difference is that Microsoft and Intel stand in a “complements” relationship
rather than a buyer/seller relationship.

13.8. Two-part tariffs. Suppose that a manufacturer sells to n retailers by means of a

two-part tariff (f, w): a fix fee f and a wholesale price w. Explain the intuition of the
result that the greater the degree of retailer competition, the greater the optimal
wholesale price.
Answer: Suppose there are two downstream firms, R1 and R2 ; and that M ’s marginal

111
cost is c, whereas each Ri has no other variable cost than the wholesale price w (and no
capacity constraints, that is, retailers can buy as much as they want from the
manufacturer). Without retailer competition setting w = c leads the retailer to set
p = pM . With retailer competition, w = c would lead to a retail price below monopoly
level. This is because, at p = pM , each retailer would have an incentive to lower price and
steal market share from the rival retailer.
Specifically, suppose that retailers compete à la Bertrand. Then in equilibrium retail
price will be equal to the wholesale price (recall that the wholesale price is the retailer’s
marginal cost). If w = c, then p = c and both manufacturer and retailers make zero
profits. Under Bertrand competition, the optimal solution would be to set w = pM
instead. For this wholesale price, retailers set, in equilibrium, p = pM , and the
manufacturer gets monopoly profits (in this case, the franchise fee would be zero).
It can be shown that, when retailers compete à la Cournot (an intermediate level of
competition), then the manufacturer’s optimal solution is to set a wholesale price (strictly)
between w and pM (see Exercise 13.9). Together, these results suggest that the greater the
degree of competition between retailers, the higher the optimal wholesale price.

13.9. Downstream Cournot competition. Consider an industry with one upstream firm
and n downstream firms that compete à la Cournot. Show that the optimal wholesale
price is strictly between marginal cost and monopoly price.
Answer: As in the text, suppose that the upstream firm offers retailers a contract
stipulating a fixed fee, f , as well as a wholesale price, w. From Section 8.2, we know that
the equilibrium price under Cournot competition is given by pN = n1 a + n2 w, where w is
the effective marginal cost paid by retailers. Output per firm is given by
 2
q N = n1 (a − pN ) = n+1
1
(a − w). Finally, equilibrium profit per firm is π R = a−wn+1 . This
 2
implies that the upstream firm can ask for as much as f = a−w n+1 as a fixed fee.
The upstream firm’s total profit is therefore given by
 2 !

a − w a − w
π M = n (w − c) q N + π R = n (w − c) +
n+1 n+1
Maximizing with respect to w, we get the optimal value
n−1 n+1
w= a+ c
2n 2n
Notice that the optimal w is a convex combination of a and c, that is, the coefficients of a
and c add up to 1. Moreover, from Section 3.2 we know that monopoly profit is given by
pM = 21 a + 12 c. Since the relative weight of w on c is greater than the weight of pM on c
(and c < a), it follows that w is less than monopoly price. By the same argument, it is
also clear that w is greater than marginal cost.

13.10. Asset specificity and vertical integration. The value created by the
√ √
collaboration between firms M and R is given by v = eM + eR , where eM and eR are
investments (in $) by firms M and R. After investment levels have been chosen, M and R
negotiate the division of v, each firm keeping 50%.
(a) Determine the equilibrium values of ei (i = M, R).

112
Answer: Firm M maximizes v/2 − eM , that is,
1 √ √

2 eM + eR − eM

The first-order condition is given by

1 √
4 / eM −1=0

or simply
1
eM =
16
Similarly,
1
eR =
16
(b) Suppose that Firm M and R merge. Determine Firm M &R’s
optimal choices of ei .
Answer: Firm M &R maximizes
√ √
eM + eR − eM − eR

The first-order conditions are given by

1 √
2 / ei −1=0

or simply
1
ei =
4
i = M, R.
(c) Compare total payoff under vertical separation and under vertical
integration.
Answer: Total payoff is given by
√ √
eM + eR − eM − eR

Under vertical separation this is equal to

q q
1 1 1 1 3
16 + 16 − 16 − 16 = 8

Under vertical integration this is equal to

q q
1 1 1 1 1
4 + 4 − 4 − 4 = 2

(d) Comment on your answer to the previous question in light of the

discussion in Section 13.1, Box 13.1, and Section 5.1.
Answer: Vertical integration leads to an 33% increase in total payoff. This is due to the
fact that it internalizes the investment externality inherent to the choice of ei : part of the
benefit from increasing ei is captured by firm j, which in turn leads to under-investment.
This problem is very similar to the free-riding problem discussed in Section 5.1: part of
the benefit from agent i’s action is captured by another agent.
14 Product Differentiation

14.1. IBM mainframes. Empirical evidence suggests that, during the 1970s, a firm with
an IBM 1400 was as likely as any other firm to purchase an IBM when making a new
purchase, while a firm with an IBM 360 was more likely to purchase an IBM than a firm
that did not own an IBM 360. Software for the IBM 1400 could not run on the succeeding
generations of IBM models (360, 370, 3000, and 4300), while software for the IBM 360
could run on the 370, 3000 and 4300.37 How do you interpret these results?
Answer: These results suggest how backwards compatibility influences the degree of
switching cost. Switching away from an IBM 1400 was less costly because there was no
backwards compatibility between later models and the software developed for the IBM
1400. The same was not thue for the 360,370, 3000 and 4300 models. Consumers who
bought one of these models had a higher opportunity cost of switching to a non-IBM
computer. As we would expect, these consumers were more likely to buy IBM in the
future than other consumers.

14.2. Price dispersion. “Price dispersion is a manifestation — and indeed it is a

measure — of ignorance in the market.”38 Do you agree? Compare this explanations with
possible alternative explanations for price dispersion.
Answer: If we consider search costs as being a measure of market ignorance, then indeed
the above claim holds. Alternative explanations for price dispersion may be: price
discrimination, different regulatory or taxation regimes (geographical price dispersion) or
different shopping experience (see the example for CDs bought in a small music shop or in
a supermarket).

14.3. Uninformative advertising and market efficiency. Explain how advertising

expenditures with no direct informational content can increase market efficiency.
Answer: As discussed in Section 14.3, advertising expenditures may signal product
quality. In the presence of repeat purchases, a firm that produces a high-quality good and
sells the good not only in the present but also in the future, will have more to gain from
getting customers to try its product than a firm that produces a low-quality good. This is
because once a good is purchased, consumers become aware of its quality; and if it’s high,
they will repeat the purchase in future periods. If, however, a consumer does not get to
try the good in the present, in the future he or she will still be uncertain about the good’s

114
quality. Therefore, high-quality goods producers will try to lure customers in the present
since their gain is higher. They thus have an incentive to differentiate themselves from
low-quality goods producers.
Although advertising has no direct informational content, the equilibrium with
advertising may be more efficient than the equilibrium without advertising. Absent
advertising, high-quality firms have no incentive to produce, since they cannot
differentiate themselves; their products are ex-ante identical to the ones produced by
low-quality firms. Therefore, if consumers value high-quality goods, even if there are
savings in advertising expenditures, the overall efficiency effect may be negative, due to
the loss in the availability of high-quality goods.

14.4. Advertising and brand switching. Empirical evidence suggests that the
probability of a household switching to a different brand of breakfast cereal is increasing
in the advertising intensity of that brand. However, the effect of advertising is
significantly lower for households who have previously tried that brand.39 What does this
suggest about the nature of advertising expenditures (persuasion vs information)?
Answer: The evidence is consistent with advertising having an informative role:
consumers who are already experienced with the good are therefore less sensitive because
they require no additional information. Notice this is consistent with advertising not
having any informative content. As shown in Section 14.3, advertising may be a signal of
product quality; this type of advertising leads consumers to experience a good and thus be
informed about its characteristics. As a result, even though advertising does not convey
any hard information about product characteristics, indirectly it leads consumers to
become informed.

14.5. Advertising intensity. Consider the following industries: pharmaceuticals, cement,

perfumes, fast food, compact cars. How would you expect them to be ordered by
advertising intensity? Why?
Answer: According to the Equation (14.6), advertising intensity is proportional to the
demand elasticity of advertising expenditures and inversely proportional to the price
elasticity of demand. Price elasticity of demand is lowest for pharmaceuticals and
perfumes, highest for cement. Advertising elasticity is lowest for cement, highest for
perfumes (and some pharmaceuticals). We would expect advertising intensity to be
highest for pharmaceuticals and perfumes, lowest for cement, intermediate for fast food
and compact cars.

14.6. Product positioning and price competition. Consider a duopoly where horizontal
product differentiation is important. Firms first simultaneously choose their product
locations, then simultaneously set prices in an infinite series of periods. Suppose that
firms collude in prices in the second stage and anticipate they will do so at the
product-positioning stage. In this context, what do you expect the degree of product
differentiation to be?40
Answer: Suppose that consumer preferences are uniformly distributed along a interval in
product space (as in the Hotelling model). If firms do not compete on price, then demand
is entirely determined by product position. Given firm j’s product position, firm i’s best

115
response is to locate next to firm j, to the right of j if j’s position is to the left of the mid
point, to the left of j if j’s position is to the right of the mid point. As a result, the Nash
equilibrium of the product-positioning game has firms locating at the center of the
product space; that is, minimum differentiation takes place in equilibrium.

14.7. Price competition with search costs. Twenty-five different stores sell the same
product in a given area to a population of two thousand consumers. Consumers are
equally likely to first visit any of the twenty-five stores. Half of the consumers have no
search costs and purchase at the lowest price so long as it is lower than $45. The other
half is willing to buy one unit of the product up to a maximum of $70 and must incur a
cost of $44 in order to find out about the prices charged by other stores. Each store can
sell up to 90 units and has a unit cost of $25.
(a) Show that, in equilibrium, there exist at most two different
prices.
Answer: Suppose there are three different equilibrium prices. Half the consumers
purchase at the lowest price, so the the two higher prices are only paid by the
high-search-cost consumers. Since these consumers are uniformly distributed across stores,
one of the higher prices must be a sub-optimal strategy: either buyers turn away from the
higher price, in which case the seller is better off by lowering price; or buyers buy at the
lowest price, in which case the seller is better off by increasing price. See also Salop,
Steven, and Joseph Stiglitz (1976), “Bargains and Ripoffs,” Review of Economic Studies
44, 493–510; or the summary discussion in Chapter 8 of Varian, Hal (1978),
Microeconomic Analysis, New York: Norton.
(b) Show that, if there exist two different equilibrium prices, then
the higher price must be 70.
Answer: If the high price is lower than 70, a firm that deviates by slightly increasing price
does not lose market share since consumers are not willing to pay the search cost.
Therefore, the firm is strictly better off. Hence, all firms would want to deviate upwards,
so that the high price must be 70.
(c) Show that the following is an equilibrium: 5 firms set a price of
70 and the remaining 20 firms set a price of 45.
Answer: First notice that, given the search costs for first type of consumers, we can safely
assume that these consumers will not search, rather will compare price to their willingness
to pay (70).
At the proposed prices, profits are as follows: for a firm setting p = 70, demand is
given by 1,000/25=40 and total profit is 40 × (70 − 25) = 1, 800. For a firm setting p = 45,
total demand is 1, 000/25 + 1, 000/20 = 90 and total profit is 90 × (45 − 25) = 1, 800.
A p = 45 firm could deviate by setting a lower price. It would get more demand but,
since it is selling at capacity, profit would be lower. It could set a higher price but would
then only keep the high valuation consumers. It could at most make a profit equal to the
profit currently earned by the p = 70 firms, which in turn is equal to its current profit. We
thus conclude that such firm would not want to charge a different price.
A p = 70 firm could deviate by setting a lower price. Any price below 70 and above
45 leads to the same demand but a lower margin. By setting a price equal or lower than

116
45, the firm would get less than what p = 45 firms currently get, which in turn is the same
as a p = 70 firm currently gets.

14.8. Demand for Yoplait yogurt. Statistical analysis suggests that the demand for
Yoplait 150 yogurt is given by a constant term + 1.85 × Advertising exposure − 0.24 ×
Advertising exposure × Number of previous purchases, where “advertising exposure” is
the number of 30-second ads for Yoplait 150 observed by each consumer during the week
of the shopping trip.41 How do these results address the debate over persuasion vs
information effects of advertising?
Answer: The first coefficient, positive, indicates that the more Yoplait spends on
advertising, the greater the probability of a purchase. The second coefficient, negative,
indicates that the more accustomed a consumer becomes to Yoplait, the less advertising
expenditures will influence his of her decision. This result is consistent with the view that
advertising has an informative role (in this case, information about the product’s
existence).
15 Innovation

15.1. Perfect competition and innovation. “Perfect competition is not only impossible
but inferior, and has no title to being set up as a model of ideal efficiency.” Do you agree?
Why or why not?
Answer: In a static sense, perfect competition is the most efficient industry structure,
since it maximizes social welfare. In a dynamic sense, however, perfect competition is not
necessarily the ideal model, since it may be less conducive to technological progress than
situations that allow for some (temporary) degree of market power. The latter would not
necessarily be monopoly markets (in a static sense) but rather forms of oligopoly, in which
firms compete not only in quantity (or price) but also in R&D, so that they can outpower
rivals in the future and gain some market power.

15.2. Innovation and market structure. “R&D competition implies a dynamic system
whereby industries tend to become more and more concentrated.” Do you agree? Why or
why not?
Answer: If preemption motives are very strong, then the tendency is for technology
leaders to cement their leadership. However, if the replacement effect is strong; or if
organizational inertia is very important; then laggards are more likely to innovate than
leaders.

15.3. Value of innovation. Derive equation (15.1). Assuming that the values of p0 , p1
are set by a profit-maximizing monopolist, show that the value of innovation is greater the
greater a1 or the lower c1 .
Answer: Total profit is given by total output, q, times the difference between price and
marginal cost, p − c. Consumer surplus, in turn, is given by the area of the triangle
between price and the inverse demand curve:

CS = 12 (a − p) Q

Adding consumer surplus and profits, and noting that Q = a − p, we conclude that social
surplus is given by
S = 12 (a − p)2 + (p − c) (a − p) (15.19)
Substituting the post- and pre-innovation values ai , ci , pi for a, c, p, we get (15.1).

118
 If p is set by a monopolist seller, then (cf Section 3.2)
1
p= 2 (a + c)

Substituting for p in (15.19) and simplifying we get

S= 3
8 (a − c)2

The value of innovation is thus given by

V = 3
8 (a1 − c1 )2 − 83 (a0 − c0 )2

Clearly, V is increasing in a1 and decreasing in c1 .

15.4. Cost reduction with Bertrand competition. Two firms are engaged in Bertrand
competition. There are 10,000 people in the population, each of whom is willing to pay at
most 10 for at most one unit of the good. Currently, both firms have a constant marginal
cost of 5.
(a) What is the equilibrium in this market? What are the firms’
profits?
Answer: Both firms charge p = 5 and earn π = 0.
(b) Suppose that one firm can adopt a new technology that lowers its
marginal cost to 3. What is the equilibrium now? How much would
this firm be willing to pay for this new technology?
Answer: The firm with the lower cost technology charges a fraction of a cent less than
p = 5 and sells to all 10,000 customers. Its profits are π = [(5 − 3) · 10, 000] = 20, 000. It
would be willing to pay up to 20,000 for this technology.
(c) Suppose the new technology mentioned in (b) is available to both
firms. The cost to a firm of purchasing this technology is 10,000. The
game is now played in two stages. First, the firms simultaneously
decide whether to adopt the new technology or not. Then, in the
second stage, firms set prices simultaneously. Assume that each firm
knows whether or not its rival acquired the new technology when
choosing its prices. What is (are) the Nash equilibrium (equilibria) of
this game?
Answer: There are two pure-strategy equilibria: (1) firm 1 invests in the low-cost
technology and firm 2 does not, and (2) firm 2 invests in the low-cost technology and firm
1 does not. It is not an equilibrium for both firms to invest or for neither firm to invest.
(There is also a mixed strategy equilibrium in which each firm invests with probability
0.5.)

15.5. Word-of-mouth diffusion of innovations. Show that the model of diffusion by

word-of-mouth communication implies the adoption path given by (15.3). (Note: this
problem is quite challenging, mathematically speaking.)
Answer: A new user adopts if and only if a user is matched with a non-user. This
happens with probability 2 xt (1 − xt ). This implies that the growth in xt is proportional

119
to the product xt (1 − xt ).
∂ xt
∝ xt (1 − xt )
∂t
where ∝ means “proportional to.” Suppose that xt has the functional form:
1
xt =
(15.20)
1 + exp − (t − β)
where β is a constant. Then
 2
1
∂ xt  
=
 exp(t − β)
∂t  1 + exp − (t − β) 

Since

exp − (t − β)
1 − xt =

1 + exp − (t − β)
I conclude that
∂ xt
= xt (1 − xt )
∂t
which confirms (15.20) satisfies the desired property. Knowing that xt = x0 when t = t0 , I
know that
1
x0 =

1 + exp − (t0 − β)
which implies

1
1 + exp − (t0 − β) =
x0

1 − x0
exp − (t0 − β) =
x0
−(t0 − β) = ln(1 − x0 ) − ln(x0 )
and finally
β = t0 + ln(1 − x0 ) − ln(x0 )

15.6. Strength and length of patent rights.42 Many standardization agreements require
that patent holders cross-license their patents on reasonable and non-discriminatory terms
(sometimes denoted by the acronyms RAND or FRAND). Suppose consumer demand for a
patented innovation is given by D(p). Let c be production cost and pM monopoly price.
Suppose that, if the patent is licensed to a competitor, then firms compete in prices (a la
Bertrand), knowing that the second firm’s cost includes the per-unit license f to pay the
patent holder. Finally, let the patent last for T periods.
(a) Show that, if f = pM − c, then firm profits and consumer welfare
are the same with and without patent licensing.
Answer: The patent owner’s marginal cost is f . The rival’s marginal cost is c + f = pM .
In equilibrium the patent owner sets p = pM and collects all market demand. This is the
same as the monopoly solution (no patent licensing.

120
 (b) Show that lowering the license fee infinitesimally from f = pM −c
increases consumer welfare without decreasing patent value.
Answer: This is a consequence of the fact that profits are a concave function of price and
pM , the monopoly price, is the value that maximizes profits. In the neighborhood of the
optimal solution, the derivative of profit with respect to price is approximately equal to
zero: a slightly lower (or higher) price leads to an approximately equal profit.
(c) Show that, by decreasing the license fee and increasing patent
length, a new patent system can be obtained that provides the same
reward to patent holders and makes consumers strictly better off.
Answer: In terms of post-innovation value, society has much more to gain from a
weakening of the patent system than the monopolist has to lose. This contrasts with
changing the duration of the patent: in terms of post-innovation value, what a monopolist
has to gain from a longer patent is proportional to what society has to lose: double the
patent life and you double both the patent value and total allocative inefficiency. Putting
these facts together suggests that an optimal patent system should make patents very long
but very weak. This would seem to be the best compromise between making patents
valuable and minimizing the allocative inefficiency that they induce.
(d) What aspects of the innovation reward system may the above
analysis miss out (open question)?
Answer: One of the purposes of patents, in addition to giving incentives for innovation, is
to publicize the relevant information regarding the product or process in question. In fact,
it is a requirement for a patent that it should contain “a written description . . . in such
full, clear, concise and exact terms as to enable any person skilled in the art . . . to make
and use the same” (U.S. Patent Code, Section 112). Among other things, this allows the
information possessed by one party to be available to everyone else, thus accelerating the
aggregate rate of innovation.
The disclosure requirement may, however, create the wrong incentives for i̊nvestors.
Suppose that a laboratory has made a basic innovation, one that can be used as a basis
for several other derived innovations. For example, a technique for locating and purifying
human genes (basic innovation) may be used in the development of a series of medical
drugs (derived innovations). If patent protection is very weak — in the sense described
above —, then the inventor will have little incentive to patent. In fact, by keeping its
invention secret, it will have a better chance at winning the race to develop the derived
innovations that flow from the basic one. But, from a social point of view, such delay in
making information public may be very costly. In fact, it implies that only one firm — as
opposed to several firms — will be engaged in research directed at derived innovations.43
An optimal patent system should trade off this effect against the allocative efficiency effect
considered above.

15.7. Patent thickets. Firm X produces a certain smartphone for which market
demand is given by Q = a − p. Production cost consists of licensing n patents required to
produce the gadget. Each patent is owned by a different firm and all license fees fi are set
simultaneously. Given the values of fi (which we assume are per-unit fees), firm X sets the

121
smartphone price and consumers decide how much to pay.
(a) Show that, in equilibrium, each patent’s license fee is given by
a
f = n+1 .
P
Answer: Let c = fi be the firm’s cost, that is, the sum of all license fees it must pay per
unit that it sells. From Section 3.2, we know that optimal price is given by (a + c)/2. This
implies that q = (a − c)/2 smartphones will be sold. Therefore, patent i license revenues
are given by !
a−c 1
X
πi = fi = 2 fi a − fk [1.5]
2
k
Maximizing πi with respect to fi implies
!
X
1
2 a− fk [1.5] − 12 fi = 0
k

By symmetry, fi = f and πi = π. This implies

1

1
2 a − nf − 2 f = 0

or thus
a
f=
n+1
 2
1 a
π=
2 n+1

Now suppose that the n patent holders form a pool and jointly set license fees.
(b) Determine the optimal license fees set by the pool.
Answer: Assuming that there is a license pool is equivalent to assuming that there is only
one license fee to pay, the license for the bundle of n patents. This corresponds to setting
n = 1 in the above equations, which yields
a
fp =
2
1  a 2
πp =
2 2
where the subscript p stands for “patent pool.”
(c) Show that patent holders, firm X and smartphone buyers are all
better off if a patent pool is formed.
Answer: Each individual patent holder receives 1/n of the total license fee, that it,
 2
πp 1 a
π= = √
n 2 2 n
This is a greater number than under no patent pool if and only if
√
2 n<n+1

122
which is true for all n > 1. Firm X’s cost under no patent pool is given by
na
c = nf =
n+1
This is greater than fp , firm X’s cost when there is a patent pool, if and only if

n 1
>
n+1 2
which is true for all n > 1. Finally, since cost is lower under a patent, so is the final
product price, and so consumers are also better off.
16 Networks

16.1. B2B. You have created a business-to-business (B2B) Internet venture directed at
an industry with exactly fifty (50) identical firms. Your services allow these firms to do
business with each other more efficiently as members of your trading network. You plan to
sell access to your service for a price p per member firm. Each firm’s benefit from the
service is given by 2 n, where n is the number of other firms joining the B2B network as a
member. So, if 21 firms join your service, each places a value of 2 × 20 or 40 on
membership in your network.
Suppose for part (a) that you set the price, p, and then firms simultaneously and
independently decide whether or not to join as members.
(a) Show that, for a price greater than zero and lower than 98,
there exist exactly two Nash equilibria in the simultaneous-move game
played by firms deciding whether or not to join the network as mem-
bers.
Answer: Suppose that no firm joins the network. Then the benefit for an individual firm
to join the network is zero. If price is positive, the net benefit is negative, which implies
that it is a best response not to join the network, which in turn confirms the conjecture
that no firm joins the network. We thus have a Nash equilibrium where no firm joins the
network for any positive price.
Suppose now that each firm conjectures that all of the other firms will join the
network. The expected benefit from joining the network is therefore 98 = 2 × 49. If price
is less than 98, the net benefit is positive, which in turn confirms the conjecture that all
firms join the network. We thus have a Nash equilibrium where all firms join the network
for a price less than 98.
Suppose that you are able to persuade 10 firms to join the network at an initial
stage. At a second stage, you set a price for the remaining 40 firms. These 40 firms then
simultaneously decide whether to join the network (as in (a)).

Suppose for part (b) that you are able to persuade 10 firms to join your network at an
initial stage as “Charter Members.” At a second stage, you set a price for the remaining
40 firms. These 40 firms then simultaneously decide (as in part (a)) whether to join your

124
network as regular members.
(b) For each price p, determine the equilibria of the game played
between the remaining 40 firms in the second stage.
Answer: Each of the second-mover firms knows that the number of adopters is at least 10.
It follows that the benefit from joining the network is at least 2 × 10 = 20. Therefore, if
price is less than 20, then the zero-adoption equilibrium is no longer a Nash equilibrium.
Only the full-adoption equilibrium remains. For higher prices, however, the two equilibria
are possible, for the same reasons as in (a).
Suppose that, when there are multiple Nash equilibria, firms behave conservatively
and conjecture that the low-adoption equilibrium will be played. (Note that, by definition
of Nash equilibrium, this conjecture is self-fulfilling.)

Finally, consider the same situation as in part (b), but suppose that, when there are
multiple Nash equilibria, firms behave conservatively and conjecture that the low-adoption
Nash equilibrium will be played. (Note that, by the definition of Nash equilibrium, this
conjecture is self-fulfilling.)
(c) How much would you be willing to pay (in total to all 10 early
adopters) in order to persuade the first 10 firms to join the network
as Charter Members?
Answer: If no firm joins the network in the first stage, then the game in the second stage
is as described in (a). Since for any positive price there are two equilibria and firms
behave “conservatively”, it follows that no firm joins the network in the second stage and
profits are zero. If however the 10 firms do join the network at the initial stage, then, in
the second stage, you can set a price of up to 20 and know that all firms will join the
network (since this is the only Nash equilibrium and firms know that; in fact, joining the
network would be a dominant strategy). For a price P = 20, this leads to profits
20 × 40 = 600. We conclude that you should be willing to spend up to 600 to persuade the
first 10 firms to join the network.

16.2. Spreadsheet network effects. Empirical evidence suggests that, between 1986
and 1991, consumers were willing to pay a significant premium for spreadsheets that were
compatible with the Lotus platform, the dominant spreadsheet during that period.44
What type of network externalities is this evidence of?
Answer: One can think of this both as a type of direct and indirect externality. For
example, exchanging files with users of the Lotus package, one has to own a compatible
spreadsheet. This is a case of direct network externality. On the other hand, developments
in the Lotus software can (potentially) be easily adapted and adopted by compatible
software products, which is a case of indirect network externality.

16.3. Home computers. People are more likely to buy their first home computer in
areas where a high fraction of households already own computers or where a large share of
their friends and family own computers: a ten percent greater penetration in the
surrounding city is associated with a one percent higher adoption rate.45 How can this be
explained by network externalities? What alternative explanations are there?

125
Answer: The larger the number of households/friends that own computers, the larger the
the possibilities for direct communication among users and the greater utility one derives
from owning a computer. Communication among users may consist of email exchange,
learning how to use a given software, exchange of files, and so forth.
An alternative explanation is that certain areas attract more “sophisticaded” users
than other areas. If this were the case, those areas would have a higher penetration rate
and a higher adoption rate. But then the correlation between penetration and adoption
would be just that—correlation, not causality. Another situation of
correlation-not-causality is when a higher penetration implies a greater degree of
competition in the computer market, lower prices and a higher adoption rate.
See the cited reference for additional alternative explanations.

16.4. ATMs. In the early days of Automated Teller Machines (ATMs), there were very
few interbank networks, that is, each bank’s network was incompatible with the other
banks’. Empirical evidence shows that banks with a larger network of branches adopted
ATMs earlier. To what extent can network externalities explain this observation?46
Answer: Network effects imply that the value of using a given ATM system is increasing
in the number of ATM machines compatible with that system. Part of the value created
by a network is gained by the consumer, part by the banks. Assuming (as is empirically
observed) that the cost of adoption is decreasing over time, the above observation implies
that the critical moment in time at which it pays to adopt ATMs is earlier the greater the
network effect. Since the number of branches is a good proxy for the number of ATMs,
this implies that banks with a greater number of ATM machines are likely to adopt first,
as the evidence shows was indeed the case.

16.5. Apple and Microsoft. How would you respond to the following quotation:
Apple Computer, the company that brought you the idiot-friendly Macintosh,
is staring at bankruptcy. Meanwhile, the great army of technocrats at
Microsoft, which only last year managed to reproduce the look and feel of a
1980’s Mac, lumbers on, invincible.
A bad break for Apple? A rare exception to the Darwinian rules in which the
best products win the hearts and dollars of consumers?47

Answer: This is an instance of how network externalities work and how “the best
technology” does not always win. As seen in the text, it is possible that the lock-in to the
Microsoft operating system resulted from a series of “small historial events,” not from the
inherent superiority of the Microsoft operating system.

16.6. Excess inertia and excess momentum. Explain why the market adoption of a
new technology may be too fast or to slow.
Answer: The adoption of new a technology may feature excess inertia because of
incomplete information about the preferences of future potential adopters. Even a slight
probability that future users are conservative and therefore are better off not adopting the
new technology, can act as a deterrent for early users that would prefer the new
technology.

126
 On the other hand, excess momentum may appear when sticking to the old
technology is not a dominant strategy. That is, if initial users prefer the new technology,
future users will also choose the new technology, despite the fact that the utility from
sticking to the old technology if the initial users were also conservative, is much higher.

16.7. Compatibility decisions. Company A has just developed a new technology.

Company B approaches Company A, stating it has developed its own version of the
technology and proposing a compromise that would make the two technologies compatible
with each other. What advice would you give Company A?
Answer: If the firms decide to go alone, standards competition reduces the
product-market size for both of them, and they may end up loosing, since users have the
easy option of staying with the old technology (see the example in box 17.3). Therefore,
compatibility is the prefered action.
On the other hand, if product competition is fierce when the products are
compatible, then the two firms will get duopoly profits which are lower than the
expectation of profits from refusing the compromise (case in which each firm receives
monopoly profits with probability 0.5). In this case, the decission to go alone is better.

16.8. Wireless device. You are marketing a new wireless information device (WID).
Consumers differ in their willingness to pay for the device. (No one needs more than one.)
All consumers value owning a WID more highly, the larger is the total number of
consumers using such devices. Denote the expected total number of WID users by ne ,
which we also can call the “expected size of the WID network.”
If all consumers expect the size of the WID network to be ne , and the price of the
device is p, then the number of users who will want to buy the device (i.e., the total
quantity demanded) is given by n = 100 − p + vne , where 0 < v < 1. (Note that this is a
standard linear relationship between price and unit sales for any given level of expected
network size, ne .)
(a) Interpret the parameter v. What factors influence v?
Answer: The parameter v is the network benefit contributed by each additional WID
sold. One way to think about this is as follows. When an additional WID is sold, ne
increases by one. This raises the benefit to all consumers from owning a WID since now
there is one more WID out there with which they might communicate. Adding up this
small improvement in the value of a WID for all consumers gives us the parameter v.
Suppose that your marginal cost per WID is 20. Suppose also that consumers are
quite sophisticated and form accurate expectations about the size of the WID network, for
any price p that you might set.

Suppose that your marginal cost per WID is 20. Suppose also that consumers are quite
sophisticated and form accurate expectations about the size of the WID network, for any
price p that you might set, so that n must equal ne .
(b) What is the profit-maximizing price of WIDs? How many are
sold, and what profits do you earn?
Answer: Since consumers have accurate expectations we can set ne equal to n and invert

127
the total demand for WIDs to get p = 100 − (1 − v) n as the inverse demand curve and a
marginal revenue curve of MR = 100 − 2(1 − v) n. The optimal number of WIDs to sell is
that which equates this marginal revenue with marginal cost. That is,
100 − 2 (1 − v) n = 20 or n = (80)/2 (1 − v). The profit maximizing price is then
p = 100 − (1 − v) (80)/2 (1 − v) = 60 and profits are (60 − 20) 80/2 (1 − v) = 1600/(1 − v).

Suppose that you could improve the performance of your WID communications network
and thus enhance the network effects, raising v from 1/3 to 1/2.
(c) How much would you pay to develop this enhancement?
Answer: When v = 1/3, the firm’s profits are 1600/(1 − .333) = $2, 400. When v = 1/2,
the firm’s profits are 1600/(1 − .5) = $3, 200. Therefore, the most the firm would pay to
develop this enhancement is $3,200-$2,400=$800.

16.9. Compress and Squeeze. Two firms, Compress and Squeeze, offer incompatible
software products that encrypt and shrink the size of large data files for safe storage
and/or faster transmission. This software category exhibits strong network effects, since
users seek to send files to each other, and a file saved in one format cannot be retrieved
using the other format. The marginal cost of serving one customer is $40 for either firm.
To keep things simple, suppose that there are only two customers, “Pioneer” and
“Follower,” and two time periods, “This Year” and “Next Year.” As the name suggests,
Pioneer moves first, picking one format This Year. Pioneer cannot change her choice once
it is made. In contrast, Follower picks Next Year. Follower will be aware of Pioneer’s pick
when the time comes for Follower to pick. The annual interest rate is 20% for both
Compress and Squeeze and Pioneer.
Pioneer regards Compress and Squeeze as equally attractive products. Pioneer values
either product at $100 during This Year (before Follower enters the market), and at $100
during Next Year if Follower does not pick the same product. If Follower does pick the
same product Next Year, Pioneer’s value during Next Year will be $136. (In other words,
the network effect is worth $36 to Pioneer.)
Follower has very similar preferences. If Follower picks the same product Next Year
as Pioneer did This Year, Follower values that product at $136. Alternatively, if Follower
picks a different product Next Year than Pioneer did This Year, the value to Follower of
that product will be only $100.
Finally, suppose that Compress and Squeeze simultaneously set prices This Year at
which they offer their products to Pioneer. (One could just as well say that they bid for
Pioneer’s business.) Then Compress and Squeeze simultaneously set prices Next Year at
which they will offer their products to Follower.
For simplicity, please assume that Pioneer will pick Compress if Pioneer is just
indifferent between Compress and Squeeze, and that Follower will pick the same product
as Pioneer if Follower is indifferent between Compress and Squeeze given the values they
offer and the prices they charge.
(a) What prices will Compress and Squeeze set Next Year in bidding
to win Follower’s business if Compress wins Pioneer’s business This
Year?
Answer: If Compress wins during the first period, then Compress offers an extra $36

128
value over Squeeze to Follower. Squeeze will compete as best possible by offering its
product at cost, $40, but Compress can win by charging $76. (We could make this $75.95,
but the numbers are simplified by breaking ties in favor of Compress.)
(b) What prices will Compress and Squeeze set This Year in bidding
to win Pioneer’s business?
Answer: The equilibrium derived in A generates profits Next Year to Compress of $36,
which are equal to $30 in This Year dollars (given the 20% interest rate). By symmetry,
Squeeze would also enjoy profits of $36 Next Year if Pioneer picks Squeeze this year. This
implies that both Compress and Squeeze are prepared to set a price as low as $10 to win
Pioneer’s business: losing Pioneer’s business means they will lose Follower’s business as
well and earn zero; bidding $10 means losing $30 This Year and earning profits of $36
Next Year, which gives zero in present discounted value.
(c) What product will Pioneer buy, and what product will Follower
buy?
Answer: From the answer to B, we conclude that the Nash Equilibrium involves a bid of
$10 by each firm to serve Pioneer. Compress thus will win Pioneer’s business, by our
tie-breaking convention. Then Squeeze will bid $40 to serve Follower, and Compress will
bid $76 to serve Follower. Compress will win, so both customers will buy from Compress.
(d) What are the resulting payoffs of Compress, Squeeze, Pioneer,
and Follower?
Answer: Both firms earn zero profits in present discounted value. All of the profits are
dissipated by bidding for Pioneer’s business, since Pioneer “tips” the market towards one
product or the other. Pioneer gets a surplus of $90 This Year and $136 Next Year.
Follower gets surplus of $60 Next Year.
(e) Describe in words the advantages of early or late adopters iden-
tified in this problem.
Answer: Pioneer enjoys a nice strategic advantage by virtue of its ability to “tip” the
market, i.e., to influence subsequent adopters.
(f) How does all of this change if there is rapid technological progress
so that costs Next Year are much lower than costs This Year?
Answer: With rapid technological progress, prices fall rapidly and Follower could well do
better than Pioneer, simply because Follower can buy when the product is much cheaper
to produce (or of higher quality). From the customer’s perspective, waiting for products
to improve must be balanced against the benefits of adopting early and thus enjoying very
strong price competition between incompatible suppliers seeking to build their installed
bases and thus gain competitive advantage.
(g) How does your analysis change if the (marginal) cost of serving
a customer is only 20 rather than 40?
Answer: If marginal costs are only $20, then the price during This Year would be -$10.
The problem here is that many “phantom” customers could appear, take the $10 subsidy
to use the product, and then disappear. Actually paying customers to take your product
can be a very dangerous strategy. Are you building an installed base of users or just

129
giving away money to opportunistic “fake” customers?

16.10. WalkDVD. Technological progress (of a sort) has led to the WalkDVD. As the
name suggests, this is a miniature DVD player. It is attached to a pair of headphones and
special viewing glasses which, together, allow for highly realistic sound and image effects,
as well as easy mobility. Three firms, Son, Tosh and Phil, are planning to launch their
WalkDVD players. There are two possible formats to choose from, S and T, and the three
competitors have not agreed on which standard to adopt. Son prefers standard S, whereas
Tosh prefers standard T. Phil does not have any strong preference other than being
compatible with the other firms. Specifically, the payoffs for each player as a function of
the standard they adopt and the number of firms that adopt the same standard are given
by the table below. For example, the value 200 in the Son row and S2 column means that
if Son chooses the S standard and two firms choose the S standard, then Son’s payoff is
200.

Firm S1 S2 S3 T1 T2 T2
Son 100 200 250 40 80 110
Tosh 40 80 110 100 200 250
Phil 60 100 120 60 100 120

Suppose that all three firms simultaneously choose which standard to adopt.
(a) Show that “all firms choosing S” and “all firms choosing T” are
both Nash equilibria of this game.
Answer: Suppose all firms choose S. By unilaterally deviating and choosing T instead,
Son would get 40 instead of 250; Tosh would get 100 instead of 110; and Phil would get 60
instead of 120. Since all would stand to lose, we conclude that all with S constitutes a
Nash equilibrium. By the same token, all choosing T is also a Nash equilibrium.
(b) Determine whether there are any other Nash equilibria in this
simultaneous-move game
Answer: The only other possible (pure-strategy) Nash equilibria are for two firms to
choose one standard and one to choose the other one. But such a situation cannot be a
Nash equilibrium: the firm that is the sole adopter of one of the standards would be
better off by joining the other firms. We conclude that there are no (pure-strategy) Nash
equilibria in addition to the ones derived in the previous question.

Son has just acquired a firm that manufactures DVDs for the S format. For all practical
purposes, this implies that Son is committed to the S format. It is now up to Tosh and
Phil to simultaneously decide which format to choose.
(c) Write down the 2x2 payoff matrix for the game now played by
Tosh and Phil. Find the Nash equilibrium of this game.
Answer: The game is as follows:
Phil

130
 S T
120 60
S
110 80
Tosh
100 100
T
100 200

Notice that Phil has a (weakly) dominant strategy: to choose S. Even if Phil assigns the
lowest positive probability that Tosh is going along with S, it is strictly better off (in
expected value) by choosing S. Knowing this, Tosh should choose S, since, conditional on
Phil choosing S, payoff is greater with S (110) than with T (100). We conclude that both
firms choose S.
Notice that (T,T) is also a Nash equilibrium. However, the discussion above implies
that it would not be a very “reasonable” Nash equilibrium.
(d) Do you think Son’s move was a good one? How would your an-
swer differ if Phil had a slight preference for the T format (e.g., assume
that payoffs for T1, T2 and T3 are 70, 110 and 130, respectively)?
Answer: Son’s move was a brilliant one. In the simultaneous-move game, there are two
Nash equilibria, one that is good for Son, one that is not so good. By moving ahead of the
other players, Son is effectively able to lead the industry to adopt its preferred standard.
In the event Phil prefers the T standard, things are different. The game played
between Tosh and Phil is now the following
Phil
S T
120 70
S
110 80
Tosh
100 110
T
100 200

There are now two Nash equilibria: (S,S) and (T,T). Moreover, joint payoff are greater in
the (T,T) equilibrium (310) than in the (S,S) equilibrium (230). It is not unlikely that
Tosh and Phil choose T, in which case Son is worse off by choosing S than by choosing T
as well.
(e) Suppose now that all firms’ payoffs are like Phil in the table
above. You are Son. If you could choose, would you rather move
before Tosh and Phil, or after them? Contrast your answer to what
you have learned from the answers to parts (c) and (d).
Answer: If all firms have payoffs as Phil, then a particular firm would prefer to move after
the other firms have moved than to move at the same time. As in part A, there are two
Nash equilibria in the simultaneous-move game. Moreover, both equilibria yield each firm
the same payoff. If firms are able to coordinate perfectly on which equilibrium to choose,
then moving at the same time or after does not make a difference. If however there is a
small chance that coordination will fail, then moving later is (weakly) better, as it reduces
the probability of mis-coordination.

131
 16.11. After-sales service. Consider the market for a given piece of hardware — a
photocopier of brand x, for example — that needs after-sale servicing. Suppose that there
is free entry into this after-market. Servicing photocopiers implies a fixed cost of F and a
marginal cost of c per unit of service provided. Total demand for servicing is given by
D = S(a − p), where p is price and S the number of photocopier owners. Finally, suppose
that firms in the after-market compete à la Cournot.
(a) Show that consumer surplus (per consumer) in the after-market
is given by
r !2
1 F
U= a−c−
2 S
an increasing, concave function of S. (Hint: apply the results on
Cournot competition with free entry derived in Chapter 10. Take into
account the fact consumer surplus per consumer is given by (a−p)2 /2.)
 q 
Q (a−c)S a−c 2 F S
Answer: We have p = a − S and qi = n+1 , ( n+1 ) = S and n = (a − c) F − 1 (see
∗ ∗ ∗2
derivation in 14.10). Consumer surplus is given by CS = (p(0)−p 2
)Q
= Q2S . Therefore,
consumer surplus per consumer is q q
Q∗2 n2 (a−c)2 S 2 n2 F
U = CSS = 2S 2
= 2S 2 (n+1)2
= 2S = 1
2 [(a − c) S
F − 1]2 F = 1 (a − c −
S 2
F 2
S) .

(b) Relate this result to the discussion of indirect network external-

ities (at the beginning of the chapter).
Answer: This is a case of indirect network externalities. The greater the market size (S),
the greater the need for after-sale services, and hence the greater competition in the
after-sale market. This increase in competition lowers the price for after-sale services and
increases consumer surplus.

132