distribution of σ: so the expected willingness to pay of the marginal consumer is increas-
ing in α. Conversely, for a sale probability above 1/2, the marginal consumer’s expected
willingness to pay is decreasing in α. Thus in the figure, α
0
< α
00
. This means that the
better informed is the consumer, the steeper is the inverse demand curve. In particular,
if the consumer has no information, α ¼0, then the inverse demand is flat at willingness
to pay 1/2.
If the firm can choose any α 2 0, 1½, it chooses either α ¼0orα ¼1. Suppose, on the
contrary, that the firm chooses α 2 0, 1ðÞ. If it sells with probability q < 1=2, then if it
makes the demand curve steeper by providing more information, it can sell with the same
probability at a higher price. Similarly, if it sells with probability q > 1=2, it can increase its
profit by providing less information. As a result, the firm chooses between full informa-
tion and no information. As
Johnson and Myatt (2006) emphasize, no information means
that the inverse demand is flat and the firm’s strategy is to sell to the entire buyer pop-
ulation at an intermediate price, whereas with full information the firm chooses a steep
inverse demand that allows for charging a high price to the consumers in the “niche” of
the product.
Which factors determine whether the firm chooses no information or on the contrary
full information? First consider changes in marginal cost. If it is zero, then by providing no
information, the firm can charge a price of 1/2 and sell with probability 1. This yields the
perfect price discrimination profit 1/2: indeed, perfect discrimination would involve
charging a price of r to the consumer for all realizations, which would yield the same
expected profit. Then the no-information profit clearly exceeds that under full informa-
tion, which is the uniform price monopoly profit against the full information demand,
1/4. It can be shown that the firm’s profit-maximizing behavior is to hide information if
0.5
0.625
0.75
1
p
0.25
0.375
0.5 1
q
P (q, α )
P (q, α )
Figure 4.2 Improved match information, where P(q, a) denotes the inverse demand when the signal is
informative with probability a.
148
Handbook of Media Economics
marginal cost is low and reveal full information if marginal cost is large.
22
Lewis and
Sappington (1994)
also argue that the firm finds it profitable to fully reveal information
only if match values are heterogeneous enough.
23
The original analysis by Lewis and Sappington (1994) actually postulates a very gen-
eral form for the informative signal.
24
Still, the firm cannot choose its characteristics (it
only chooses the probability that a consumer receives it). As a result, they do not char-
acterize the optimal informative signal for the firm. It turns out that this question has a
very simple answer in their experience good setting. Consider a simple signal that only
tells the consumer whether or not her match exceeds marginal cost c: formally σ 2 ‘, h
fg
and σ ¼h with probability 1 if r c, σ ¼‘ with probability 1 if r < c. Now suppose the
firm charges a price of Erjr cðÞ. Then the consumer buys if and only if she learns that
r c. Furthermore her expected surplus from buying is zero. This implements the perfect
price discrimination outcome: the firm only sells to the consumer if she is willing to pay
more than marginal cost and it captures the entire social surplus. It is therefore the best the
firm can achieve.
25
Hence the optimal solution involves partial information disclosure, in
contrast with the finding of
Lewis and Sappington (1994) and Johnson and Myatt (2006).
This first-best solution should be viewed as a benchmark. There is no guarantee that the
firm has access to appropriate disclosure strategies to implement it. Note however that,
because the firm perfectly controls the consumer’s access to product information, it can
implement this solution merely by selling the access to information to the consumer at
some price that extracts all ex ante surplus and then price at marginal cost.
Meurer and Stahl (1994) and Anderson and Renault (2009) study match information
advertising in a discrete-choice, differentiated product duopoly setting. In both papers,
match disclosure decisions are taken in a first stage prior to price competition and partial
22
To see this, note that, using the envelop theorem on the full information profit, it is convex and decreasing
in c, and it has just one crossing point with the no-information profit, which is linear and decreasing
(because the no-information profit crosses from above).
23
A simple intuition for this is that, if matches are all very close to 1/2 and c < 1 =2 so that the firm can sell
while providing no information, then all matches are above marginal cost and no information implements
the perfect discrimination profit. If, on the contrary, matches are widely spread around 1/2, then with no
information the firm sells to a significant mass of consumers whose willingness to pay is below marginal
cost, so that it would be better that they are excluded from buying. If it fully reveals information, the firm
can screen out those consumers and charge a price substantially above 1/2 while still selling to the large
mass of consumers with very high valuations.
24
They merely require a first-order stochastic dominance between the consumer’s posterior beliefs after
observing a more or less favorable signal. The version presented here can be found in
Johnson and
Myatt (2006)
.
25
This result was first derived by Saak (2006). It is also related to the analysis in Anderson and Renault
(2006)
, who show, in a search good setting, that this type of threshold signal does at least as well as
any other informative signal (see
Section 4.3.3.2). Their argument can easily be adapted to the simpler
case of experience goods.
149
Advertising in Markets
match advertising is ruled out. In Meurer and Stahl (1994), match values for the two
products are perfectly and negatively correlated so that match disclosure by either firm
fully informs a consumer. By contrast,
Anderson and Renault (2009) assume i.i.d. match
values but they allow a firm to choose between providing no information, information on
its own match only, or information on both matches thus using comparative advertising.
Anderson and Renault (2009) assume zero advertising costs. They first consider a sit-
uation with symmetric firms where, without any match information, both products
would be perceived as identical by consumers. This would result in Bertrand competition
with zero profits. More match information always increases both firms’ profit so it is a
weakly dominant strategy to reveal own-match information. As a result, full match infor-
mation is disclosed, whether comparative advertising is barred or not.
Meurer and Stahl
(1994)
also find that match information is profit enhancing. They, however, consider
costly advertising reach so that there is not enough of it from the producers’ point of
view.
26
Comparative advertising becomes a very relevant practice when firms are ex ante
asymmetric enough, either in terms of production costs or because there is a systematic
quality advantage Δq > 0 in favor of one of the products that consumers take into
account in their purchase decision, in addition to the match difference and the price dif-
ference. Then comparative advertising is used by the “weak” firm with the small market
share (high costs/low quality) so as to attract those consumers who have a very strong
match preference for its product. The “strong” firm, however, prefers that limited match
information be conveyed to consumers so they make their purchase decision mostly on
the basis of the quality and/or price difference for which it has a large advantage. When
comparative advertising is used by its competitor, it lowers its price, which benefits most
consumers because they buy from the large firm. This, combined with the better match
information imparted by comparative advertising, improves consumer surplus over what
it would be if comparative advertising was disallowed. Total surplus, however, deterio-
rates because too many consumers buy the low-quality product.
27
A few observations are noteworthy regarding the comparison between monopoly and
oligopoly. First, competition may provide added incentives for the firms to disclose
match information. Both in
Meurer and Stahl (1994) and the symmetric version of
Anderson and Renault (2009), full product information is disclosed, whereas a monopoly
firm may choose to withhold such information (e.g., when its marginal cost is low
26
They also find that advertising may be excessive or insufficient relative to the social optimum. See Bagwell
(2007)
for more details.
27
Emons and Fluet (2012) consider a setting where consumers know horizontal match but not quality. They
compare direct advertising where each firm may only advertise its own quality to comparative advertising
where a firm advertises the quality difference. In the latter case, only one firm advertises in equilibrium.
They analyze both the case of direct quality information and the case of quality signaling through ad
expenditures.
150
Handbook of Media Economics
enough).
28
The simple intuition for this is that product information relaxes price com-
petition. This price-increasing impact of information also arises under monopoly: once
information is revealed, a firm caters mostly to those consumers who have a strong pref-
erence for its product (over the competitor or the outside option). However, when firms
are asymmetric,
Anderson and Renault (2009) show that more match information in the
form of comparative advertising may induce a drop in price for the firm with the large
market share.
I now turn to the disclosure of match information for a search good so that the con-
sumer may obtain product information before purchase even if the firm does not
advertise it.
4.3.3.2 Search Goods and Advertising Content
The focus so far has been on experience goods, implying that the firm has full control
over the product information that is available to consumers prior to purchase. Further-
more, the analysis in the previous subsection implicitly assumes that price information is
freely available to the consumer. It is therefore not possible to discuss the availability of
price information in advertisements. The framework I introduce next allows for both
analyzing advertising of search goods and discussing the advertiser’s choice of including
price and/or product information in ads.
A search characteristics as defined by
Nelson (1970) is such that the consumer may
observe it prior to purchasing the product. If such information could be accessed at
no cost, then there would be no role for advertising product information. The discussion
of search characteristics advertising in
Nelson (1974) clearly incorporates some costs for
the consumer to acquire the pre-purchase information: he explicitly mentions
“transportation costs of consumers” (p. 730). These costs may naturally be construed
as visit costs required to get to the store (or on the seller’s web page), inspect the product,
and then possibly purchase it.
Anderson and Renault (2006) consider a monopoly setting
that captures this simple idea. It is a straightforward extension of the monopoly setting
introduced above. Let us now assume that the buyer may only purchase the product after
incurring a visit cost γ > 0. Once this cost is sunk, and prior to deciding whether to buy or
not, she becomes perfectly informed about the product and hence knows the realization
of r perfectly as well as the price charged by the firm. The firm may use advertising to
inform the consumer about the product and/or the price before she decides whether
to visit or not. The main objective is to determine the optimal choice of ad content
for the firm. Because the consumer is initially uninformed about r, all consumer types
are ex ante identical. I now assume that marginal cost is c ¼0 so that, given the standard
uniform distribution of the match, the unique monopoly price is 1/2.
28
It is, however, not the case that competition necessarily leads to more information revelation, as I explain
in
Section 4.3.4.
151
Advertising in Markets
If γ is small enough, the firm may obtain the monopoly profit without resorting to any
advertising. If the consumer observes no information before deciding whether to visit,
the firm cannot infer anything about the match value from observing a visit. Hence,
it charges the monopoly price 1/2. Expecting this, it is optimal for the consumer to visit
as long as the visit cost is less than her expected surplus from visiting and then buying if
and only if r 1=2: this expectation is merely the standard consumer surplus, which is
here 1/8. Hence, in order for advertising to have a role, visit costs should be at least 1/8.
I now turn to an important insight, which is a reformulation of the unraveling result of
Stiglitz (1979). Suppose that the firm only advertises product information and provides
full information to the consumer. The consumer, when deciding whether to visit, knows
the realization of r but not the price. In equilibrium she rationally anticipates that the firm
will charge some price p
. It is then optimal to visit if and only if r p
+ γ.Ifp
+ γ < 1
then the consumer visits whenever her match is large enough. The firm can then infer
that the visiting consumer is willing to pay at least p
+ γ. But this means that it could up
its price to this level and still sell to the consumer, thus making more profit than at price
p
. Hence p
cannot be an equilibrium price. This means that, in equilibrium, the con-
sumer must expect a price above 1 γ so she does not visit, no matter how high her
match value is. As a result the market unravels.
29
Such advertising is therefore not prof-
itable for the firm. In short, the consumer expects to be held up.
However, a firm need not always include a price in its ad in order to circumvent the
holdup problem. Suppose the firm can post an ad that merely informs the consumer
whether her match is above or below the monopoly price, 1/2. Then if the consumer
learns she is willing to pay at least 1/2, she expects her match to be uniform on [1/2, 1]
and choose to visit as long as γ is at most her expected surplus at price 1/2,
Ð
1
1=2
r 1=2
ðÞ
2dr ¼1=4. A consumer learning that her match is below 1/2 will not visit.
The firm anticipating that a consumer who visits is willing to pay at least 1/2 does not
choose to charge less. Charging a larger price would also yield ^r a lower profit because the
demand elasticity at 1/2 is 1. It is therefore credible that the firm will charge 1/2 to a
consumer who visits and we have an equilibrium. Thus, using this simple threshold dis-
closure strategy enables the firm to retain its monopoly profit for visit costs as high as 1/4.
For larger visit costs, it is no longer possible for the firm to sustain monopoly profit.
Nor is it possible to survive without resorting to price advertising. This is because the
consumer is no longer willing to visit, expecting the monopoly price even if she is reas-
sured that she is willing to pay that price. The firm therefore needs to further sweeten the
29
This argument can readily be generalized to mixed strategies, oligopoly, and ex ante heterogeneous con-
sumer types, as long as they all have a strictly positive visit cost. Such unraveling, however, would not arise
if the buyer had a price-sensitive demand rather than a unit demand and the visit cost was not too large.
Alternatively, if the firm sells multiple products as in
Rhodes (2015), then the holdup problem is circum-
vented because the firm cannot infer a precise enough lower bound on a consumer’s willingness to pay for
each of the products.
152
Handbook of Media Economics
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.