consumer is perfectly informed and may buy at no additional cost. As pointed out by
Shapiro (1980), this can be accounted for in the demand-shifting monopoly setting by
assuming that the pre-advertising valuation is zero for those consumers who are initially
unaware of the firm’s existence, and then using the post-advertising valuations to evaluate
welfare. For instance, in the two-consumer setting, if consumer 1 is initially aware of the
firm’s existence and then advertising allows the firm to reach consumer 2, then we have
r
1
1ðÞ¼r
1
0ðÞand r
2
1ðÞ> r
2
0ðÞ¼0. Now also recall that in my discussion of Section 4.2.4,
following
Tirole (1988), I explain that it is typically ambiguous whether market-provided
advertising is excessive, because of the two countervailing effects, “non-appropriation of
surplus” and “market stealing.” Under monopoly only the first effect remains because
there are no competitors to steal the market from. This suggests that the firm does
not advertise enough because it cannot fully appropriate the surplus it generates through
advertising.
Returning to the application where advertising reaches consumer 2 in addition to
consumer 1, either r
2
1ðÞr
1
0ðÞ, in which case the monopoly price would be weakly
smaller with advertising, or r
2
1ðÞ> r
1
0ðÞ, in which case the firm charges r
1
1ðÞ¼r
1
0ðÞ
with advertising so price is unaffected. Hence the analysis of price increasing advertising
of
Dixit and Norman (1978) never applies. Furthermore, it is readily seen that advertising
is weakly under-provided by the firm and the only case where the firm’s behavior is
socially optimal is when r
1
1ðÞ¼r
2
1ðÞ, a case where the firm can extract the entire social
surplus through its monopoly price. If we allow for any form of heterogeneity in tastes, so
the firm cannot fully appropriate consumer surplus, then monopoly advertising is insuf-
ficient as predicted by the discussion of
Section 4.2.4.
Now consider the provision of product information. As I briefly mention in the con-
cluding remarks to
Section 4.3, there is no clear-cut general conclusion regarding the
firms’ incentives to provide product information as compared to what would be socially
optimal. For instance,
Koessler and Renault (2012, p. 645) provide an example where the
firm wishes to disclose full certified information (and this is a unique equilibrium) and yet
it would be preferable for social welfare that it is incapable of certifying product infor-
mation (even if certification and disclosure involve no cost). And there are some obvious
cases where the ability of a firm to certify and disclose product information at no cost is
socially desirable. Here I just consider the canonical case of quality disclosure, as in
Sections 4.3.4.1 and 4.5.1.2.
Favorable quality information should ameliorate the product’s appeal to consumers
and therefore shift demand outward as has been assumed above. Furthermore,
post-advertising valuations are clearly the relevant measures of gross consumer surplus.
Provided that price increases as a result and that the order of consumer valuations is
the same whether consumers know that quality is good or not, then the analysis of
Dixit and Norman (1978) shows that the monopoly firm excessively provides quality
information. However, a full welfare analysis must also account for the impact of
189Advertising in Markets
information revelation when quality turns out to be lower than expected by consumers.
This is what is done in the analysis of quality disclosure in
Section 4.5.1.2. I now consider
a reinterpretation of the two-consumer setting above that allows for a full welfare analysis
of monopoly quality disclosure.
Assume the product’s quality is either high or low, where q 2 0, 1ðÞis the probability
that quality is high. Then, for i ¼1, 2, r
i
(0) is the valuation of consumer i if she does not
know quality while r
i
(1) is consumer i’s valuation for the high-quality product. Assume
that the firm sells only to consumer 1 if consumers do not know quality, and sells to both
consumers if quality is revealed and is high. Further assume that r
2
1ðÞr
1
0ðÞand that
consumer 2 is the marginal consumer when quality is known to be high. Then the perfect
information price of the high-quality product is at least as large as the price charged by the
monopoly firm when its quality is not known. From the analysis above, the high-quality
firm’s incremental profit from disclosing quality is at least as large as the social surplus it
generates by revealing its quality and selling to one more consumer. Let us now assume
that the low-quality firm’s quantity resulting from information disclosure is no larger than
the quantity sold when quality information is not disclosed. Further assume that the
monopolist under-provides the low-quality product even when quality is unknown
so that any drop in the quantity sold as a result of quality revelation diminishes social sur-
plus. Then quality disclosure decreases social surplus if quality is low. As a result, it is
unambiguous that the high-quality firm’s incentives to provide quality information
are excessive as compared to what would be socially optimal.
The above result that the disclosure of quality information by the monopoly firm is
excessive coincides with the result derived in
Section 4.5.1.2 in my discussion of
Jovanovic (1982). The underlying motive for excessive disclosure is, however, quite dif-
ferent. In
Section 4.5.1.2, demand is inelastic and quality revelation can only contribute
to social surplus by preventing the sale of some units of the low-quality product that gen-
erates a negative surplus. Here, demand is elastic and all units of the low-quality product
that are sold with or without information generate some positive surplus. The point is to
show that under the assumptions that induce over-provision of persuasive advertising in
Dixit and Norman (1978) there is an excessive provision of quality information that is
even more severe under fairly reasonable conditions (that the monopolist sells an insuf-
ficient amount of the low-quality product even if consumers do not know quality and
does not sell more if quality is revealed to be low).
Returning now to persuasive advertising, the difficulty in evaluating its welfare impli-
cations is a result of the disconnection between consumer tastes and consumer behavior:
there is no unambiguous manner to infer one from the other. It is also clear that in order
for persuasive advertising to be possible, consumer rationality must be altered in some
way. One possible route to perform a proper welfare analysis is to assume stable prefer-
ences and model the bounded rationality explicitly. I explore this direction in the next
subsection by looking at some recent contributions.
190 Handbook of Media Economics
4.6.2 Consumer Naivety
There is obviously a great variety of manners in which we might wish to think of con-
sumer naivety. Here I discuss two formulations that have been proposed in some recent
and influential research with some applications to advertising: consumer unawareness (or
myopia) and coarse thinking.
Here again I use a simple monopoly framework to illustrate these ideas. A monopoly
firm with zero costs is selling a product to a consumer with unit demand. The product is
characterized by two characteristics, (x
1
, x
2
), where x
i
, i ¼1, 2, is either 0 or 1.
Consumer unawareness and shrouded attributes. Assume that t he value t o the consumer
of buying the product is given by ^r ¼u
X
i¼1,2
x
i
Δ
i
, with u > 0andwhereΔ
i
> 0rep-
resents the loss in utility for t he consumer of consuming a product with characteristic
x
i
¼1 rat her than x
i
¼0. In other words, the consumer considers each characteristic
independently, in evaluating how it affects her utility. She may, however, be unaware
that attribute 2 exists. This idea has been formalized in some general game theory con-
texts by various authors (see, for instance,
Geanakoplos, 1989 or Heifetzetal.,2006)
Whether she is aware or not depends on the firm’s communication. The firm selling
product (x
1
, x
2
) can send a certified message m 2 x
1
, ∅ðÞ, ∅, x
2
ðÞ, x
1
, x
2
ðÞ
fg
,where
∅ in the ith po sition in dicat es tha t the ith charact eristics are not mentioned. Now
assume t hat the cons umer is aware of characteristic 1 but remains unaware of charac-
teristic 2 unless it is mentioned by the firm. Letting A be the set of characteristics that are
mentioned in message m, t he consu mer’ s willingnes s to pay may then be writ ten a s
rx
1
, x
2
, m
ðÞ
¼u
X
i2 1fg[A
x
i
Δ
i
.
If the consumer was, from the outset, fully aware of the existence of the two char-
acteristics, then this would be a special case of the persuasion game of quality disclosure
described in
Section 4.3.4. The unique sequential equilibrium outcome would then be
that the firm fully reveals (x
1
, x
2
) and charges the consumer her true valuation r. Now
assuming that there is initial consumer unawareness, consider the following candidate
equilibrium. The high-quality firm type (0,0) announces m ¼ 0, ∅ðÞand so does firm
type (0,1). In both instances, the consumer is reassured that x
1
¼0, and since this is all
she cares about, the firm can charge a price of u
¯
and she buys. The other two types
announce message m ¼ 1, ∅ðÞand charge u Δ
1
. Obviously, no firm type could gain
by disclosing more or less information (assuming that when the firm does not disclose
the first characteristics, the consumer infers that x
1
¼1).
64
In this equilibrium, the con-
sumer pays her true valuation only if the product has the highest quality or if it is product
(1,0). If the product is either (0,1) or (1,1), she pays more than her valuation.
64
A proper formal analysis would require writing an appropriate definition for the equilibrium concept.
Heifetz et al. (2011) show more formally that unraveling may break down if there is some unawareness
on the buyer’s side.
191
Advertising in Markets
This simple example illustrates that, when consumers are not fully aware of all the
potential attributes, a firm may select to “shroud” some of these attributes. This is espe-
cially the case if finding out that these attributes exist can only deteriorate the consumer’s
expected valuation for the product, as compared to what she thought, based on the attri-
butes she was initially aware of. This may explain why, despite the persuasion game unra-
veling result, some negative information might remain hidden. This idea of shrouded
attributes has been applied by
Gabaix and Laibson (2006) to price information. Obvi-
ously, consumers are aware that price is a relevant dimension in determining the utility
associated with purchasing a product. However, relevant price information is often more
complex than a single price quote. Specifically, Gabaix and Laibson consider add-ons,
which are items that are complementary to the purchased product and that the consumer
may end up purchasing from the same provider. They mention such examples as printer
cartridges or bank account services and penalties. A rational consumer anticipates the
pricing of these items if the firm does not advertise it. She also foresees that, once the
base product has been bought, there will be some cost in finding out about other sources
that might provide those additional items at a cheaper price. This creates a potential
holdup problem similar to the one discussed in
Section 4.3.3.2, which the firm might
wish to fight through price advertising, especially if it faces competition.
Gabaix and Laibson, however, argue that some consumers might be “myopic” about
the pricing of add-ons and this may provide a rationale for firms to keep quiet about the
price of these additional complementary items. In the modeling of
Gabaix and Laibson
(2006)
, myopic consumers merely ignore the pricing of add-ons when deciding whether
to buy the base product, as long as the firm makes no mention of its add-on pricing.
Hence, myopic consumers may be viewed as being “unaware” of add-on prices unless
one of the competitors advertises them. Gabaix and Laibson show that, even with com-
petition, there may be an equilibrium where all firms shroud add-on prices.
65
In sum, unawareness is an argument for why advertisers might choose to hide some
relevant attributes. As such, it might provide some support for imposing some forced dis-
closure rules. I now present an alternative form of consumer naivety that could explain
why it might be profitable for advertisers to claim some seemingly irrelevant attributes.
Coarse thinking. In the simple setup outlined at the beginning of the subsection,
assume that x
2
reflects some category the product might belong to. Further assume that
the consumer’s true valuation is given by ^r ¼u + x
1
x
2
Δ, where Δ > 0 is the additional
utility for the consumer if x
1
¼x
2
¼1. The expression for r indicates that attribute x
1
affects the consumer’s welfare only if x
2
¼1.
Mullainathan et al. (2008) call the consumer a “coarse thinker” when she does not
know x
2
. In other words, she does not know which product category she is dealing with.
65
Their modeling also includes some ex ante investment that non-myopic consumers can undertake so as to
be able to avoid the high add-on prices they anticipate from the seller of the base product.
192
Handbook of Media Economics
In their application to “branding,” they propose as an example that some consumers
might confuse such brand labeling as “California Burgundy” for some cheap California
wine with the certified labeling “Burgundy” that applies to quality wines from the French
Burgundy region. In the four-product setting above, products (0,0) and (1,0) would be
two cheap California wines with comparable physical characteristics but the producer
might try to differentiate one of them (say (1,0)) by calling it California Burgundy. Prod-
ucts (0,1) and (1,1) would be two wines characterized by a quality difference so that only
product (1,1) can claim a certification label such as Burgundy. Formally, assume that the
producer of the generic cheap California wine, x
2
¼0, can choose a message
m 2 CB, DR
fg
, where CB stands for “California Burgundy” and DR stands for
“Delicious Red” (following the example in
Mullainathan et al., 2008). By contrast,
the wine producer in the alternative category, x
2
¼1, chooses a message m 2 B, T
fg
if it is (1,1) and m 2 T
fg
if it is (0,1), where B stands for Burgundy and T stands for Table.
The key point here is that only product (1,1) can use message B. The coarse thinker,
however, not knowing x
2
, confuses messages B or CB. She only realizes that x
1
¼1 when
reading such a message but is unsure whether this certifies a high quality or not.
If the consumer is not a coarse thinker, Mullainathan et al. call her a Bayesian, in which
case she knows which category of product she is dealing with, so she knows x
2
.Now
assume x
2
¼0. If a Bayesian consumer is endowed with beliefs about the product’s char-
acteristics x
1
, she updates her beliefs after observing the message m, depending on the equi-
librium messages she expects from products (0,0) and (0,1). Her willingness to pay,
however, does not depend on these beliefs. It is rmðÞ¼u for all m 2 CR, DR
fg
.Bycon-
trast, a coarse thinker believes that m ¼CB might certify quality in the event that x
2
¼1.
Assume now, as in Mullainathan et al., that the probability assigned by a coarse thinking
consumer to x
2
¼1 is exogenously fixed at some level strictly between zero and one.
66
A producer of cheap California wine will be able to charge a higher price to that consumer
if it announces m ¼CB, rather than m ¼DR. Mullainathan et al. show that, if the con-
sumer population comprises some Bayesians and some coarse thinkers, then it is optimal
for a monopoly producer of cheap California wine to propose two brands: Delicious
Red that is purchased by Bayesian consumers and California Burgundy that is purchased
by coarse thinkers where the latter brand is sold at a higher price than the former.
66
Mullainathan et al. (2008) also mention the possibility that coarse thinkers could be more sophisticated and
update their beliefs in view of the message they receive. For instance, if they believe that product (1,1) is
much more unlikely than product (0,1), then they would drastically revise the probability that x
2
¼1
downward after observing m ¼CB (assuming that m ¼CB is product (1,1)’s equilibrium strategy).
Lauga (2012) explores a similar idea in a setting where consumers are confused about the past experience
they have had with a product. When they remember a favorable experience, they cannot tell whether it
was their actual experience or whether they have been influenced by some advertising campaign. They
update their beliefs about the product quality accordingly, taking into account the firm’s equilibrium
advertising behavior.
193
Advertising in Markets
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.