Screening for Abdominal Aortic Aneurysm - US Preventive Services

16 Pages·2005·0.2 MB·English

informs
Vol.24,No.3,Summer2005,pp.461–476 ®
issn0732-2399(cid:1)eissn1526-548X(cid:1)05(cid:1)2403(cid:1)0461 doi10.1287/mksc.1050.0117
©2005INFORMS
The Targeting of Advertising
Ganesh Iyer
HaasSchoolofBusiness,UniversityofCalifornia,Berkeley,Berkeley,California94720-1900,
[email protected]
David Soberman
INSEAD,BoulevarddeConstance,Fontainebleau,France77305,[email protected]
J. Miguel Villas-Boas
HaasSchoolofBusiness,UniversityofCalifornia,Berkeley,Berkeley,California94720-1900,
[email protected]
A
n important question that firms face in advertising is developing effective media strategy. Major improve-
ments in the quality of consumer information and the growth of targeted media vehicles allow firms to
precisely target advertising to consumer segments within a market. This paper examines advertising strategy
when competing firms can target advertising to different groups of consumers within a market. With targeted
advertising, we find that firms advertise more to consumers who have a strong preference for their product
thantocomparisonshopperswhocanbeattractedtothecompetition.Advertisinglesstocomparisonshoppers
can be seen as a way for firms to endogenously increase differentiation in the market. In addition, targeting
allows the firm to eliminate “wasted” advertising to consumers whose preferences do not match a product’s
attributes. As a result, the targeting of advertising increases equilibrium profits. The model demonstrates how
advertisingstrategiesareaffectedbyfirmsbeingabletotargetpricing.Targetadvertisingleadstohigherprofits,
regardless of whether or not the firms have the ability to set targeted prices, and the targeting of advertising
canbemorevaluableforfirmsinacompetitiveenvironmentthantheabilitytotargetpricing.
Keywords: mediaprecision;advertising;targeting;pricediscrimination
History: This paper was received May 7, 2003, and was with the authors 9 months for 2 revisions; processed
byChakravarthiNarasimhan.
1. Introduction is the fragmentation of existing media (broadcast TV,
Advertising is one of the most important decisions a for example) and a multitude of new advertising
marketer makes, and media purchasing is the largest media (the Internet, satellite shopping channels, and
elementofadvertisingspending.Ensuringthatmedia infomercials). Sophisticated media buying now pro-
is bought effectively and not directed toward the videsfirmswiththeabilitytotargetspecificsegments
“wrongpeople”hasalwaysbeenachallengeformar- within a market (see “Infinite Variety,” The Economist,
keters.1 Traditionally, the objective in media planning November 19, 1998). Because firms need to ensure
was to minimize wasted advertising by reducing the that marketing spending has impact, it is not surpris-
quantity of advertising sent to consumers who are ing that they are increasingly active in the use of tar-
not active in the category. However, firms can now geted advertising.
do much better than reduce advertising to nonusers. In media planning, firm objectives are often to
They have both the know-how and the means to target advertising to specific consumer groups. For
target advertising to segments of consumers within example, consider the U.S. light beer market in
a market. This ability comes from two key changes
whichMillerLiteandCoorsLightaremajorcompeti-
in the marketing environment. Today, firms have
tors. The light beer market is comprised of distinct
much better information on consumers, their prefer-
demographic groups that vary in their consump-
ences, and their media habits (see “Star Turn,” The
tion profile. Miller Lite, the “diet beer,” has tradi-
Economist, March 9, 2000). This is the result of signif-
tionally been directed to mature male beer drinkers
icant improvements in the ability to collect and pro-
in their mid- to late 30s who are concerned about
cess consumer-level information. The second change
their waistline. In contrast, Coors Light has been
more popular among young and relatively new beer
1This is a classic concern and goes back to at least John Wana-
drinkers (men and women in their early 20s). But
maker’s (a 19th-century department store owner) comment “Half
a substantial proportion of light beer consumption
the money I spend on advertising is wasted and the trouble is I
don’tknowwhichhalf.” resides in the intermediate segment comprised of
461
Iyer,Soberman,andVillas-Boas: TheTargetingofAdvertising
462 MarketingScience24(3),pp.461–476,©2005INFORMS
young adults in their late 20s to early 30s. These con- firm has a group of consumers who have a strong
sumers are more uncommitted in their brand pref- preferenceforitsproduct,i.e.,theyonlyconsiderbuy-
erence and are indifferent between the two brands.2 ing from that firm (up to a reservation price). There
An important question for firms is the decision about is also a group of consumers who compare the prices
how to allocate media budgets between segments atbothfirmsandbuyatthelowestprice.Advertising
where they have a strong franchise and segments of is costly and the cost of informing a group of con-
uncommitted consumers who choose between com- sumers is directly proportional to its size. The target-
peting brands. On the one hand, it can be argued ingofadvertisingimpliesthatfirmscandesignmedia
that concentrating advertising on consumers who are vehiclestotargetadvertisingmessagestospecificseg-
strongly predisposed to buy a firm’s product should ments in the market. A firm that cannot target adver-
be advantageous (for Miller, this would mean target- tising advertises uniformly to the entire market.
ing advertising effort on mature male beer drinkers), We show that when firms have the ability to target
given that these consumers are more disposed to buy advertising,eachfirmadvertisesmoretothesegment
and are willing to pay higher prices. On the other
thathasastrongpreferenceforitsproductthantothe
hand, competition is highest for consumers who do
segment of consumers who comparison shop. When
not have a strong preference for one of the compet-
comparison shoppers are informed about both prod-
ingproducts(inthelightbeerexample,thiswouldbe
ucts, they perceive no differentiation between them
theintermediatesegment).Withoutastrongadvertis-
and this leads to intense price competition. Firms
ing effort, these consumers may be lost to the com-
respondbyreducingadvertisingtocomparisonshop-
petition. Will the attractiveness of an intermediate
pers. Consequently, there are times when compar-
segment with weak preferences lead to aggressive
ison shoppers are informed about only one firm’s
advertising by both firms or will firms limit competi-
product. In this situation, that firm has monopoly
tion for these consumers with lower advertising? We
power over the comparison shopper segment. Indi-
considerthisquestionwithamodelofadifferentiated
rectly, this reduces the intensity of price competition.
marketwithtwocompetitivefirms,eachofwhichsell
Thus, advertising less to comparison shoppers is an
a single product. We examine how the ability to tar-
indirect way of creating market differentiation. The
getadvertisingtospecificsegmentsaffectsadvertising
targeting of advertising also provides a direct bene-
and pricing decisions.
fit of eliminating “wasted” advertising to consumers
The following questions are analyzed in this paper.
who prefer to buy the competing product. For these
When firms have the ability to choose different levels
reasons, the ability to target advertising increases the
(media weights) of advertising to different consumer
equilibrium profits of firms.
segments, how will they choose media weights?
When firms move from a strategy of uniform
Shouldafirmadvertisemoretoconsumerswhohave
advertising to targeted advertising, the total amount
a strong preference for its product or to consumers
spent on advertising can either increase or decrease.
who are more likely to comparison shop amongst
When advertising is expensive, the inability to target
alternatives? How are equilibrium pricing and profits
advertising leads firms to choose low levels of adver-
inamarketaffectedbythefirms’abilitytotargettheir
tising.Whilethismeanslesswastedadvertising,firms
advertising?Wealsoexaminehowtheabilitytotarget
advertising affects the level of advertising spending are also not able to realize the demand potential in
by firms. Recent advances in consumer information the market because few consumers are informed. In
and database technologies also mean that firms can thiscase,targetinghelpsfirmsrealizehigherdemand
price discriminate and offer different prices to differ- and firms increase their advertising expenditures. In
entgroupsofconsumers.Wethenaskhowtheability contrast,whenadvertisingisinexpensive,thenafirm
to target advertising interacts with targeted pricing. chooses high advertising levels with uniform adver-
We consider a model where a firm’s advertising tising. This implies that the extent of wastage is sig-
provides complete information about its products nificant and the ability to target advertising leads to
to potential consumers, but advertising may be too lower advertising expenditures.
costly for all competitors to always advertise. Each We also analyze how targeted advertising interacts
with targeted pricing. Our analysis shows that in a
competitive environment, the ability to target adver-
2See the discussion “Competition: A Whole New Ball Game in
tising is more important for profits than the abil-
Beer,”Fortune,September19,1994,p.79,andLee,Thomas,“Miller’s
TimeMayBeRunningOut:Brewer’sSalesRemainFlatAmidTalk ity to target pricing. When firms have the ability to
ThatPhilipMorrisWillSelltoForeignFirm,”St.LouisPost-Dispatch, choosedifferentadvertisinglevelsfordifferentgroups
March10,2002,p.E1.Roy(2000)couldpotentiallybeseenasajus-
of consumers, it leads to higher profits independent
tificationforwhythehighpreferencesegmentsforMillerLiteand
ofwhetherornotfirmshavetheabilitytosettargeted
CoorsLightaredifferent.Thisdifferencecouldalsobeseenasjust
abasicproductdifferentiationdecision. prices. In contrast, the ability to target prices creates
Iyer,Soberman,andVillas-Boas: TheTargetingofAdvertising
MarketingScience24(3),pp.461–476,©2005INFORMS 463
increased competition for comparison shoppers and after observing the competitor’s advertising. In con-
no improvement in equilibrium profits. trast, this paper considers a differentiated market
We examine the market outcomes when firms (that reduces to a homogeneous good if h = 0, as
investtoobtaintheabilitytotargetadvertising.Given described later) where firms can target consumers
the increased profits associated with targeting adver- according to their preferences and set prices without
tising, both firms will acquire targeting capability if knowing the competitor’s advertising.4 In addition to
the fixed cost to obtain it is sufficiently low. Similarly, characterizing the targeting equilibrium, we consider
both firms choose not to target advertising when the the question in both uniform pricing and targeted
fixedcostisveryhigh.Butinterestingly,whenthecost pricing contexts. Roy (2000) assumes that consumers
of targeting is in an intermediate range, asymmetric have unique addresses (which are unrelated to con-
firms arise endogenously. While one firm invests to sumer preferences because consumers have homoge-
obtain targeting capability, the other chooses not to nous preferences) and argues that firms choose to
invest. Differences in the ability to target advertising advertise to different individual consumers. This idea
are also a way to reduce competition for comparison might be seen as related to the result in this paper
shoppers. Finally, we examine the case of imperfect that firms advertise less to the comparison shopper
targeting in which advertising by a firm to a specific segment than to the segment of consumers who have
segmentleakstoothersegmentsandshowsthatleak- a high preference for the firm. However, the result in
age leads to lower equilibrium firm profits. Roy (2000) depends on the assumptions that pricing
Several papers have looked at the impact of adver- decisions are made after observing the competitor’s
tising on product information and pricing. In par- advertising and that firms are able to target adver-
ticular, Butters (1977) proposes a message-sending tising to individual consumers. In addition, Roy’s
model,whereadvertisingprovidesinformationabout (2000) model generates an infinite number of equilib-
the existence of products (and their characteristics) ria, all of which depend on significant coordination
andthehigherthelevelofadvertisingafirmchooses, between firms (firms target consumers with no over-
the more likely a representative consumer is exposed lap). Stegeman (1991) considers the welfare implica-
to it. Grossman and Shapiro (1984), Stahl (1994), and tions of informative advertising in a model with a
Soberman (2004) extend this model to markets with large number of competitors selling a homogeneous
horizontal differentiation and analyze the impact of good. Consumers may have different valuations for
informative advertising on market competition and the good and firms can target advertising to con-
the provision of variety. All of these papers assume sumers with different valuations. However, because
that advertising is uniform throughout the market.3 the model pertains to a homogeneous good, there is
Esteban et al. (2001) allow different levels of adver- no possibility of consumers having different prefer-
tising to be directed at different segments (or loca- ences for the competing products in the market.
tions) within the market. That paper considers a In this literature, targeted marketing activity has
monopolistic firm that faces a market where cus- been analyzed in context of other marketing ele-
tomers have heterogeneous reservation prices, and ments. Price discrimination based on customer recog-
argues that the monopolist will direct heavier adver- nition has been examined by Villas-Boas (1999, 2004)
tising weights to the consumers who are willing to and Fudenberg and Tirole (2000). Previous research
pay more for the product, and that the overall level has also examined location-specific pricing (Thisse
of advertising falls with targeting. Roy (2000) consid- and Vives 1988, Shaffer and Zhang 1995), the role
ers the competition for a homogeneous good where of imperfect customer addressability (Chen and Iyer
firms can target consumers, and compete on prices
4Several of the main results also generalize to the symmetric
3Rajivetal.(2002)modelpricepromotionaladvertisingstrategies equilibrium in the case in which firms set prices after observ-
whenfirmsareasymmetricalongaqualitydimension.Villas-Boas ing the competitor’s advertising, given that in such an equilib-
(1993) considers dynamic competitive effects with advertising rium, the advertising strategy is in mixed strategies. Analyzing
pulsing (see also Villas-Boas 1992 for other applications of the the case in which firms set prices after observing the competi-
same framework and Dubé et al. 2004 for an empirical analy- tor’sadvertisingmayperhapsbemoreappropriateforthecaseof
sis). Vakratsas et al. (2004) investigates the shape of the advertis- large visible advertising campaigns, where a firm can better infer
ing response functions that could justify pulsing. Consumers can the competitor’s advertising spending ex ante before the pricing
alsopotentiallyfindinformationaboutexistingproductattributes, decision.However,itistypicalinmanycasesforfirmstonothave
includingprice,throughsearch(Kuksov2004)andthroughtheuse goodknowledgeoftheircompetitor’sadvertisingplanorbudgets.
ofcomparativeshoppingmechanisms(IyerandPazgal2003).Bass Indeed,firmsarecautiousaboutnotlettingtheircompetitorslearn
et al. (2004) consider dynamic competition with both generic and abouttheiradvertisingplans.Thecaseofpricingwithoutobserving
brandadvertising.ShafferandZettelmeyer(2004)consideradver- thecompetitor’sadvertisingisalsorelevantforthecaseoflessvis-
tising in a distribution channel. Baye and Morgan (2004) consider ibleormoretargetedadvertisingordirectmailings.Butters(1977),
theeffectofuniformadvertisingonthecreationofbrandawareness Grossman and Shapiro (1984), and Stahl (1994) consider pricing
andpricecompetitioninonlinemarkets. withoutobservingthecompetitor’sadvertising.
Iyer,Soberman,andVillas-Boas: TheTargetingofAdvertising
464 MarketingScience24(3),pp.461–476,©2005INFORMS
2002), the imperfect targeting of prices to customers toward these consumers provides them with infor-
(Chen et al. 2001), and the impact of targeted prod- mation on the product and its characteristics. For
uct modifications (Iyer and Soberman 2000). Some of example: Does the product possess the attributes that
the nature of the effects in this literature is discussed the consumer requires to consider it for purchase?
in§3.3.Thispapercontributestothisresearchbyana- Of course, this simply implies that advertising facil-
lyzingtheimpactoftargetedadvertisinginacompet- itates consideration of the product by the consumer.
itive setting. The rest of this paper runs as follows. If the product does not fit a consumer’s needs or if
Section 2 describes the basic model. In §3, we present the price is too high, she will not buy. Note that we
the main results of this paper and some anecdotal areassumingthatadvertisingisnecessaryforthecon-
information from retail markets to support the analy- sumer to be in the market and to consider the prod-
sis. Finally, §4 presents concluding remarks. uct. Later in §3.5, we extend the model to consider
the case in which some proportion of the consumers
are informed and would consider buying the product
2. The Model
even without receiving advertising.
We develop a model of a market with two firms,
The characterization of advertising is consistent
i = 1(cid:3)2. Each firm produces its product at a con-
with behavioral research that documents how adver-
stant marginal cost of production, which is assumed
tising makes a product and its characteristics salient
to be zero without loss of generality.5 We start by
intheconsumers’memory.This,inturn,enhancesthe
describing the consumer market. The market is com-
likelihood that consumers consider the product if its
prised of a unit mass of consumers. Each consumer
characteristics do indeed match consumer tastes (see
has a demand of at most one unit of the product.
MitraandLynch1995).6 Fornewproducts,awareness
Consumers have a common reservation price r for
isclearlythefirststageincreatingdemandforaprod-
the product. Assume that each firm has a segment
uct. Consumers also use advertising for new prod-
of consumers who have high preference for its prod-
uctstoobtaininformationaboutkeyproductfeatures.
uct in the sense that they consider buying only from
Theformulationisalsoconsistentwiththeroleadver-
that firm as long as the price at the firm is below
tising plays in mature product categories. Keeping a
the reservation price r. The proportion of these con-
product “top of mind” and priming the consumer to
sumers per firm is given by h. The remaining con-
consider it is critical in established categories such as
sumers are comparison shoppers who are indifferent
beer and soft drinks. For example, in the soft drinks
betweenthefirmsandprefertobuytheproductwith
market, one might argue that the product features
the lowest price (as long as this price is below the
of Coke and Pepsi are known to most consumers.
reservation price). The size of this segment s is given
Yet, these brands spend a significant amount of their
by s = 1 − 2h. The role of this segment is to rep-
budgetonreminderadvertisingaimedatkeepingthe
resent consumers who have less intense preference
brand top of mind. In our model, this simply means
for either brand. Note that h represents the extent
that advertising increases the consideration of the
of ex ante market differentiation, with higher val-
product by consumers. Advertising could also have
ues representing greater differentiation between the
other roles not considered here such as changing the
firms, because more consumers would have different consumervaluationforproducts,possiblyindifferent
preferences across firms. When h=0, all consumers ways across consumer types.
comparisonshopbetweenthetwofirmsandthecom- We assume that the cost to advertise to the entire
petition between the firms reduces to Bertrand price marketisA.Whenadvertisingcanbetargetedtopar-
competition. ticular segments in the market, we assume that the
Consumers are endowed with preferences over costtoadvertisetoeachsegmentislinearlyrelatedto
product attributes, but without advertising, do not its size.7 Therefore, if a firm is able to target advertis-
know which products exist or their characteristics ing,thecostsareAhforthehighpreferenceconsumer
(they do not search for information about products).
The role of advertising about a product is to con-
6AnandandShachar(2001)alsoshowthiseffectwithactualdata.
vey information that the product exists and its prod-
Furthermore, advertising can be seen as creating heterogeneity in
uct attributes (which might also include the price), the set of products that consumers consider depending upon the
so that an originally uninformed consumer can eval- numberoffirmsfromwhomtheconsumersreceiveadvertising.As
uate its degree of preference for the product and showninMehtaetal.(2003),therecanbesubstantialheterogeneity
intheconsiderationsetsofconsumersinamarket.
decide whether to buy it or not. Advertising directed
7Some research has discussed the possibility of the response
to advertising being S-shaped or nonlinear. See, for example,
5ThemodelandresultscanbeextendedtoamarketwithN firms, ThompsonandTeng(1984),EastlackandRao(1986),Mahajanand
whereeachfirmhasdemandfromahighpreferencesegmentand Muller (1986), Rao (1986), Sasieni (1989). The advertising technol-
a comparison shopping segment that is common, along the same ogy in the model and its results can be both consistent with the
linesasinVarian(1980). caseofextremeS-shapeandwiththecaseoflinearcosts.
Iyer,Soberman,andVillas-Boas: TheTargetingofAdvertising
MarketingScience24(3),pp.461–476,©2005INFORMS 465
segment and As for the comparison shopping seg- Lemma 1. Whenhr>A,firmsadvertiseinequilibrium
ment. There is some discussion that targeted media with probability one. When hr ≤A, then the equilibrium
vehicles are more costly on a per consumer basis will involve firms using mixed advertising strategies.
(Esteban et al. 2001). Incorporating this effect into the Firms always advertise if the guaranteed profits
model would just make the targeting of advertising from the high preference segment are large enough
less profitable without affecting the main messages to cover the cost of advertising. This happens when
of this paper. This paper considers the fixed costs of the extent of differentiation (cid:11)h(cid:12) or the reservation
obtainingtargetingcapabilityin§3.4.Notethatafirm pricearesufficientlyhigh.Forthecaseinwhichfirms
does not have an incentive to target advertising to advertise with probability one, simple calculations
thehsegmentofitscompetitorsincethoseconsumers will verify that (cid:17)F(cid:11)p(cid:12)/(cid:17)h<0. Thus, the average price
will not buy its product. We consider advertising that charged by a firm increases with the extent of differ-
informs all of a given segment or none of it.8 entiation between the firms (i.e., a larger h).
The interesting case is when firms do not find it
optimal to advertise with probability one. In other
3. EquilibriumAnalysisofAdvertising
words, in less differentiated markets, if the reser-
andPriceCompetition vation price for the product is small compared to
We start the analysis with the case when firms do the cost of advertising, firms use mixed strategies in
not have the ability to target advertising or pricing advertising. To focus on some basic effects of adver-
tising, our model assumes that each period is inde-
to specific segments of the market. It provides a base
pendent and that there are no carry-over effects for
case that we use to interpret the impact of targeting.
consumers.10 We can interpret the probability with
which firms advertise as the intensity of advertising
3.1. Uniform Advertising and Price Competition
withinaplanningperiod.Basically,throughitsadver-
Considerthatinequilibriumthefirmsadvertise.With
tisingfrequency,afirmdeterminesthelikelihoodthat
uniformadvertising,firmscanreachtheentiremarket
a consumer becomes informed (or aware of the prod-
for a cost A. The price equilibrium will then be in
uct) during the period. The more intense the adver-
mixedstrategies.Thereasoningisasfollows:Suppose
tising is, the higher the likelihood that the consumers
that one firm; say, Firm 2, chooses a price p that is
2 become informed.
not too low; then Firm 1 can undercut p to attract
2 For this case, there is a unique symmetric equilib-
all the comparison shoppers (these consumers make
rium. To derive the equilibrium when firms employ
a comparison and will choose Firm 1 because its p
1 mixed strategies in advertising, define (cid:20) as the prob-
is slightly lower). Otherwise, Firm 1 will set prices at
ability of advertising by a firm. From the property
the reservation price to maximize the profit from its
of a mixed strategy equilibrium, the profits between
informed h consumers. In either possibility, Firm 2’s
advertising and not advertising are equal, which
best response is then not to charge p , and we end up implies the following equilibrium condition, hp +
2
withapricemixedstrategyequilibrium(Varian1980). (cid:11)1−(cid:20)(cid:12)sp+(cid:20)sp(cid:11)1−F(cid:11)p(cid:12)(cid:12)−A=0. From this, if A>
Denote the c.d.f. (cumulative distribution func- r(cid:11)1−h(cid:12), then the firms will not advertise; i.e., adver-
tion) of the mixed strategy price distribution without tising is not feasible. When A(cid:21)(cid:11)rh(cid:3)r(cid:11)1−h(cid:12)(cid:12), adver-
advertising to be F(cid:11)p(cid:12). In a symmetric equilibrium tising strategies are mixed: advertising costs are low
i
(cid:11)F(cid:11)p(cid:12)=F(cid:11)p(cid:12)(cid:12), the profit of a firm when charging a enough such that advertising is efficient (with proba-
i
price p in the mixed strategy profile will be given by bility less than one), but not so low that firms choose
(cid:13)(cid:11)p(cid:12)=hp+sp(cid:14)1−F(cid:11)p(cid:12)(cid:15)−A. Using standard analysis to advertise with probability one. The equilibrium
(e.g., Varian 1980, Narasimhan 1988, Baye et al. 1992), solution leads to Proposition 1.
the equilibrium profit is the guaranteed profit that a Proposition 1. When hr ≤ A, and with uniform
firmcanrealizebychargingthereservationpriceand advertising, the equilibrium profits are zero and the equi-
selling only to its h segment, (cid:13)(cid:11)r(cid:12)=hr −A.9 If it is librium probability with which firms advertise is (cid:20)∗ =
greater than the profit associated with not advertis- 1−(cid:11)A−hr(cid:12)/sr. In addition, firms employ mixed pricing
ing, i.e., zero, then firms always advertise in equilib- strategies with c.d.f.
rium.Equilibriumadvertisingisthuscharacterizedby (cid:1) (cid:2) (cid:1) (cid:2)
r−p A A
Lemma 1. F∗(cid:11)p(cid:12)=1− for p∈ (cid:3)r (cid:24)
p (cid:11)1−h(cid:12)r−A 1−h
8Iyeretal.(2002)extendsthemodeltoacontinuousrepresentation
10It would be interesting to also study the impact of carry-over
ofadvertisingwithcoststhatareconvexintheproportionofcon-
effects for consumers in this context of targeted advertising. Note
sumers reached within a segment, and shows that the results are
that there is some evidence that in many low involvement cat-
similarwhenfirmscanadvertisetoanyproportionofasegment.
egories (like cookies, potato chips, and ready-to-eat cereal), the
9See Zhao (2004) for a descriptive analysis of price dispersion in maindriverofbrandchoiceistop-of-mindawareness(Dicksonand
thegrocerychannelthatisconsistentwithsuchamodel. Sawyer1990).
Iyer,Soberman,andVillas-Boas: TheTargetingofAdvertising
466 MarketingScience24(3),pp.461–476,©2005INFORMS
The equilibrium probability (or frequency) of beingabletodirectadvertisingtothehighpreference
advertisingdecreaseswiththecostofadvertisingand and to the comparison shopper segments separately.
increases with the reservation price. It is also easy Given our assumption that the cost of advertising is
to see that (cid:17)F(cid:11)p(cid:12)/(cid:17)h < 0 and (cid:17)F(cid:11)p(cid:12)/(cid:17)A < 0. Thus, proportionaltotheconsumersreached,thecostoftar-
the expected price increases with both market differ- geting the h segment of a firm is hA, while the cost
entiation (the size of the h segment) and advertis- of targeting the comparison shopping segment is sA.
ing costs. The relationship between (cid:20)∗ and market Because firms can choose to advertise to the high
differentiation is more interesting: the frequency of preferenceconsumersonlyandchargethereservation
advertising decreases with the size of the comparison price, the guaranteed profits from the h segment are
shopping segment (i.e., lower differentiation) when h(cid:11)r−A(cid:12). Thus firms always advertise to their h con-
A<r/2. However, advertising frequency increases in sumers as long as r >A. For the rest of the analy-
the size of the comparison shopping segment when sis, we assume that this holds.12 Note that with the
A>r/2. This reversal can be explained by two effects ability to target advertising, firms do not advertise to
thathigheradvertisingfrequencieshaveonthenature the other firms’ h consumers as these consumers will
of competition. First, higher advertising frequencies not buy. Next, consider advertising to the compari-
increase the fraction of comparison shoppers that son shopping segment. In general, advertising to this
are informed about both firms (cid:11)(cid:20)2(cid:12). This raises the segment involves mixed advertising strategies. Sup-
incentivetopriceaggressivelybecausefullyinformed pose that both firms advertise with probability one.
comparison shoppers compare prices and buy from Then, if advertising is costly, either of the firms has
the firm with the lowest price. The second effect an incentive to deviate by marginally reducing the
of increased advertising frequency is that more of frequency of advertising. While the firm’s expected
the total market is able to buy each firm’s product, demandfromthecomparisonshoppingsegmentgoes
(cid:17)(cid:11)h+s(cid:12)/(cid:17)s>0. This provides a demand benefit to down by a small amount, all profits from this seg-
each firm. ment are dissipated when it is fully informed 100%
When costs of advertising are low (cid:11)A<r/2(cid:12), firms of the time. As a result, a firm will save on the cost
advertise aggressively. In this case, a reduction in of advertising by reducing its frequency of advertis-
market differentiation (i.e., decreases in h, the size of ing. Writing the probability of advertising to com-
the loyal segment) has two effects. First, it increases parison shoppers as (cid:25), the profit function for a firm
the fraction of each firm’s demand that is com- when advertising to s is (cid:13)(cid:11)p(cid:12) = hp + (cid:11)1 − (cid:25)(cid:12)sp +
peted for (each firm has more comparison shoppers (cid:25)sp(cid:14)1−F(cid:11)p(cid:12)(cid:15)−A(cid:11)h+s(cid:12).Proposition2summarizesthe
relative to high preference consumers). Second, it equilibrium with targeted advertising.
increases total demand available to each firm (h+s
Proposition 2. Whenadvertisingcanbetargeted,and
increases). However, reduced profits from the first
r>A, the equilibrium profit is h(cid:11)r−A(cid:12) and firms adver-
effect are larger than the positive effect of a higher
tise to their h consumers with probability one and to com-
potential market. As a result, the optimal advertising
parison shoppers with a probability of (cid:25)∗ =1−A/r. In
leveldrops.Here,firmsmanagenoncooperativelythe
addition, firms employ mixed strategy pricing with c.d.f.
degree of competition in the market by reducing the
proportion of fully informed comparative shoppers. (cid:1) (cid:2)
rh+As r−p hr+As
The inverse applies when advertising costs are suf- F∗(cid:11)p(cid:12)=1− for p∈ (cid:3)r (cid:24)
s(cid:11)r−A(cid:12) p h+s
ficiently high (cid:11)A>r/2(cid:12). In this case, the benefit of
increased demand outweighs the competitive effect. First, note that the probability of advertising to
As a result, the optimal level of advertising is higher the comparison shoppers (cid:25)∗ is strictly less than one.
when the size of the comparison shopping segment Therefore, when advertising can be targeted, firms
increases. advertise more to their respective high preference
segments than to comparison shoppers. By targeting
3.2. Competition with Targeted Advertising
advertising to consumers who have a strong prefer-
We now analyze the main issue of this paper per-
ence for its product, a firm increases the consumer
taining to the ability of firms to target advertising
surplusitextractsfromthemarket.Eitherfirmhasan
to particular segments of the market. The advertis-
incentive to advertise to comparison shoppers with
ing targeting technology being considered implies
a probability less than one. The effect of advertis-
more precise media vehicles that allow firms to tar-
ing with a probability less than one is to reduce
getadvertisingtospecificsegmentsofthemarketand
better information on consumer preferences across
segments.11 In the model, this translates to the firms 12This simply means that the reservation value of all consumers
who require advertising to become informed is greater than the
cost of advertising. Otherwise, firms will not advertise, implying
11Roy(2000)canbeseenaslookingonlyatthefirsteffect. thedegeneratecasewherefirmsfindadvertisinginfeasible.
Iyer,Soberman,andVillas-Boas: TheTargetingofAdvertising
MarketingScience24(3),pp.461–476,©2005INFORMS 467
competition for comparison shoppers. In fact, the products. This leads to an overall increase in adver-
competing firm enjoys monopoly power over these tising expenditure.
consumers when it is advertising but the focal firm is It is useful to compare the above results to the
not. This has the indirect effect of reducing the inten- monopoly analysis of Esteban et al. (2001) who
sity of price competition (which allows higher profits findthattargetingdecreasesadvertisingexpenditures.
tobeearnedfromthehighpreferencesegment).Thus, This idea is similar to the first part of Proposition 3
advertisingwithprobabilitylessthanonehelpsafirm in the sense that with targeted advertising firms can
to endogenously create differentiation in the compet- avoid advertising to consumers with lower willing-
itivepartofthemarket.Furthermore,thedirecteffect ness to pay for the product (given their other alter-
oftargetedadvertisingistoeliminatewastagecaused natives). However, our analysis shows that there are
by advertising that falls on the competitor’s h seg- indeed conditions under which the inverse can hap-
ment. Consequently, as Proposition 2 shows, the abil- penandadvertisingexpendituresincreasewhenfirms
ity to target advertising to specific segments leads to have the ability to target advertising.
anincreaseinprofitoverthecaseofuniformadvertis- Targeted advertising also increases the average
ing.Notethattheadvertisingintensitytothecompar- prices that firms charge. With targeted advertising,
ison shopping segment increases with the reservation a firm always advertises to its h segment, while
price because there is more surplus to extract from advertising with probability (cid:25) to comparison shop-
consumers who are reached by advertising. Targeted pers.Consequently,thereisreducedpricecompetition
advertising also has interesting effects on advertising betweenfirms,leadingtohigheraveragepricesbeing
spending and pricing. charged in equilibrium.
Proposition 3. Compared to the case of uniform 3.3. Comparing Targeted Prices and
advertising, total advertising expenditures are lower with Targeted Advertising
targeted advertising when A < r/2 and higher when Until now, we have focused on markets where firms
A>r/2. had the ability to target advertising but could only
compete with uniform pricing strategies. This is the
Advertising expenditures decrease with targeted
mainstream case of most product markets, where
advertising when A<r/2, i.e., when advertising is
firms target advertising to different consumer seg-
relativelyinexpensive.However,wealsofindthattar-
ments through the media plan and products are
geting can lead to an increase in advertising expen-
sold to consumers through traditional retail channels.
ditures when A > r/2. This phenomenon obtains
However, with the growth of the Internet and bet-
because of the competitive context of our model and
ter point-of-sale technologies, firms increasingly have
the resulting interaction of advertising and price.
theabilitytopricediscriminateandtargetspecialized
The analysis highlights two effects of targeting
prices to different segments.
advertising. The first is reduced wastage and the sec- In this section, we examine the effect of targeted
ond is the creation of a more effective marketing pricing and ask how it interacts with the ability of
instrument. In particular, when a firm cannot target firms to target advertising. A natural way to begin
itsadvertising,itcannoteliminatewastedadvertising this investigation is to ask what happens if firms
to the h customers of the competitor. When adver- could target price, but were restricted to uniform
tising is inexpensive, a firm will choose high levels advertising.Thiscaseallowsustoteaseouttheeffects
(i.e., frequency) of advertising, all else being equal. of advertising targeting relative to that of pricing.
Therefore, without the ability to target, inexpensive The case of uniform advertising and targeted pric-
advertising means that the extent of wastage is sig- ing applies to situations where the media options to
nificant. The ability to target advertising allows the reach a target population are limited, yet consumers
firm to eliminate this wastage leading to a decrease areeasytoclassifyatthetimeofpurchase. Forexam-
in the overall level of expenditure. In contrast, when ple, a major problem for firms in developing coun-
advertising is expensive, firms choose low levels of tries is finding media vehicles that deliver a targeted
advertising under uniform advertising. Advertising audience.Ontheotherhand,variousformsofpricing
is an ineffective marketing instrument because it is (volume discounts, bundling, coupons) often allow
both expensive and much of it goes to the wrong these firms to tailor prices based on customer type.
potential consumers. As a result, many customers Inthissituation,theabilitytotargetpricesisstronger
who would be willing to pay the equilibrium price than the ability to target advertising.
are uninformed, and thus do not buy. In this case, Recall that when a firm advertises without target-
the ability to target advertising allows firms to real- ing, the profit from charging the reservation price
ize higher demand by increasing advertising to the is hr −A. Therefore, following Lemma 1, if hr >A,
part of the market that has interest in their respective then firms advertise with probability one. If hr <A,
Iyer,Soberman,andVillas-Boas: TheTargetingofAdvertising
468 MarketingScience24(3),pp.461–476,©2005INFORMS
thenfirmsemploymixedadvertisingstrategies.Simi- unaffected by gaining the ability to target prices
lar to §3.1, we solve for a symmetric equilibrium and regardless of whether firms can target advertising
denote (cid:26) as the probability of advertising for this or not. The reason is that the attractiveness of the
u
uniform advertising case. We then write the profit comparison-shopping segment fully determines the
of a firm when it advertises as hr + (cid:11)1 − (cid:26) (cid:12)sp + incentive to advertise to it and it is a function of two
u
(cid:26) sp(cid:11)1−F(cid:11)p(cid:12)(cid:12)−A. The equilibrium profit in this case things, the size of the segment and the reservation
u
is zero, while the equilibrium probability of advertis- price comparison shoppers are willing to pay. This
ing is (cid:26)∗=1−(cid:11)A−hr(cid:12)/(cid:11)sr(cid:12). Comparing this with the incentive is independent of whether firms can target
u
case of uniform advertising and pricing, we see that pricing or not. The difference in the two worlds (uni-
the incentive to advertise (uniformly) is unaffected in form versus targeted pricing) is that with uniform
this model by the ability to set targeted prices (the pricing, the incentive to cut price is reduced because
equilibrium advertising is identical to the case of uni- profitislostonhighpreferenceconsumerswhenprice
form pricing derived in §3.1). The equilibrium prof- is lowered. Of course, firms will only reduce price to
its also do not change from the uniform price case. the point where the profits they earn by capturing
This is because while targeted pricing allows firms to increaseddemandisatleastashighastheguaranteed
increase the price charged to the high preference con- profit.
sumers (to the reservation price r), it also increases In contrast, in the world of targeted pricing and
competition for the comparison shoppers relative to targeted advertising, competition in the comparison
the base case. In this model, these effects cancel out shopping segment is decoupled from the high pref-
andinequilibrium,firmsdonotbenefitfromtargeted erence segments. While the incentive to advertise is
pricing versus the base case. With targeted pricing, unchanged by targeted pricing, the incentive to price
thecomparisonshoppersarebetteroff,whilethehigh aggressively is higher. As a result, the average price
preferencesegmentisworseoffandpaysthereserva- for comparison shoppers is lower in the targeted
tion price. pricingworld.13 Ofcourse,theselowerpricesareper-
We now consider the case where firms can tar- fectly offset by higher prices that are charged to high
get both advertising and pricing. This case is directly preference consumers (they always pay r).
applicable to direct marketers who offer tailored Similar to §3.2 where advertising can be targeted
prices to consumers based on the increased availabil- but prices are uniform, firms advertise to their h seg-
ity of individual-level consumer information. Ana- ment with probability one and the probability of
lyzing this problem helps us to understand how the advertising to comparison shoppers is identical. The
ability to target advertising interacts with a firm’s contrasting effects of targeting for both pricing and
ability to target pricing. When firms can target both advertising are summarized in Table 1. The benefit
price and advertising, each firm can guarantee itself of targeted pricing is the ability to charge reservation
a profit of h(cid:11)r − A(cid:12). This is because the firm can
prices and extract surplus from the high preference
choose to send advertising only to their h segment
segment. However, targeted pricing also increases
andchargethereservationprice.Similarto§3.2,firms
price competition for comparison shoppers because
do not advertise to the h consumers of the com-
a firm can reduce price to these consumers without
petitor and employ a mixed advertising strategy to
reducingthepricetoitshsegment.Theresultsshown
the comparison shopping segment. We can write the
in Table 1 demonstrate that these effects cancel out.
following equilibrium condition for the comparison
In this model, the profits of firms are unaffected by
shopping segment (where (cid:26) is the probability of
t having the ability to set targeted prices regardless of
advertisingtocomparisonshoppersinthiscaseoftar-
whether advertising is uniform or targeted (see also
geted advertising): (cid:11)1−(cid:26)(cid:12)sp+(cid:26)sp(cid:11)1−F(cid:11)p(cid:12)(cid:12)−As=0.
t t Winter 1997, Corts 1998).
Proposition 4 characterizes the equilibrium.
3.4. Incentives to Invest in Targeting Capability
Proposition 4. When advertising and pricing can be
targeted,theequilibriumprofitish(cid:11)r−A(cid:12)andfirmsadver- We now consider the situation where firms incur a
tise to their h consumers with a probability one and to fixed cost to acquire the ability to target their adver-
comparison shoppers with a probability of (cid:26) =1−A/r. tising.Mostoftenthisconsistsofpurchasingtargeting
t
In addition, firms employ mixed pricing strategies with information from market research firms, purchas-
F(cid:11)p(cid:12)=0 for p<A, F(cid:11)p(cid:12)=(cid:11)r(cid:11)p−A(cid:12)(cid:12)/(cid:11)p(cid:11)r−A(cid:12)(cid:12) for p∈ ing information technology to better understand the
(cid:14)A(cid:3)r(cid:15), and F(cid:11)p(cid:12)=1 for p>r.
In this setting, neither the advertising strategy nor 13Notethatthepricingdistributionwithuniformpricingfirstorder
stochastically dominates the pricing distribution for comparison
profits are affected when firms that can target adver-
shopperswithtargetedpricing.Thisimpliesthattheaverageprice
tising obtain the ability to target prices. Moreover,
underuniformpricingisstrictlygreaterthantheaveragepricefor
the advertising intensity to comparison shoppers is comparisonshoppersundertargetedpricing.
Iyer,Soberman,andVillas-Boas: TheTargetingofAdvertising
MarketingScience24(3),pp.461–476,©2005INFORMS 469
Table1 EquilibriumOutcomesasaFunctionofTargeting(cid:1)(cid:2)=0(cid:3)∗ shoppers. When Firm 2 is already reaching all the
consumers in the market, reducing the advertising to
AdvertisingprobabilitiesbysegmentandprofitsRange:
comparisonshoppershelpsFirm1toreducethelevel
A>hr
of market competition. Thus, there are three possible
Case1: Case2: Case3: Case4: cases: two cases where either one of the firms adver-
Uniform Targeted Uniform Targeted
tises with probability less than one (while the other
advertising, advertising, advertising, advertising,
advertises with probability one) and the third case in
uniform uniform targeted targeted
Advertising pricing pricing pricing pricing which both the firms advertise with probability less
than one. The derivation of all the cases are provided
A−hr A−hr
Advertising(cid:1)h(cid:3) 1− 1 1− 1 in the appendix. Proposition 5 provides the details of
sr sr
the equilibrium. The superscript n on the profit for
A−hr A A−hr A
Advertising(cid:1)s(cid:3) 1− 1− 1− 1− Firm 1 indicates that the expression pertains to the
sr r sr r
price and advertising subgame before the investment
Profits 0 h(cid:1)r−A(cid:3) 0 h(cid:1)r−A(cid:3)
decision f.
∗Withtargetedpricing,thepricetothehsegmentisrandthepricetothe
ssegmentisinmixedstrategies. Proposition 5. WhenonlyFirm1targetsitsadvertis-
ing, there are two possible types of equilibria: either (cid:25) <1
1
and (cid:20) =1 or (cid:25) =1 and (cid:20) <1. Furthermore, Firm 1
mediabehaviorofconsumers,orincurringthecostof 2 1 2
always advertises to its h segment with probability one.
using an advertising agency.
(1) For low cost of advertising 0<A<hr, the equilib-
Assume that firms can make an ex ante invest-
riuminvolves(cid:25) =1−A/r and(cid:20) =1.Firm1’sprofitsare
ment f to acquire the ability to target advertising. 1 2
(cid:13)n=h(cid:11)r−A(cid:12)andFirm2’sprofitsare(cid:13) =rh−A(cid:11)1−s(cid:12).
This game can be represented as a two-stage game, 1 2
(2) For high cost of advertising A>r/2, the equilib-
where firms first decide whether or not to invest in
riuminvolves(cid:25) =1and(cid:20) =1−(cid:11)A−hr(cid:12)/(cid:11)sr(cid:12).Firm1’s
targeting and then compete in advertising and price. 1 2
equilibrium profits are (cid:13)n=A−A(cid:11)h+s(cid:12), while Firm 2
Toanalyzethissituation,wefirstidentifytheoptimal 1
makes zero profit.
strategies as a function of firm capabilities. Note that
(3) For intermediate costs hr <A<r/2, both types of
the optimal strategies when both firms use uniform
equilibriaarepossible.Buttheequilibriumwith(cid:25) <1and
advertising and when both firms target advertising 1
(cid:20) =1 Pareto dominates the equilibrium with (cid:25) =1 and
are described in §§3.1 and 3.2. Thus, to complete the 2 1
(cid:20) <1.
analysis, we analyze the case where a firm with tar- 2
geting capability (say, Firm 1) faces a firm that can When the costs of advertising are sufficiently low
only advertise uniformly (Firm 2). We first solve the (cid:11)A<hr(cid:12), the equilibrium involves (cid:25) <1 and (cid:20) =1.
1 2
price and advertising subgame and then analyze the With lower costs of advertising, the firm with uni-
decision to make the investments to target advertis- form advertising always advertises. In response, the
ing.14 Let (cid:25) be the probability that Firm 1 adver- firm with the ability to target advertising chooses
1
tises to comparison shoppers (it advertises to its high (cid:25) <1toreducecompetitionforcomparisonshoppers
1
preference segment with probability 1) and (cid:20) be ((cid:25) also decreases in A in this range). At the other
2 1
the probability that Firm 2 advertises uniformly to extreme, when the cost of advertising is sufficiently
the market. In this situation, when both firms adver- high (cid:11)A>r/2(cid:12) the equilibrium involves (cid:25) =1 and
1
tise to comparison shoppers, the firms’ prices are in (cid:20) <1. The firm with uniform advertising finds it too
2
mixed strategies, because each firm has an incentive expensive to advertise with probability one. In con-
to undercut the other to attract comparison shop- trast, the ability to target advertising and eliminate
pers. We start the equilibrium characterization with wastedadvertisingallowsFirm1toalwaysadvertise.
Lemma 2. Finally, in the intermediate range of A, both types
of equilibria are possible. However, the equilibrium
Lemma 2. The outcome with both (cid:20) =1 and (cid:25) =1
2 1 with the targeting firm advertising with probability
cannot be part of the equilibrium.
lessthanoneandtheuniformfirmalwaysadvertising
Suppose Firm 2 (the uniform advertising firm) is Pareto dominant. While analyzing the decision to
advertises with probability one. Then, Firm 1 (the investintargeting,wepicktheParetodominantequi-
targeting firm) earns a higher profit by advertising libriumastherelevantonewhenadvertisingcostsare
with a probability less than one to the comparison in the intermediate range.
Theaboveresultshighlightsomeinterestingaspects
of competition between the two firms that have dif-
14Asmentionedintheprevioussection,werestrictourattentionto
ferent capabilities. For A above r/2, the inability of
therangeofadvertisingcosts,whichruleoutthedegeneratecase
Firm 2 to always advertise confers a positive exter-
wherefirmswithuniformadvertisingabilitydonotadvertise;i.e.,
A<(cid:11)1−h(cid:12)r. nality on Firm 1. Firm 1 makes A−A(cid:11)h+s(cid:12), which
Iyer,Soberman,andVillas-Boas: TheTargetingofAdvertising
470 MarketingScience24(3),pp.461–476,©2005INFORMS
is strictly greater than the profit earned by only serv- uniformly.15 The analysis demonstrates that the bene-
ingitshighpreferencesegment.Inotherwords(from fitsoftargetingaregreaterforafirmthatfacesacom-
theperspectiveofFirm1),allpotentialprofitoncom- petitor that uses uniform advertising than for a firm
parisonshoppersisdissipatedwhenadvertisingcosts that faces a competitor that has targeting capability.
are low enough because Firm 2 finds it optimal to
alwaysadvertise.Whenadvertisingcostsarehigh,the 3.5. Positive Endowment of Consumer
reduced advertising by Firm 2 mitigates the compe- Information
tition for the comparison shoppers. Firm 1’s profit is In the preceding analysis, we assume that without
increasing in A when A>r/2. Here, even though an advertising from a firm, consumers are not informed
increaseinAmakesitmoreexpensiveforFirm1(the ofthefirm’sproductanddonotconsideritspurchase.
target advertising firm) to advertise, it also has the In this section, we relax this assumption to allow for
effectofmakingFirm2(theuniformadvertisingfirm) consumer product knowledge even in the absence
to advertise less. For Firm 1, the impact on profits of of advertising. This reflects the fact that in many
havingaweakercompetitoroutweighstheaddedcost markets, consumers have knowledge about products
of communicating with the market. and might consider a product even in the absence of
We now analyze the decisions of the firms to advertising. In particular, let a fraction of the high
h
invest f to obtain targeting capability. Figure 1 illus- preference consumers of each firm be informed with-
tratesthepayoffsofthefirmsbasedontheirdecisions out advertising, while a fraction of the comparison
s
to either invest or not invest in targeting capability. shoppers are similarly informed about both prod-
In this Figure, (cid:13) is the profit where both firms use ucts. This implies that each firm will have a group
u
uniform advertising, (cid:13) is the profit where both firms of (cid:11)1− (cid:12)h high preference consumers who are unin-
t h
usetargetedadvertising,(cid:13) istheprofitofafirmwith formed and who need advertising to be activated to
a
targetingcapabilitywhenitscompetitordoesnot,and consider buying the product. Similarly, the fraction
(cid:13) istheprofitofafirmthatusesuniformadvertising (cid:11)1− (cid:12)s of comparison shoppers need advertising to
d s
against a firm that targets its advertising (all profit be informed and to consider buying one of the two
quantities are net of f). products.
Because a firm can potentially sell even without
Proposition 6.
advertising, the pricing strategy of a firm is condi-
(1) When0<A<r/2,bothfirmswilltargetiff <Ah,
tional on whether it is advertising or not. Denote
onlyonefirmwilltargetiff ∈(cid:14)Ah(cid:3)A(cid:11)1−h(cid:12)(cid:15),andneither
the c.d.f. of firm i’s mixed strategy price distribution
firm will target if f >A(cid:11)1−h(cid:12).
without advertising to be G(cid:11)p(cid:12) and with advertising
(2) When A > r/2, both firms will target if f < i
to be F(cid:11)p(cid:12). Consider the case of uniform advertising.
h(cid:11)r−A(cid:12), only one firm will target if f ∈(cid:14)h(cid:11)r−A(cid:12)(cid:3)Ah(cid:15), i
If both firms are not advertising, then a firm’s profit
and neither firm will target if f >Ah.
function is (cid:13)(cid:11)p(cid:12)= hp+ sp(cid:11)1−G(cid:11)p(cid:12)(cid:12). In this case,
For the entire range of advertising costs there is i h s j
a firm by charging the reservation price can guar-
a consistent pattern of equilibrium outcomes. Three
antee itself a profit of hr. Next, when both firms
types of equilibrium outcomes are possible. When f h
advertise in equilibrium, the profit of Firm i while
issufficientlylow,theequilibriuminvolvesbothfirms
charging p will be (cid:13)(cid:11)p(cid:12)=hp+sp(cid:11)1−F(cid:11)p(cid:12)(cid:12)−A. Thus,
investing in targeting. On the other hand, if the costs i j
the firm while advertising can charge the reservation
of targeting are high, both firms will choose to use
priceandguaranteeitselfaprofitofhr−Aandifthis
uniform advertising and not invest in targeting. But
profit is greater than hr, firms will always adver-
themoreinterestingpointisthatwhentargetingcosts h
tise in equilibrium. Then, as in the previous analysis,
are in an intermediate range, there is an asymmetric
wehavethatfirmswilladvertisewithprobabilityone
equilibrium. In other words, ex ante identical firms
if A<hr(cid:11)1− (cid:12). If advertising is sufficiently expen-
differentiate in the decision to acquire the ability to h
sive and A≥hr(cid:11)1− (cid:12), the equilibrium will involve
target advertising: while one firm makes the invest- h
mixed advertising strategies. For this case, the profit
mentf,theotherchoosesnottoinvestandadvertises
of a firm is then (cid:13)(cid:11)p(cid:12)=hp+ sp(cid:14)(cid:11)1−(cid:20)(cid:12)(cid:11)1−G(cid:11)p(cid:12)(cid:12)+
i s j j
(cid:20)(cid:11)1−F(cid:11)p(cid:12)(cid:12)(cid:15)+(cid:11)1− (cid:12)(cid:14)(cid:11)1−(cid:20)(cid:12)sp+(cid:20)sp(cid:11)1−F(cid:11)p(cid:12)(cid:12)(cid:15)−
j j s j j j
Figure1 NormalFormofDecisiontoInvesttoObtainTargeting A. From this the symmetric equilibrium condition
Capability
Firm2
15This might be seen as related to Mills and Smith (1996) who
Uniform Targeted
argue that asymmetric firms arise endogenously if the fixed costs
to acquire a lower marginal cost of production are in an interme-
Uniform (cid:13) (cid:3)(cid:13) (cid:13) (cid:3)(cid:13) −f
u u d a diate range. Note, however, that while a firm having lower costs
Firm1 alwayshurtsthecompetitor,inthispaper,afirminvestingintar-
Targeted (cid:13) −f(cid:3)(cid:13) (cid:13) −f(cid:3)(cid:13) −f geting ability benefits the competitor if the competitor does not
a d t t
havetargetingability.