Advertising Policies: Comparative versus Informative Advertising in Oligopolistic Markets.

Transcription

1 Advertsng Polces: Comparatve versus Informatve Advertsng n Olgopolstc Markets. Alprant Mara Mtrokostas Evangelos Petraks Emmanuel Unversty of Crete Unversty of Cyprus Unversty of Crete June 009 Abstract The present paper nvestgates frms ncentves to nvest n nformatve and/or n comparatve advertsng, n an olgopolstc market wth horzontal and vertcal product dfferentaton. We show that, frms optmal endogenous choce s to nvest n a combnaton of nformatve and comparatve advertsng, where the nvestment levels of comparatve advertsng are postvely related to the degree of substtutablty between products, whle the nvestment levels of nformatve advertsng are negatvely connected to the degree of substtutablty. Moreover, the comparson wth the case of mere nformatve advertsng and the case wthout any advertsng actvtes reveals that the use of comparatve advertsng leads to lower frms proftablty and therefore, t can be characterzed as wasteful advertsng. Fnally, the mpact of advertsng on socal welfare s also dscussed. Keywords: Informatve Advertsng, Comparatve Advertsng, Olgopoly, Product Dfferentaton. Department of Economcs, Unversty of Crete; e-mal: alprant@stud.soc.uoc.gr Department of Economcs, Unversty of Cyprus; e-mal: mtrokostase@econ.soc.uoc.gr Department of Economcs, Unversty of Crete; e-mal: petraks@econ.soc.uoc.gr

2 1. Introducton. Undoubtedly, from 1971 where the Federal Trade Commsson (FTC) began to encourage the use of comparatve advertsng, whch s the form of advertsng that compares rvals brands on obectvely measurable attrbutes or prce, and dentfes the rval brand by name, llustraton or other dstnctve nformaton 1, as a legal practce whch mproves market competton, ths alternatve way of advertsng has receved large publcty and has became a prevalent marketng practce. The advertsng campagns of Peps and Coke, Unlevel and Procter & Gamble, or Duckng Donuts and Starbucks are only few examples of the extended use of comparatve advertsng n several dfferent ndustry sectors. However, despte the large popularty of ths aggressve marketng practce, ts effectveness s stll under consderaton. On the one hand, there are some emprcal evdence that show that the comparatve ads tent to be more effectve than the non comparatve n nducng attenton to the ad, message and brand awareness, favourable brand atttudes and purchase ntentons [Grewal et al.(1997), Jung and Sharon (00)]. On the other hand, comparatve advertsng, apart from legal rsks, can have several nocuous mpacts snce t enhances consumers mstrust and may leads to msdentfcatons of the sponsorng brands [Etgar (1980), Wlke and Farrs (1975), Prasad (1976), Barone and Mnard (1999)]. Gven the above evdence a number of questons arse about the economc analyss of comparatve advertsng comparng to the tradtonal non comparatve forms of advertsng that n our model are specfed to nformatve advertsng. What are the frms ncentves to nvest n comparatve and/or n nformatve advertsng n order to promote ther products? Whch s the optmal level of nvestment n each type of advertsng? How the frms nvestments n these strategc marketng practces affect the market outcomes and the socal welfare?. The present paper ams to address the above questons n a smple duopoly market wth horzontally and 1 Statement of polcy regardng comparatve advertsng, Federal Trade Commsson, Washngton, D.C., August13, 1979 Pechmann and Stewart (1990) showed that at the Unted States the 60 percent of all advertsng contaned ndrect comparatve clams, 0 percent contaned drect comparatve clams and only 0 percent contaned no comparatve clams whle, Muehlng et al. (1990) suggested that almost 40 percent of all advertsng content s comparatve. The dfference between drect and ndrect comparatve ads s based on whether the compettor s explctly named (or precsely ndentfed by logos and mages) or not.

3 vertcally dfferentated products, where consumers do not possess any nformaton about products. The basc dea behnd our model s that frms strategcally use nformatve advertsng n order to transmt nformaton to consumers about the product exstence, locaton and attrbutes and as consequence, to help them dentfy the product that covers better ther needs. On the contrary, comparatve advertsng s strategcally used by the frms n order to present ther product as superor to the rvals one. Therefore, the use of comparatve advertsng not only ncreases consumers valuaton for the advertsed product but also dmnshes consumers valuaton of the targeted product. However, despte the dfferences between these two marketng strateges they are both accompaned by a suffcently hgh advertsng cost. We develop a smple theoretc model, ncorporatng all the above features of the two types of advertsng, where frms compete by choosng ther nvestment levels n each type of advertsng and ther outputs. In partcular, these decsons are modelled as a two stage game where at each stage moves are smultaneous. Our man fndngs suggest that at the equlbrum frms have always strong ncentves to nvest n both nformatve and comparatve advertsng. The latter result, comes to a contradcton wth the semnal work of Anderson and Renault (006) who clam that when the products are of same qualty, frms have strong ncentves to advertse only ther own product specfcaton, and that there s no need for comparatve advertsng. Further, we show that the frms expendtures on comparatve advertsng are postvely related to the degree of product substtutablty whle, the expendture levels on nformatve advertsng are negatvely connected to degree of substtutablty between products. Moreover, by comparng the equlbrum outcomes of our basc model, where frms nvest n an combnaton of nformatve and comparatve advertsng, wth the equlbrum outcomes of the benchmark case wthout advertsng actvtes and the case of mere nformatve advertsng, we fnd that the frms nvestment n both types of advertsng always leads to a prsoners dlemma stuaton and frms conclude to be worst off. Thus, we are n a poston to say that comparatve advertsng can be characterzed as wasteful advertsng snce, t enhances market competton, actuate frms to overnvest n advertsng and leads them to lower profts. However, these frms strategc marketng actvtes seems to be consumers benefcal snce they lead 3

4 to better nformed consumers and therefore, to hgher net consumers surplus. The above results also hold under Bertrand Competton. To the best of our knowledge, n economcs the aspect of comparatve advertsng has receved lttle attenton and the exstng theoretcal lterature on ths topc s lmted. Aluf and Shy (001) usng a Hotellng type model, where the use of comparatve advertsng ncreases the transport cost to the rval s product, show that ths form of advertsng weakens prce competton by enhancng the degree of product dfferentaton n consumers eyes and thus, leads to hgher prces and profts. In a dfferent ven, Bargozz et al.(006) examne comparatve advertsng as a mean to sgnal qualty by consderng a market where an entrant decdes between the use of generc advertsng (whch s a standard money burnng to sgnal qualty) and comparatve advertsng n order to confront an ncumbent whose qualty s known 3. They conclude that the entrant s ncentves to use comparatve advertsng are close related wth the qualty of hs product and the penalty that he s gong to pay f the content hs advertsng campagn s manpulatve 4. Further, they show that comparatve advertsng s seemed by consumers as a credble source of nformaton snce consumers nfer the use of comparatve nstead of generc advertsng as a warranty of qualty. From an alternatve perspectve, Anderson and Renault (006) nvestgate frms ncentves to mpart product nformaton through non comparatve or comparatve advertsng, and show that only a frm whch produce a low qualty product has strong ncentves to use comparatve advertsng (f t s legal) n order to survve n the market by targetng the frm of the hgh qualty product. However, such comparatve advertsng has detrmental mpact on the welfare snce t leads more consumers to consume a low qualty product. Although Anderson s and Renault s (006) work s a sgnfcant frst step to better understandng the frms ncentves to use comparatve nstead of non comparatve advertsng, ther modellng approach s specfed only to the nformatve character of comparatve advertsng and falures to capture ts persuasve character. 3 The sgnalng role of advertsng s based on the dea that hgh advertsng spendng works as a devce desgned to sgnal hgh qualty (e.g Nelson, 1974; Khlstrom and Rordan, 1984; Mlgrom and Roberts, 1986) 4 They assumed that when the entrant uses comparatve advertsng, the ncumbent has the opportunty to go to court, and obtan gans f the court verdct s that the entrant s qualty s low. 4

5 Thus, the present paper s novel n several dmensons. Frst, unlke the bulk of the lterature whch examnes the aspect of comparatve advertsng only exogenously, we analyze endogenously the frms ncentves to nvest n comparatve or/and n nformatve advertsng. Moreover, the present paper, contrary to the exstng lterature, gves answers not only about the type of advertsng but also about the expendture levels on each type of advertsng that frms are wllng to undertake n order to promote ther products. Further, n our duopolstc model, followng Häckner (000) along wth Garella and Petraks (006), we assume a consumers utlty functon whch ncludes both horzontal and vertcal product dfferentaton, where the vertcal dfferentaton can be perceved by consumers through the use of advertsng, a fact that seeks to capture the persuasve role of comparatve advertsng. The rest of the paper proceeds as follows. In the next secton we ntroduce our basc model along wth the benchmark case wthout any advertsng actvtes. In the secton 3, we adduce the equlbrum analyss and the man results of our research. In secton 4, we carry out the welfare analyss. In subsecton 5.1 we consder the comparson between our basc model and the case of mere nformatve advertsng, whle n the subsecton 5. we analyze how the equlbrum results change under Bertrand competton. Fnally, secton 6 concludes.. The Basc Model. We consder a market whch conssts of two frms, denoted by, =1,,, each producng one brand of dfferentated good. On the demand sde, we assume a unt mass populaton of consumers composed by ndvduals who have homogenous preferences. In partcular, followng Häckner (000), the utlty functon of the representatve consumer s gven by: 1 U ( q, q ) = ( α + + κ κ ) q+ ( α+ + κ κ ) q [ q + q + γqq ] + m (1) where, q, =1, represents the quantty of good, bought by the representatve consumer and m s the respectve quantty of the composte good. The parameter γ [0,1] denotes the degree of product substtutablty. When γ = 0 the goods are ndependent whle, when γ = 1 the goods are perfect substtutes. Further, represents the level of nvestment n nformatve advertsng whle, κ denotes the level of nvestment n comparatve advertsng. In partcular, the frms nvestment n 5

6 nformatve advertsng gves consumers the relatve nformaton to dentfy the product that covers better ther needs. Thus, t ncreases consumers valuaton whch shfts the demand curve outwards. On the other hand, comparatve advertsng has a dual effect. It has a postve mpact on the advertsed frm s demand snce, t ncreases the consumers valuaton of the advertsed product, whle, t has a detrmental mpact on the rval s frms demand snce, t decreases the consumers valuaton of the targeted product. Maxmzaton of utlty (1) wth respect to q, q gves the (nverse) demand functons for the representatve consumer: p = α + + κ κ q γq, =1,, ; () where p, p are the frms prces, whle the prce of the composte good has been normalzed to unty. We assume that both frms are endowed wth dentcal constant returns to scale producton technologes, wth ther margnal producton cost denoted by c (c α). Further, we hypothesze a quadratc cost of advertsng, gven by: b( + κ ), where the parameter b represents the effectveness of the advertsng technology. The hgher the parameter b s; the lower s the effectveness of advertsng. Equvalently, the hgher the parameter b s, the hgher are the requred nvestment levels n advertsng to obtan a gven demand shft, ceters parbus. Hence, the total cost functon of frm s gven by: C (.) = c q + b( + κ ). As t follows, the frm s net profts can be expressed as: Π = ( α + + κ κ q γq ) q cq b( + κ ) =1,,; (3) We consder a two stage game wth the followng tmng. In the frst stage, both frms decde, ndependently and smultaneously, upon the nvestment level of each type of advertsng that they are gong to undertake n order to promote ther products. In the second stage, gven the decsons of the prevous stage, they compete by settng ther outputs. We solve the game by applyng subgame perfecton..1 The Benchmark Case wthout any nvestment n advertsng. Before proceedng to the equlbrum analyss of the basc model, we brefly dscuss the benchmark case where frms do not undertake any advertsng actvtes, thus 6

7 = =0, κ =κ =0. Hence, the market outcomes can be descrbed by the standard Cournot game wth dfferentated products, where each frm chooses ts output, n order to maxmze ts profts, Π C = ( α q γq ) q cq. Takng the frst order condtons, we evaluate the reacton functon of each frm, whch s gven by: γq c ( ) C α q = R q = (4) Due to symmetry, the equlbrum prce, output and proft are, respectvely, c q = a C, + γ (1+ γ)c = a+ p C C C (α - c), Π = ( q ) = (5) + γ (+ γ) Fnally, by assumng a unt mass populaton of consumers composed by ndvduals who have dentcal preferences, t turns out that each consumer buys a quantty q=q C from each good. Further, usng eq.(1) and eq.(5), t can be checked that the net consumers surplus and the total welfare are, respectvely, (α - c) (+ γ) (α - c) TW (6) (+ γ) C C CS net = (1+ γ ) qc = (1+ γ ), = (3+ γ ) 3. Equlbrum Analyss. In the last stage of the game each frm chooses ts output q, takng as gven the rval s output q along wth the expenses on each type of advertzng (,, κ, ), decded n the frst stage of the game, n order to maxmze ts profts gven by eq.(3). Takng the frst order condtons (foc) of eq.(3), we evaluate frm s reacton functon whch s gven by: q α c γq + + κ κ = R ( q ) = (7) Clearly from the above expresson an ncrease n frm s nformatve and/or comparatve advertsng nvestment levels tends to ncrease ts level of producton and thus, the equlbrum output. On the other hand, equlbrum output tends to decrease when the rval s nvestment n comparatve advertsng ncreases. 7

8 By solvng the system of reacton functons eq. (7), we obtan the equlbrum output of the second stage, ( γ )( α c) + ( + κ κ ) γ ( + κ κ ) q (.) = (8) 4 γ In the frst stage of the game, frm chooses the expendture level of each type of advertsng (, κ ) n order to maxmze ts profts. Thus, frm s maxmzaton problem s: * [( γ )( α c) + ( + κ κ ) γ ( + κ κ )] maxπ (.) =, κ (4 γ ) b( κ + ) (9) Applyng frst order condtons, we get the best reply functons for both nformatve and comparatve advertsng gven by: [( γ )(α - c) + (+ γ )( κ κ ) γ ] ( ) = (10) b( γ 4) 4 ( γ )(α - c) ( γ + ) κ γ + κ ( κ ) = (11) [ b( γ ) 1]( γ + ) From the above equatons t s clear that both nformatve and comparatve advertsng effort that frm s wllng to undertake, decrease as the effectveness of advertsng technology decrease (for the hgher values of b), ceters parbus. In addton, we observe strategc substtutablty between nformatve and comparatve advertsng snce the nvestment level of both nformatve and comparatve advertsng tend to decrease as the correspondng rval s advertsng levels tend to decrease. By mposng symmetry and solvng the frst order condtons system, we obtan frm s equlbrum nvestment level for both nformatve and comparatve advertsng, * (α - c) = > 0 (1) b( γ )( γ + ) * (α - c)( γ + ) κ = > 0 (13) b( γ )( γ + ) 8

9 From eq.(1) and eq.(13) one can easly observe that at the equlbrum there always exst strong ncentves for the competng frms to nvest both n nformatve and n comparatve advertsng. The ntuton behnd ths result s that frms are wllng to undertake both types of advertsng n order to ncrease ther demand not only by attractng consumers through the use of nformatve advertsng but also by decreasng the rval s demand through the comparson. Hence, the followng proposton can be stated: Proposton 1. In equlbrum, frms optmal endogenous choce s to nvest n a combnaton of nformatve and comparatve advertsng, gven that the parameter of advertsng technology effectveness (b) s suffcently hgh. Further, by eq. (1) and eq.(13) t s clear that as the substtutablty between products ncreases,.e, as the parameter γ 1, frms nvestment levels n nformatve advertsng decrease whle, the nvestment levels n comparatve advertsng ncrease. The ntuton s as follows. It s well known that when the products tend to be close substtutes the market competton augments. Thus, each frm has a strong ncentve to nvest more n comparatve advertsng than n nformatve n order to ncrease ts demand and decrease the demand of ts rval by convncng consumers that ts product s superor than the rval s one. Moreover, t can be easly testfed by the eq.(1) and eq.(13) that the more effectve the advertsng technology s,.e, the lower values of b, the hgher s the nvestment n advertsng, ceters parbus. The followng lemma summarzes: Lemma 1. a) The equlbrum nvestment level of nformatve advertsng s decreasng n the degree of substtutablty between products (γ) and n the effectveness degree of advertsng technology (b). Therefore, t s hgher when goods tend to be ndependent and advertsng technology s effcent. b) The equlbrum nvestment level of comparatve advertsng s ncreasng n the degree of substtutablty between products and s therefore hghest when the goods are perfect substtutes whle, t s decreasng n the effectveness degree of advertsng technology (b). 9

10 Pluggng eq.(1) and (13) nto (6) and (3), frm s equlbrum output and profts are gven, b(α - c)( γ 4) = > 0 (14) b( γ )( γ + ) + q Π (α - c) {b[b( γ 4) γ ( γ + 4) 8]} 3 = > 0, for b > (15) [ b( γ )( γ + ) + ] Note that the necessary stablty condton at the equlbrum, that s a condton n order the frms profts to be postve, s b >3/. Gven the eq.(15) t can be testfed after some manpulatons, that frms equlbrum profts tend to ncrease as the effectveness of the advertsng technology decreases, ceters parbus. The ntuton behnd ths result s based on two alternatves aspects. Frstly, as we have prevously shown, each frm s expendtures on comparatve advertsng tend to decrease when the advertsng technology tends to be neffectve. Therefore, the lower frms nvestment n comparatve advertsng declnes the dmnshng effect on each frm s demand due to rval s comparatve advertsng and thus, t acts benefcally to profts. On the contrary, an neffectve advertsng technology leads also to lower frms nvestment n nformatve advertsng whch n turn, t tends to decrease the benefcal effect of nformatve advertsng on each frm s demand and thus, to decrease profts. However, our equlbrum results reveal that the benefcal effect of the low effectveness of advertsng technology domnates snce t prevent frms from overnvestment n advertsng. In other words, a reducton n b enhances market competton and as a consequence the frms equlbrum profts are reduced. The latter result s n the same sprt as Bester and Petraks (1995) and Peters (1984) who clam that frms are better off under advertsng restrctons. The followng proposton summarzes: Proposton. Equlbrum output s decreasng both n the degree of substtutablty between products γ and n the advertsng effectveness parameter b, whle equlbrum profts are decreasng n γ but ncreasng n b. 10

11 3.1 A Comparson We turn out now to the comparson of the equlbrum outcomes of our basc model and the benchmark case wthout advertsng actvtes. It s obvous that eq.(7) can also be wrtten as: α c γq + κ κ q = R ( q ) = + (16) C Comparng R q ) wth the reacton functon of the no advertsng case R q ), n ( whch only the left term of eq.(16) appears, we observe the dual effect that the use of advertsng has over the output. On the one hand, frm s expendtures on nformatve and comparatve advertsng (, κ ) tend to augment ts demand and thus, ts producton and equlbrum output. On the contrary, the rval s frm nvestment n comparatve advertsng (κ ) tends to decrease equlbrum output. However, at the equlbrum due to the symmetry of the game we have that the optmal expendture on comparatve advertsng s equal for both frms, κ * = κ * = κ *, thus, the effect of each frm s nvestment n comparatve advertsng tends to neutralze one another and as a consequence, q >q C always holds. Further, by evaluatng the dfference between the equlbrum profts of the case where frms nvest both n comparatve and nformatve advertsng gven by eq.(15) and the profts of the benchmark case gven n eq.(5) we have that, ( Π Π C ( α c) [ bγ ( γ + ) (8+ γ ) + 4] = < 0 ( γ + ) [ b( γ )( γ + ) + ] (17) Clearly from the above equaton, snce both the nomnator and the denomnator are postve for all the gven values of the parameters b and γ, C Π > Π always holds. Thus, t s obvous that frms conclude n a prsoners dlemma stuaton snce, the nvestment n both nformatve and comparatve advertsng ncreases market competton and leads frms to be worst off. As a consequence, the optmal decson of frms to nvest n a combnaton of comparatve and nformatve advertsng can be characterzed as the case of wasteful advertsng 5. The followng lemma summarzes: 5 The term was frst ntroduced by Pgou 194, n order to descrbe the Prsoner s Dlemma whch arses when competng frms n a market nvest equal efforts n advertsng n order to attract the favor 11

12 Lemma. a) The equlbrum output when frms nvest both n nformatve and comparatve advertsng s always hgher than that of the case wthout any advertsng actvtes. b) Frms equlbrum profts when they do not undertake any advertsng actvtes always exceed the equlbrum profts when frms nvest both n nformatve and comparatve advertsng. 4. Welfare Analyss. In ths secton we dscuss the mpact of the frms decson to nvestment n both nformatve and comparatve advertsng on the socal welfare. Total welfare s defned as the sum of consumers and producers surplus: TW = + (18) CS net * Π wth CS net and Π * correspondng to the net consumers surplus and the overall market profts respectvely. In partcular, the net consumer surplus for the representatve consumer s gven by the followng expresson: CS net 1 ( α+ + κ κ ) q+ ( a+ + κ κ) q ( q + q + qq ) pq p q = γ (19) In equlbrum, due to symmetry, we have that * * κ = κ = wrtten as: * κ and p = q = = q q, * * * =, = = p p. Thus, after some manpulatons eq.(19) can be CS net γ = (1+ )[ q ] (0) From the eq. (0) we observe that snce CS net s postvely connected to the equlbrum output, t tends to decrease as the parameters γ and b ncrease. The ntuton behnd ths s based on two dfferent aspects. Frst, for a gven effectveness of advertsng, as the products tend to be homogeneous frms tend to nvest more n comparatve and less n nformatve advertsng whle, the overall expendture on of the publc from the others. As Pgou frst showed ths concludes n a Prsoner s Dlemma where none of the frms gans anythng at all. 1

13 advertsng tend to decrease. Hence, snce the use of comparatve advertsng ams manly to alter the consumers valuaton about the rval s frm product than to transmt nformaton, along wth the reducton on the nvestment n advertsng, we conclude n less nformed consumers whch has a detrmental mpact on the net consumers surplus. Second, as the products tend to be perfect substtutes, the equlbrum output of each frm decreases whch tends to dmnsh the net consumers surplus. Further, wth respect to eq. (18), (0) and (14), the total welfare can be wrtten as: ( α c) b{[b( γ + 3)( γ - 4) -16+ γ ( γ + 4)]} TW = (1) [b( γ - )( γ + ) + ] As t can be easly testfed by eq.(1) the total welfare, when frms nvest both n nformatve and comparatve advertsng, s decreasng as goods tend to be homogenous and as effectveness of advertsng technology decrease,.e., for hgher values of b. We summarze our fndngs n the follow proposton: Proposton 3. In equlbrum, both the socal welfare and the net consumers surplus are decreasng n the degree of substtutablty between products γ and n the advertsng effectveness parameter b. Let us now compare our results regardng socal welfare to the benchmark case wthout advertsng actvtes. The followng lemma summarzes our fndngs: Lemma 3. a) The net consumers surplus when frms nvest n advertsng (nformatve and comparatve) always exceeds the net consumers surplus when frms do not undertake any advertsng actvtes. b) The socal welfare when frms nvest n advertsng and products are not close substtutes s hgher than that of the no advertsng case whle, t s lower when the products tend to be close substtutes. We turn now to dscuss the man arguments that drve our results. By usng the eq.(6) and the eq. (0), and snce q >q C always holds, t can be easly checked that 13

14 CS > CS for all the gven values of the parameters γ and b. The ratonale behnd net C net ths result s based on the dual benefcally effect of advertsng. Frstly, frms nvestment n advertsng leads to better nformed consumers whch acts benefcally to the net consumers surplus. Secondly, the ncreased competton among the frms, leads to hgher total producton, that make consumers better off. Further, by takng the dfference between eq. (6), we have that: TW gven by eq. (1) and C TW gven by TW TW C 4(3+ γ ) + b( γ + ) (3γ + 6γ 4) = ( γ + ) [ b( γ )( γ + ) + ] () From the above equaton t can be easly testfed, after some manpulatons, that C TW > TW when the products are not close substtutes whle, TW C > TW CΙ when the products tend to be close substtutes. Clearly, when the products are not close substtutes the benefcally effect of the hgher net consumers surplus when frms nvest n advertsng domnates the dmnshng effect of the lower frms proftablty. On the contrary, as the substtutablty between products ncreases the dmnshng effect of the lower profts domnates the benefcally effect of the hgher net consumers surplus. 5. Extensons 5.1 The case of mere nformatve advertsng. In ths secton we consder the case where frms nvest only n nformatve advertsng. In other words, κ =κ =0 6. Hence, keepng all the other modelng specfcatons fxed, frm s nverse demand s gven now by: p IN = α + q γq. Thus, n the fnal market competton stage, where frms compete by settng ther outputs, frm solves the maxmzaton program, IN maxπ = ( α + q γq ) q cq (3) q 6 Ths confguraton also reflects the case where consumers perceve comparatve advertsng as a manpulatng frms marketng practce whch does not capture any trustworthy nformaton. In ths case the results concde wth the ones obtaned n ths subsecton. 14

15 where 1 and are gven as frst stage choces. The best reply functon of frm as IN ( α c γq + ) functon of the rval s output q s q = R ( q ) =. Therefore by solvng the frst order condtons system, we evaluate equlbrum output, ( γ )( α c) + γ IN q = (4) 4 γ Note that equlbrum output n the case where frms nvest only n nformatve advertsng s decreasng as the rval s nvestment n advertsng ncreases and products tend to be close substtutes. Further n the frst stage of the game, the maxmzaton program of the frm s gven by: maxπ IN [( γ )( α c) + γ ] = (4 γ ) b (5) Applyng frst order condtons we have that the best reply functon n nformatve advertsng s gven by: IN IN [( γ )(α - c) γ ] ( ) = (6) b( γ 4) 4 Explotng symmetry we have that equlbrum level of nvestment n nformatve advertsng, output and profts are respectvely, (α - c) = b( γ + ) 1 IN *, b(α - c)( γ + ) = b( γ + ) 1 * q IN, IN * b(α - c) Π = (7) b( γ + ) 1 Further, the net consumers surplus and total welfare are gven respectvely by: CS IN net b = (1+ γ )(α - c) ( γ + ) [ b( γ + ) 1] b[ b(+ γ ) (3+ γ ) ](α - c) TW IN (8) [ b( γ + ) 1], = Note that the equlbrum profts, output, net consumers surplus and total welfare when frms nvest only n nformatve advertsng always exceed the equlbrum profts, output, net consumers surplus and total welfare of the benchmark case where frms do not undertake any advertsng actvtes. As a consequence, we observe that frms always have strong ncentves to nvest n nformatve advertsng. 15

16 5.1. A Comparson. We turn now to the comparson between the equlbrum outcomes when frms nvest both n nformatve and comparatve advertsng and when they nvest only n nformatve advertsng. Usng eq.(14) and eq.(7) t can be easly checked that the equlbrum output when frms nvest n an combnaton of nformatve and comparatve advertsng always exceeds the equlbrum output of the sngle nformatve advertsng case, equlbrum profts n these two cases, we have that, q >q IN always holds. Further, consderng the IN b{(+ γ ) [b( γ + γ + ) ]}(α - c) Π Π = (30) [ b( γ + ) 1][ b( γ )( γ + ) + ] From the above equaton t s obvous that IN Π >Π holds for all the gven values of parameters γ and b. Moreover the dfference between equlbrum profts s ncreasng n γ whle t s decreasng n b. Therefore t s hghest when goods are perfect substtutes, ceters parbus. The ratonale behnd ths result s straghtforward. The use of comparatve advertsng s consdered as a qute aggressve marketng practce; as a consequence, t leads to fercer market competton and lower frms proftablty. Clearly, we observe that a prsoners dlemma exsts snce, IN Π >Π always holds. Consderng now the socetal effects of comparatve advertsng we have that, snce the equlbrum net consumers surplus can be expressed as γ CS ( 1 )[ q * ] net = + and q >q IN always holds, CS > CS for all the gven values of γ and b. Therefore, the ncreased market competton, due to the use of comparatve advertsng, turns out to be consumers benefcal. Fnally, by takng the dfference between eq.(8) and eq.(1) and after some manpulatons t can be checked that the total welfare n the case of mere nformatve advertsng always exceeds the total welfare when frms nvest both n nformatve and comparatve advertsng, IN TW > TW. Hence, t s obvous that the effect of the hgher frms proftablty when frms nvest only n nformatve advertsng domnates the effect of the hgher net consumers surplus when frms nvest n a combnaton of nformatve and comparatve advertsng. From all the above, one can realze that the use of ths aggressve marketng practce enhances market competton and leads to lower frms proftablty and net IN net 16

17 lower socal welfare. Therefore, comparatve advertsng can be characterzed as wasteful advertsng. The above dscusson s summarzed n the followng lemma: Lemma 4. a) The equlbrum output and the net consumers surplus of the case where frms nvest n a combnaton of nformatve and comparatve advertsng are always hgher than those of the case where frms nvest only n nformatve advertsng. b) The equlbrum profts and the total welfare when frms nvest only n nformatve advertsng always exceed the equlbrum profts and the total welfare of the case where frms nvest both n nformatve and comparatve advertsng. 5.3 Bertrand Competton. In ths secton we consder the case where frms compete by choosng ther prces. In ths case each frm faces the lnear demand functon q (1 γ ) a+ γ + (1+ γ)( κ κ ) p p γ (1 γ ), 0 γ < 1. Keepng all the other modelng = specfcatons fxed, we observe that all of our man results stll hold under Bertrand market competton and the ntutve arguments are n lne wth the respectve ones n the Cournot case. In partcular, we show that frms n the equlbrum have always strong ncentves to nvest both n nformatve and comparatve advertsng for all the gven values of the substtutablty parameter γ. However, we observe that the necessary stablty condton at the equlbrum, changes from 3 b > under Cournot competton to b > 30 under Bertrand competton. Thus, our model possesses a prce settng equlbrum only f the cost of advertsng s sgnfcantly hgh. The ntuton behnd ths result s derved drectly by the fact that the Bertrand competton s harder than Cournot. Thus, as the advertsng tends to be neffectve frms tend to nvest less n advertsng whch protect them by the mpacts of the exstng prsoners dlemma. 17

18 6. Concludng Remarks. In the present paper, we have analyzed endogenously the frms ncentves to nvest n nformatve and/or n comparatve advertsng n an olgopolstc market wth horzontal and vertcal product dfferentaton, takng as basc premse, that nformatve advertsng s used by frms as a mean to transmt nformaton to consumers whle, comparatve advertsng s manly used n order to alter consumers preferences by presentng the advertsed product as superor to the rval s one. Our man fndngs suggest that, at the equlbrum frms optmal decson s to nvest n a combnaton of nformatve and comparatve advertsng, where, when the products are ndependent the frms expendtures on nformatve advertsng are equal to that of comparatve advertsng whle, when the products tend to be close substtutes frms tend to nvest more n comparatve advertsng than n nformatve. Ths result comes to a contradcton wth the exstng lterature that states that when the products are of same qualty frms have no ncentve to nvest n comparatve advertsng. Moreover, by comparng the equlbrum outcomes when frms nvest n a combnaton of nformatve and comparatve advertsng wth those of the no advertsng case and the case of mere nformatve advertsng, we show that frms endogenous optmal decson to nvest n both nformatve and comparatve advertsng always leads them to lower proftablty. Thus, frms fnd themselves n a prsoner s dlemma stuaton, and they conclude to be worst off. Hence, comparatve advertsng can be characterzed as wasteful advertsng snce both of the frms would be better off, f comparatve advertsng was prohbted. On the contrary, our analyss reveals that the use of comparatve advertsng s benefcal to consumers, snce t leads to hgher net consumers surplus due to the mproved nformaton that consumers possess through the use of advertsng. Although, our analyss was carred out for a duopolstc market, we strongly beleve that t provdes all the essental nsghts about the frms ncentves to nvest n nformatve and/or comparatve advertsng. We are also aware that the undertaken analyss has several lmtatons that emerge from the functonal forms we have use, but we strongly beleve that the use of more general modelng forms would eopardze the clarty of our fndngs wthout sgnfcantly changng ther qualtatve character. Further, gven the large publcty of comparatve advertsng, we beleve that our 18

19 analyss provde a frst theoretcal step n the better understandng of the strategc use of comparatve advertsng. However, there are several open drectons for future work. For nstance, an nterestng enough queston that arses s, how the frms ncentves to use nformatve and/or comparatve advertsng change, f we assume that a porton of consumers are already nformed about the exstng products n the market. Snce ths scenaro s not an easy task, ts analyss awats for further work. 19

20 7. References. Aluf, Y., and Shy, O., 001, Comparson Advertsng and Competton, mmeo, Unversty of Hafa. Anderson, S.P., and Renault R., (004), Negatve Advertsng a New Product!, mmeo, Unversty of Vrgna. Anderson, S.P., and Renault R., (006a), Comparatve Advertsng, mmeo, Unversty of Vrgna. Anderson, S.P., and Renault R., (006b), Advertsng Content, Amercan Economc Revew, Vol. 96, Bagwell, K., (003), The Economc Analyss of Advertsng, The Handbook of Industral Organzaton, Vol.3. Bargozz, F., Garella, P.G., and Petz, M., (007), Wth a Lttle Help from my Enemy: Comparatve Advertsng, Unversty of Bologna dscusson paper. Bargozz, F., Garella, P.G., and Petz M., (006), Advertsng and Prces as Sgnals of Qualty: Competng Aganst a Renowned Brand. Bargozz, F., and Petz M., (006), Comparatve Advertsng and Competton Polcy, n: Cho, J.P.(ed.), Recent Trends n Anttrust: Theory and Evdence, MIT Press, Barone, M.J., and Mnard, P.W., (1999), How and When Factual Ad Clams Mslead Consumers: Examnng the Deceptve Consequences of Copy X Copy Interactons for Partal Comparatve Advertsements, Journal of Marketng Research, Vol. 36(1),

21 Bester, H., and Petraks, E., (1995), Prce Competton and Advertsng n Olgopoly, European Economc Revew, Vol.39, Garella, P.G., Petraks, E., (006), Mnmum qualty standards and consumers nformaton, Economc Theory, Vol. 36()., Grewal, D.S. Kavanoor, E. Fern, C. Costley and J. Barnes, (1997), Comparatve Versus Non-comparatve Advertsng: A Meta- Analyss, Journal of Marketng, Vol.61(4), Grossman and Sharpo, (1984), Informatve Advertsng wth Dfferentated Products, The Revew of Economcs Studes, Vol.51. Häckner, J., (000), A note on prce and quantty competton n dfferentated olgopoles, Journal of Economc Theory, Vol. 93, Haller Hans and Chakrabart Subhadp, (00) An Analyss of Advertsng Wars, mmeo. Muehlng D., Stoltman J., Grossbart S., (1990), The Impact of Comparatve Advertsng on Levels of Message Involvement, Journal of Advertsng, Vol. 19, Nelson, P., (1974), Advertsng as Informaton, Journal of Poltcal Economy, Vol.84, Pechmann, C. and D.W. Steward, (1990), The Effect of Comparatve Advertsng on Attenton, Memory and Purchase Intenton, Journal of Consumer Research, Vol.17, Pechmann, C. and S. Ratneshwar, (1991), The Use of Comparatve Advertsng for Brand Postonng: Assocaton versus Dfferentaton, Journal of Consumer Research, Vol. 18,

22 Peters, M, (1984), Restrctons on prce advertsng, Journal of Poltcal Economy, Vol. 9, Pgou, A.C, (194), The Economcs of Welfare, London: Mcmllan and Co., Lmted. Second Edton. Wlke, W.L. and P.W. Farrs, (1975), Comparatve Advertsng: Problems and Potental, Journal of Marketng, Vol.39, 7-15.