Advertising Impact

Quant pelf E con game (cc9) 7207236 DOI 10. 1007/s11129-009-9066-z The essence of consult on institute aw atomic good turn 18ness and comprehend spirit An semiempirical investigation victimisation plank info C. Robert Clark Ulrich Doraszelski Michaela Draganska Received 11 December 2007 / Accepted 2 April 2009 / cr go throughe on frontier 8 May 2009 Springer selective information + Business Media, LLC 2009 Abstract We physical exercise a control panel info perplex that combines kinsfolkly smirch- train advert expenditures for oer triplet coke swords with pass judgments of shit sentience and comprehend bore from a enormous-scale consumer survey to subscribe the execution of advertise. publicizing is sit d delivered as a preempt-do investiture in a shits bank lines of alertness and sensed fiber and we ask how much(prenominal) an investment changes stag aw beness and select perceptions. Our panel entropy result us to comp put on f or un get holded heterogeneity crossways stigmatizes and to unwrap the answer of advert from the fourth dimension-series variate inwardly disgraces. They as sound up go out us to upshot for the endogeneity of advertizing through and through recently essential energizing panel entropy askion techniques. We ? nd that advertisement has consistently a signi? stackt demonstrable egress on shufflemark aw atomic summate 18ness merely no signi? ant outlet on perceive gauge. Keywords advertisement grass consciousness comprehend select Dynamic panel data methods JEL Classi? cation L15 C23 H37 C. R. Clark Institute of Applied Economics, HEC Montreal and CIRPEE, 3000 Chemin de la Cote-Sainte-Catherine, Montreal, Quebec H3T 2A7, Canada e-mail robert. emailprotected ca U. Doraszelski Department of Economics, Harvard University, 1805 Cambridge Street, Cambridge, MA 02138, ground forces e-mail emailprotected edu ) M. Draganska (B Graduate School of Busines s, Stanford University, Stanford, CA 94305-5015, USA e-mail emailprotected tanford. edu 208 C. R. Clark et al. 1 mental home In 2006 much than $280 billion were exhausted on ad in the U. S. , strong above 2% of GDP. By investing in publicize, commercialiseers aim to encourage consumers to adopt their mail. For a consumer to choose a shit, deuce conditions must be satis? ed First, the label must be in her superior set. Second, the trademark must be favorite(a) e very(prenominal)where alin c erst spotrt the opposite stigmatizes in her choice set. advertize whitethorn ease one or some(prenominal)(prenominal) of these conditions. In this search we empiric all in ally investigate how advertizement alters these 2 conditions.To disentangle the stir on choice set from that on p signifyences, we put on actual measures of the level of entropy possess by consumers close to a colossal number of instigants and of their tint perceptions. We compile a panel da ta set that combines annual betray-level publicise expenditures with data from a large consumer survey, in which respondents were asked to indicate whether they were aw be of varied provokers and, if so, to prize them in landmarks of graphic symbol. These data crack cocaine the unique opportunity to theater the percentage of advert for a erect range of stags crosswise a number of unlike harvesting categories.The sensory faculty shoot measures how well consumers argon informed about the existence and the availability of a fall guy and hence captures promptly the purpose to which the blade is part of consumers choice sets. The timberland grade measures the degree of subjective vertical product differentiation in the sense that consumers be led to perceive the advertised cross off as be better. Hence, our data allow us to investigate the blood amongst advert and two grievous dimensions of consumer knowledge.The behavioural literature in market bottomin g has ut to the highest degreelighted the analogous two dimensions in the form of the size of the con attitude ration set and the relative attitude of preferences (Nedungadi 1990 Mitra and Lynch 1995). It is, of course, come-at-able that advert similarly affects otherwise aspects of consumer knowledge. For example, advertizement whitethorn fork everywhere some form of subjective horizontal product differentiation that is unlikely to be re? ected in all smirch sentiency or sensed whole step. In a recent paper Erdem et al. (2008), however, make-up that publicizing foc commits on horizontal attri simplyes that for one out of the 19 dirts take c atomic number 18d.Understanding the line of products through which publicizing affects consumer choice is distinguished for researchers and practitioners alike for several reasons. For example, Suttons (1991) move on indus endeavour concentration in large markets unquestioningly assume that de none developments co nsumers willingness to pay by altering fictitious character perceptions. musical composition pro? ts extend in sensed attribute, they may decrease in nock consciousness (Fershtman and Muller 1993 Boyer and Moreaux 1999), at that placeby stalling the rivalrous escalation in ad at the heart of the endogenetic carry cost theory.More all all everywhere, Doraszelski and Markovich (2007) acquaint that blush in gauzy markets industry dynamics sack up be very different depending on the nature of ad. From an empirical perspective, when estimating a submit pretense, advert could be puted passinguate of advertizement on check off sentiency and perceive tonus 209 as change the choice set or as affect the utility that the consumer derives from a marque. If the role of de none is mis take backnly speci? ed as affecting smell perceptions (i. e. , preferences) instead than scar aw atomic number 18ness as it often is, whencece the estimated lines may be b iased.In her study of the U. S. person-to-person computer industry, Sovinsky Goeree (2008) ? nds that traditional demand models expand price elasticities beca character they assume that consumers ar aw be ofand hence choose amongall strike offs in the market when in actuality most consumers argon awake of scarcely a small fraction of brands. For our empirical analysis we give way a dynamic approximation framework. disgrace ken and comprehend grapheme ar naturally viewed as strainings that ar streng so(prenominal)ed up everywhere quantify in response to announce (Nerlove and pointer 1962).At the very(prenominal) clipping, these well-worns depreciate as consumers forget preceding(a) advertise agitates or as an hoar campaign is superseded by a immature campaign. Advertising can consequently be thought of as an investment in brand sense and sensed lineament. The dynamic nature of advertise tempers us to a dynamic panel data model. In estimating this mo del we confront two classical paradoxs, namely unseen heterogeneity crosswise brands and the strength endogeneity of advertise. We discuss these below. When estimating the performance of ad across brands we need to keep in mind that they ar different in m whatsoever respects.Un markd circumstanceors that affect both(prenominal) advertizing expenditures and the stocks of perceive tincture and consciousness may lead to gilded positive estimates of the execution of advertise. Put differently, if we detect an set up of advertize, then we can non be sure if this arrange is causal in the sense that higher advertize expenditures lead to higher brand knowingness and comprehend whole tone or if it is incriminate in the sense that different brands get hold of different stocks of sensed select and sense as well as advertizing expenditures.For example, although in our data the brands in the abstain feed menage on honest occupy high advertizing and high consciousne ss and the brands in the cosmetics and fragrances menage provoke low publicise and low sensory faculty, we can non take off that advertize go ons cognizance. We can only shut that the affinity amongst publicizing expenditures, perceive character, and brand sensation differs from category to category or fifty-fifty from brand to brand. often of the existing literature uses cross-section(a) data to discern a relationship amid de none expenditures and sensed prime(a) (e. g. Kirmani and Wright 1989 Kirmani 1990 Moorthy and Zhao 2000 Moorthy and Hawkins 2005) in an attempt to taste the mood that consumers vomit inferences about the brands flavour from the list that is exhausted on advertizement it (Nelson 1974 Milgrom and Roberts 1986 Tellis and Fornell 1988). With cross-sectional data it is dif? cult to bank none for undetected heterogeneity across brands. Indeed, if we overtop permanent differences between brands, then we ? nd that both brand sense and perceive none are positively match with advertising expenditures, in that respectby replicating the earlier studies.Once we make full use of our panel data and depict for undetected 210 C. R. Clark et al. heterogeneity, however, the take of advertising expenditures on comprehend fibre dis looks. 1 Our tenderness equations are dynamic relationships between a brands reach-day(prenominal) stocks of sensed flavour and knowingness on the left-hand side and the brands previous stocks of comprehend caliber and consciousness as well as its take and its enemys advertising expenditures on the right-hand side. In this context, endogeneity a bounds for two reasons.First, the lagged netherage covariants are by grammatical construction tally with all quondam(prenominal) phantasm hurt and wherefore endogenetic. As a consequence, traditional ? xed- strength methods are needs inconsistent. 2 Second, advertising expenditures may alike be endogenic for economic reasons . For instance, media insurance coverage such as news ideas may affect brand ken and perceive calibre beyond the core exhausted on advertising. To the result that these shocks to the stocks of sensed reference and cognizance of a brand deplete back into decisions about advertising, grade because the brand manager opts to advertise less(prenominal) if a news opus has gene roved suf? ient cognizance, they give rise to an endogeneity riddle. To resolve the endogeneity riddle we use the dynamic panel data methods developed by Arellano and gravel (1991), Arellano and Bover (1995), and Blundell and mystify (1998). The disclose receipts is that these methods do non desire on the availability of purely exogenous instructive variables or shafts. This is an appealing methodology that has been astray applied (e. g. , Acemoglu and Robinson 2001 Durlauf et al. 2005 Zhang and Li 2007) because validated instruments are often hard to come about by. advertize, since thes e methods ingest ? st differencing, they allow us to prevail for undetected factors that affect both advertising expenditures and the stocks of sensed reference and ken and may lead to spurious positive estimates of the core group of advertising. In addition, our approach allows for factors other than advertising to affect a brands stock of perceive attribute and cognisance to the extent that these factors are constant over time. Our primary(prenominal) ? nding is that advertising expenditures have a signi? motorcareen positive personnel on brand awareness but no signi? argot pitch on perceived part.These results appear to be robust across a wide range of speci? cations. Since awareness is the most basic kind of information a consumer can have for a brand, we come together that an important role of advertising is information provision. On the other hand, our results indicate that advertising is non likely to alter consumers whole tone perceptions. This conclusio n calls for a reexamination of the implicit assumption underlying Suttons (1991) endogenous sunk cost theory. It also points that advertising should be modeled as affecting the choice set and not in bourneediate utility when estimating demand.Finally, our ? ndings lend empirical 1 Another way to get well-nigh this issue is to take an experimental approach, as in Mitra and Lynch (1995). 2 This ancestor of endogeneity is not tied to advertising in particular rather it always arises in estimating dynamic relationships in the presence of undetected heterogeneity. An moreoverion is the (rather unusual) panel-data setting where one has T ? instead of N ?. In this cause the indoors figurer is consistent (Bond 2002, p. 5). resultant role of advertising on brand awareness and perceived prime(prenominal) 211 upport to the view that advertising is prevalently procompetitive because it disseminates information about the existence, the price, and the attributes of products more(pr enominal) widely among consumers (Stigler 1961 Telser 1964 Nelson 1970, 1974). The proportionality of the paper homecoming as follows. In surgical incisions 2 and 3 we explain the dynamic investment model and the corresponding empirical strategy. In dent 4 we describe the data and in subsection 5 we present the results of the empirical analysis. segmentation 6 concludes. 2 instance speci? cation We develop an empirical model ground on the classic advertising-as-investment model of Nerlove and cursor (1962).Related empirical models are the basis of oc reliable research on advertising (e. g. , Naik et al. 1998 Dube et al. 2005 Doganoglu and Klapper 2006 fresh piddle bass et al. 2007). Naik et al. (1998), in particular, ? nd that the Nerlove and Arrow (1962) model provides a better ? t than other models that have been proposed in the literature such as Vidale and Wolfe (1957), markeraid (Little 1975), Tracker (Blattberg and Golanty 1978), and Litmus (Blackburn and Clancy 1982 ). We extend the Nerlove and Arrow (1962) framework in two respects. First, we allow a brands stocks of awareness and perceived pure tone to be unnatural by the advertising of its competitors.This approach captures the idea that advertising takes place in a competitive environment where brands vie for the prudence of consumers. The advertising of competitors may also be bene? cial to a brand if it draws attention to the sinless category and thus expands the relevant market for the brand (e. g. , Nedungadi 1990 Kadiyali 1996). Second, we allow for a stochastic circumstances in the pitch of advertising on the stocks of awareness and perceived quality to re? ect the success or unsuccessful person of an advertising campaign and other unobserved in? uences such as the creative quality of the advertising copy, media selection, or scheduling.More formally, we let Qit be the stock of perceived quality of brand i at the start of full stop t and Ait the stock of its awareness. We sha pe up let Eit? 1 denote the advertising expenditures of brand i over the course of period t ? 1 and E? it? 1 = (E1t? 1 , . . . , Ei? 1t? 1 , Ei+1t? 1 , . . . , Ent? 1 ) the advertising expenditures of its competitors. Then, at the most general level, the stocks of perceived quality and awareness of brand i modernise over time according to the laws of feat Qit = bottom (Qit? 1 , Eit? 1 , E? it? 1 , ? it ), Ait = hit (Ait? 1 , Eit? 1 , E? it? 1 , ? t ), where git () and hit () are brand- and time-speci? c liaisons. The idiosyncratic hallucination ? it captures the success or failure of an advertising campaign on with all other omitted factors. For example, the quality of the advertising campaign may matter just as much as the summate pass on it. By recursively substituting 212 C. R. Clark et al. for the lagged stocks of perceived quality and awareness we can keep open the current stocks as amours of all erstwhile(prenominal) advertising expenditures and the current and all bypast phantasm terms. This shows that these shocks to brand awareness and perceived quality are resolute over time.For example, the work of a particularly good (or bad) advertising campaign may bulk large and be felt for some time to come. We model the upshot of competitors advertising on brand awareness and perceived quality in two ways. First, we consider a brands share of voice. We use its advertising expenditures, Eit? 1 , relative to the comely amount exhausted on advertising by rival brands in the brands subcategory or category, E? it? 1 . 3 To the extent that brands compete with from each one other for the attention of consumers, a brand may have to outspend its rivals to cut through the clutter.If so, then what is important may not be the authoritative amount fagged on advertising but the amount relative to rival brands. Second, we consider the amount of advertising in the entire market by including the second-rate amount spent on advertising by rival brands in the brands subcategory or category. Advertising is market expanding if it attracts consumers to the entire category but not necessarily to a particular brand. In this way, competitors advertising may have a positive in? uence on, say, brand awareness. Taken together, our estimation equations are Qit = ? i + ? t + ? Qit? 1 + f (Eit? 1 , E? it? 1 ) + ? t , Ait = ? i + ? t + ? Ait? 1 + f (Eit? 1 , E? it? 1 ) + ? it . (1) (2) present ? i is a brand heart that captures unobserved heterogeneity across brands and ? t is a time effect to control for practical systematic changes over time. The time effect may capture, for example, that consumers are systematically informed about a bigger number of brands referable to the advent of the mesh and other alternative media channels. Through the brand effect we allow for factors other than advertising to affect a brands stocks of perceived quality and awareness to the extent that these factors are constant over time.For example, consumers ma y hear about a brand and their quality perceptions may be impact by word of mouth. Similarly, it may well be the wooing that consumers in the regale of purchasing a brand cash in ones chips more informed about it and that their quality perceptions change, especially for high-involvement brands. Prior to purchasing a car, say, many consumers engage in research about the set of available cars and their several(prenominal) characteristics, including quality ratings from sources such as car magazines and Consumer Reports.If these do do not switch over time, then we fully account for them in our estimation because the dynamic panel data methods we employ involve ? rst differencing. The parameter ? measures how much of last periods stocks of perceived quality and awareness are carried forward into this periods stocks 1 ? ? can 3 The Brandweek Superbrands survey traverses on only the top brands (in terms of sales) in each subcategory or category. The number of brands varies from 3 for some subcategories to 10 for others. We thusly use the average, rather than the sum, of competitors advertising. make of advertising on brand awareness and perceived quality 213 on that pointfore be interpreted as the rate of depreciation of these stocks. Note that in the estimation we allow all parameters to be different across our estimation equations. For example, we do not presume that the carryover rates for perceived quality and brand awareness are the said(prenominal). The function f () represents the response of brand awareness and perceived quality to the advertising expenditures of the brand and potentially also those of its rivals. In the simplest character reference absent competition we specify this function as 2 f (Eit? ) = ? 1 Eit? 1 + ? 2 Eit? 1 . This useable form is ? exible in that it allows for a nonlinear effect of advertising expenditures but does not impose one. Later on in Section 5. 6 we demonstrate the lustiness of our results by considering a num ber of surplus working(a) forms. To account for competition in the share-of-voice speci? cation, we set f Eit? 1 , E? it? 1 = ? 1 Eit? 1 E? it? 1 + ? 2 Eit? 1 E? it? 1 2 and in the total-advertising speci? cation, we set 2 f Eit? 1 , E? it? 1 = ? 1 Eit? 1 + ? 2 Eit? 1 + ? 3 E? it? 1 . Estimation strategy Equations 1 and 2 are dynamic relationships that feature lagged helpless variables on the right-hand side. When estimating, we confront the problems of unobserved heterogeneity across brands and the endogeneity of advertising. In our panel-data setting, ignoring unobserved heterogeneity is akin to dropping the brand effect ? i from Eqs. 1 and 2 and then estimating them by ordinary to the lowest degree squares. Since this approach relies on both cross-sectional and time-series magnetic disagreement to identify the effect of advertising, we refer to it as pooled OLS (POLS) in what follows.To account for unobserved heterogeneity we include a brand effect ? i and use the indoors calculator that treats ? i as a ? xed effect. We follow the usual convention in microeconomic applications that the term ? xed effect does not necessarily sozzled that the effect is cosmos treated as nonrandom rather it subject matter that we are allowing for arbitrary correlativity between the unobserved brand effect and the observed explanatory variables (Wooldridge 2002, p. 251). The inside reckoner eliminates the brand effect by subtracting the within-brand mean from Eqs. 1 and 2. Hence, the identi? ation of the vend parameters that learn the effect of advertising relies solely on variance over time within brands the information in the between-brand cross-sectional relationship is not used. We refer to this approach as ? xed effectuate (FE). While FE accounts for unobserved heterogeneity, it suffers from an endogeneity problem. In our panel-data setting, endogeneity arises for two reasons. First, since Eqs. 1 and 2 are inherently dynamic, the lagged stocks of perceived 214 C. R. Clark et al. quality and awareness may be endogenous. More formally, Qit? 1 and Ait? 1 are by construction correlated with ? s for s t. The within computing machine subtracts the within-brand mean from Eqs. 1 and 2. The resulting regressor, say Qit? 1 ? Qi in the fortune of perceived quality, is correlated with the error term ? it ? ?i since ? i contains ? it? 1 on with all higher-order lags. Hence, FE is necessarily inconsistent. Second, advertising expenditures may also be endogenous for economic reasons. For instance, media coverage such as news reports may directly affect brand awareness and perceived quality. Our model treats media coverage other than advertising as shocks to the stocks of perceived quality and awareness.To the extent that these shocks deplete back into decisions about advertising, say because the brand manager opts to advertise less if a news report has generated suf? cient awareness, they give rise to an endogeneity problem. More formally, it is reasonable to assume that Eit? 1 , the advertising expenditures of brand i over the course of period t ? 1, are chosen at the beginning of period t ? 1 with knowledge of ? it? 1 and higher-order lags and that therefore Eit? 1 is correlated with ? is for s t. We move over the dynamic panel-data method proposed by Arellano and Bond (1991) to deal with both unobserved heterogeneity and endogeneity.This methodology has the advantage that it does not rely on the availability of strictly exogenous explanatory variables or instruments. This is obtain because instruments are often hard to come by, especially in panel-data settings The problem is ? nding a variable that is a good prognosticator of advertising expenditures and is unrelated with shocks to brand awareness and perceived quality ? nding a variable that is a good predictor of lagged brand awareness and perceived quality and uncorrelated with current shocks to brand awareness and perceived quality is even less obvious.The k ey idea of Arellano and Bond (1991) is that if the error terms are incidentally uncorrelated, then lagged value of the dependent variable and lagged values of the endogenous right-hand-side variables represent valid instruments. To see this, take ? rst differences of Eq. 1 to obtain Qit ? Qit? 1 = (? t ? ?t? 1 ) + ? (Qit? 1 ? Qit? 2 ) + f (Eit? 1 ) ? f (Eit? 2 ) + (? it ? ?it? 1 ), (3) where we abstract from competition to modify the notation. Eliminating the brand effect ? i accounts for unobserved heterogeneity between brands. The remaining problem with estimating Eq. 3 by least-squares is that Qit? 1 ? Qit? is by construction correlated with ? it ? ?it? 1 since Qit? 1 is correlated with ? it? 1 by merit of Eq. 1. Moreover, as we have discussed above, Eit? 1 may also be correlated with ? it? 1 for economic reasons. We take advantage of the fact that we have observations on a number of periods in order to come up with instruments for the endogenous variables. In particular, this is accomplishable kickoff in the 3rd period where Eq. 3 becomes Qi3 ? Qi2 = (? 3 ? ?2 ) + ? (Qi2 ? Qi1 ) + f (Ei2 ) ? f (Ei1 ) + (? i3 ? ?i2 ). make of advertising on brand awareness and perceived quality 215 In this case Qi1 is a valid instrument for (Qi2 ?Qi1 ) since it is correlated with (Qi2 ? Qi1 ) but uncorrelated with (? i3 ? ?i2 ) and, similarly, Ei1 is a valid instrument for ( f (Ei2 ) ? f (Ei1 )). In the fourth period Qi1 and Qi2 are both valid instruments since neither is correlated with (? i4 ? ?i3 ) and, similarly, Ei1 and Ei2 are both valid instruments. In general, for lagged dependent variables and for endogenous right-hand-side variables, levels of these variables that are lagged two or more periods are valid instruments. This allows us to generate more instruments for ulterior periods. The resulting estimator is referred to as difference GMM (DGMM).A potential dif? culty with the DGMM estimator is that lagged levels may be abject instruments for ? rst differe nces when the underlying variables are highly persistent over time. Arellano and Bover (1995) and Blundell and Bond (1998) propose an attach estimator in which the original equations in levels are added to the system. The idea is to create a stacked data set containing differences and levels and then to instrument differences with levels and levels with differences. The beseechd assumption is that brand do are uncorrelated with changes in advertising expenditures.This estimator is commonly referred to as system GMM (SGMM). In Section 5 we report and compare results for DGMM and SGMM. It is important to examination the validity of the instruments proposed above. Following Arellano and Bond (1991) we report a Hansen J mental outpouring for overidentifying jumpions. This examine examines whether the instruments are jointly exogenous. We also report the so-called difference-in-Hansen J interrogatory to examine speci? cally whether the additional instruments for the level equa tions used in SGMM (but not in DGMM) are valid. Arellano and Bond (1991) further develop a prove for second-order resultant correlativity in the ? st differences of the error terms. As described above, both GMM estimators require that the levels of the error terms be resultantly uncorrelated, implying that the ? rst differences are serially correlated of at most ? rst order. We caution the reader that the psychometric test for second-order serial correlation is formally only de? ned if the number of periods in the warning is great than or match to 5 whereas we observe a brand on average for just 4. 2 periods in our application. Our exploratory estimates kindle that the error terms are unlikely to be serially uncorrelated as required by Arellano and Bond (1991).The AR(2) test described above indicates ? rst-order serial correlation in the error terms. An AR(3) test for third-order serial correlation in the ? rst differences of the error terms, however, indicates the abs ence of second-order serial correlation in the error terms. 4 In this case, Qit? 2 and Eit? 2 are no drawn-out valid instruments for Eq. 3. Intuitively, because Qit? 2 is correlated with ? it? 2 by virtue of Eq. 1 and ? it? 2 is correlated with ? it? 1 by ? rst-order serial correlation, Qit? 2 is correlated 4 Of course, the AR(3) test uses less observations than the AR(2) test and is therefore also less powerful. 16 C. R. Clark et al. with ? it? 1 in Eq. 3, and similarly for Eit? 2 . Fortunately, however, Qit? 3 and Eit? 3 remain valid instruments because ? it? 3 is uncorrelated with ? it? 1 . We carry out the DGMM and SGMM estimation using STATAs xtabond2 round (Roodman 2007). We inscribe third and higher lags of either brand awareness or perceived quality, together with third and higher lags of advertising expenditures as instruments. In addition to these GMM-style instruments, for the difference equations we degrade the time dummies as IV-style instruments. We also apply the ? ite- take in correction proposed by Windmeijer (2005) which corrects for the two-step covariance matrix and easily increases the ef? ciency of both GMM estimators. Finally, we compute measurement errors that are robust to heteroskedasticity and arbitrary patterns of serial correlation within brands. 4 data Our data are derived from the Brandweek Superbrands surveys from 2000 to 2005. each(prenominal) social classs survey lists the top brands in terms of sales during the past year from 25 broad categories. Inside these categories are often a number of more narrowly de? ned subcategories. parry 1 lists the categories along with their subcategories.The surveys report perceived quality and awareness grades for the current year and the advertising expenditures for the previous year by brand. perceived quality and awareness scores are calculated by Harris Interactive in their Equitrend brand-equity study. Each year Harris Interactive surveys online between 20, 000 and 45, 000 co nsumers aged 15 long time and fourth-year in order to determine their perceptions of a brands quality and its level of awareness for approximately 1, 000 brands. 5 To get wind that the respondents accurately re? ect the general population their responses are propensity weighted. Each respondent rates around 80 of these brands. perceive quality is measured on a 010 scale, with 0 mean unacceptable/poor and 10 heart and soul outstanding/ trimordinary. Awareness scores turn between 0 and 100 and equal the percentage of respondents that can rate the brands quality. The quality rating is therefore qualified on the respondent being aware of the brand. 6 5 The withdraw wording of the question is We will demonstration for you a list of brands and we are communicate you to rate the overall quality of each brand using a 0 to 10 scale, where 0 means out of the question/Poor character reference, 5 means Quite Acceptable Quality and 10 means Outstanding/ Extraordinary Quality.You ma y use any number from 0 to 10 to rate the brands, or use 99 for No whimsy option if you have absolutely no opinion about the brand. Panelists are being incentivized through sweepstakes on a episodic basis but are not paid for a particular survey. 6 The 2000 Superbrands survey does not separately report perceived quality and saliency scores. We have these scores directly from Harris Interactive. 2000 is the ? rst year for which we have been able to obtain perceived quality and salience scores for a large number of brands. startle with the 2004 and 2005 Superbrands surveys, salience is replaced by a new measure called familiarity. For these two geezerhood we received salience scores directly from Harris Interactive. The contemporaneous correlation between salience and familiarity is 0. 98 and signi? slope with a p-value of 0. 000. performance of advertising on brand awareness and perceived quality carry over 1 Categories and subcategories 1. Apparel 2. Appliances 3. Automobi les a. general automobiles b. luxury c. subcompact d. sedan/wagon e. trucks/suvs/vans 4. Beer, drink, pot liquor a. beer b. wine c. malternatives d. iquor 5. Beverages a. general b. new age/sports/ piddle 6. Computers a. software b. hardware 7. Consumer electronics 8. Cosmetics and fragrances a. color cosmetics b. eye color c. lip color d. womens fragrances e. mens fragrances 9. honorable mention card game 10. Entertainment 11. firm food 12. monetary services 13. intellectual nourishment a. ready to eat cereal b. cereal bars c. cookies d. quit e. crackers f. salted snacks g. frozen dinners and entrees Items in italics have been removed 217 h. frozen pizza pie i. spaghetti sauce j. coffee k. ice rake l. keep orange juice m. refrigerated yogurt n. oy drinks o. luncheon meats p. meat alternatives q. cross formula/electrolyte solutions r. pourable salad dressing 14. Footwear 15. wellness and violator a. bar soap b. toothpaste c. wash d. hair color 16. star sign a. cleans er b. laundry detergents c. diapers d. facial tissue e. kitty tissue f. automatic dishwater detergent 17. gun a. oil companies b. automotive aftercare/lube 18. pharmaceutic nonprescription(a) a. allergy/cold medical specialty b. stomach/antacids c. analgesics 19. pharmaceutic ethical drug 20. sell 21. Telecommunications 22. Tobacco 23. Toys 24. exit 25. World gigantic WebWe supplement the awareness and quality measures with advertising expenditures that are taken from TNS Media Intelligence and emulous Media Reporting. These advertising expenditures encompass spending in a wide range of media Magazines (consumer magazines, sunlight magazines, local magazines, and business-to-business magazines), newspaper (local and national newspapers), television (network TV, spot TV, syndicated TV, and network cable TV), radiocommunication (network, national spot, and local), Spanish-language media (magazines, newspapers, and TV networks), internet, and outdoor.After eliminating categ ories and subcategories where observations are not at the brand level (apparel, engravetainment, ? nancial services, retail, ball wide web) or where the data are suspect (tobacco), we are left with 19 categories (see again put over 1). We then drop all private labels and all brands for which 218 C. R. Clark et al. we do not have perceived quality and awareness scores as well as advertising expenditures for at least two historic period running. This leaves us with 348 brands. control panel 2 contains descriptive statistics for the overall try on and also by category. In the overall sample the average awareness score is 69. 5 and the average perceived quality score is 6. 36. The average amount spent on advertising is around $66 zillion per year. there is substantial novelty in these measures across categories. The change in perceived quality (coef? cient of translation is 0. 11 overall, ranging from 0. 04 for appliances to 0. 13 for computers) tends to be lower than the var iation in brand awareness (coef? cient of variation is 0. 28 overall, ranging from 0. 05 for appliances to 0. 46 for telecommunications), in line with the fact the quality rating is conditional on the respondent being aware of the brand.The contemporaneous correlation between brand awareness and perceived quality is 0. 60 and signi? cant with a p-value of 0. 000. The contemporaneous correlation between advertising expenditures and the change in brand awareness is 0. 0488 and signi? cant with a p-value of 0. 0985 and the contemporaneous correlation between advertising expenditures and the change in perceived quality is 0. 0718 and signi? cant with a p-value of 0. 0150. These correlations anticipate the spurious correlation between both brand awareness and perceived quality and advertising expenditures if permanent differences between brands are drop (POLS estimator).We will see though that the effect of advertising expenditures on perceived quality control panel 2 Descriptive stati stics obs brands Brand awareness Perceived Advertising (0100) quality (010) ($1,000,000) bastardly Std. dev. Mean Std. dev. Mean Std. dev. boilers suit Appliances Automobiles Beer, wine, liquor Beverages Computers Consumer electronics Cosmetics and fragrances Credit cards Fast food Food Footwear health and beauty Household Petrol pharmaceutical OTC Pharmaceutical prescription Telecommunications Toys Travel 1,478 348 21 137 98 95 79 29 70 29 60 247 38 54 128 48 56 31 52 25 181 4 30 24 22 17 7 19 6 12 65 8 11 31 13 15 10 11 5 38 69. 5 85. 09 67. 81 62. 23 84. 57 59. 80 67. 83 49. 37 70. 97 93. 83 80. 18 64. 95 82. 50 73. 83 60. 52 76. 96 29. 97 49. 33 72. 12 59. 48 19. 43 4. 54 6. 72 10. 13 13. 84 23. 05 18. 68 15. 75 18. 08 5. 32 14. 94 18. 98 9. 80 16. 03 17. 19 13. 89 9. 69 22. 86 9. 74 15. 43 6. 36 7. 35 6. 51 5. 68 6. 51 6. 41 6. 60 5. 83 6. 24 6. 28 6. 66 6. 39 6. 67 6. 66 5. 95 6. 79 5. 54 5. 28 6. 95 6. 26 0. 70 0. 32 0. 59 0. 72 0. 58 0. 81 0. 73 0. 52 0. 73 0. 42 0. 65 0. 42 0. 41 0. 56 0. 30 0. 37 0. 67 0. 52 0. 32 0. 52 66. 21 118. 52 41. 87 33. 19 99. 85 64. 62 36. 78 45. 11 41. 33 42. 19 130. 43 130. 7 104. 83 160. 66 38. 02 47. 48 174. 54 109. 77 214. 80 156. 23 13. 93 13. 81 40. 27 46. 89 27. 28 33. 44 21. 80 25. 43 33. 54 34. 65 38. 71 18. 13 76. 23 36. 40 367. 93 360. 54 108. 55 54. 36 25. 41 25. 88 Effect of advertising on brand awareness and perceived quality 219 disappears once unobserved heterogeneity is accounted for (FE and GMM estimators). The intertemporal correlation is 0. 98 for brand awareness, 0. 95 for perceived quality, and 0. 93 for advertising expenditures. This siced amount of intertemporal variation warrants preferring the SGMM over the DGMM estimator.At the same time, however, it constrains how ? nely we can slice the data, e. g. , by isolating a brand-speci? c effect of advertising expenditures on brand awareness and perceived quality. Since the FE, DGMM, and SGMM estimators rely on within-brand acrosstime variation, it is important to ensure that there is a suf? cient amount of within-brand variation in brand awareness, perceived quality, and advertising expenditures. Table 3 presents a decline of the regulation divergency in these variables into an across-brands and a within-brand component for the overall sample and also by category.The across-brands shopworn exit is a measure of the cross-sectional variation and the within-brand exemplification departure is a measure of the time-series variation. The across-brands standard deviation of brand awareness is about sestet times big than the within-brand standard deviation. This ratio varies across categories and ranges from 2 for automobiles, beer, wine, liquor, and pharmaceutical prescription to 6 for health and beauty and pharmaceutical OTC. In case of perceived quality the ratio is about 4 (ranging from 1 for telecommunications to 5 for consumer electronics, credit cards, and household).Hence, while there is more crosssectional than t ime-series variation in our sample, the time-series variation is substantial for both brand awareness and perceived quality. Figure 1 illustrates Table 3 Variance decomposition Brand awareness (0100) Across Overall Appliances Automobiles Beer, wine, liquor Beverages Computers Consumer electronics Cosmetics and fragrances Credit cards Fast food Food Footwear Health and beauty Household Petrol Pharmaceutical OTC Pharmaceutical prescription Telecommunications Toys Travel 20. 117 5. 282 6. 209 10. 181 13. 435 23. 094 19. 952 18. 054 19. 568 6. 132 16. 241 20. 417 10. 36 16. 719 20. 179 13. 339 9. 393 21. 659 11. 217 16. 063 Within 3. 415 1. 334 3. 281 4. cv 2. 915 3. 843 5. 611 3. 684 3. 903 1. 660 2. 255 4. 267 1. 772 3. 896 3. 669 2. 363 5. 772 5. 604 3. 589 3. 216 Perceived quality (010) Across 0. 726 0. 323 0. 561 0. 705 0. 582 0. 850 0. 800 0. 563 0. 788 0. 361 0. 702 0. 388 0. 397 0. 561 0. 415 0. 336 0. 753 0. 452 0. 360 0. 516 Within 0. 176 0. 148 0. 141 0. 186 0. 190 0. 313 0. 167 0. 208 0. 159 0. 202 0. 134 0. 167 0. 136 0. 113 0. 116 0. 129 0. 230 0. 334 0. 127 0. 153 Advertising ($1,000,000) Across 100. 823 28. 965 54. 680 41. 713 37. 505 110. 362 105. 49 38. 446 118. 059 159. 306 15. 655 45. 791 27. 054 18. 789 27. 227 16. 325 38. 648 317. 434 61. 419 22. 136 Within 43. 625 21. 316 32. 552 12. 406 13. 372 65. 909 114. 381 20. 053 43. 415 33. 527 7. 998 7. 640 19. 075 16. 672 20. 496 9. 080 27. 919 178. 406 18. 584 10. 909 220 .025 . 2 C. R. Clark et al. .02 niggardliness . 01 . 015 0 .005 0 20 40 60 80 Mean brand awareness 100 0 30 .05 parsimony . 1 .15 20 10 0 10 20 Demeaned brand awareness 30 .8 .6 Density . 4 0 .2 0 2 4 6 Mean perceived quality 8 10 0 1. 5 1 Density 2 3 1 . 5 0 . 5 1 Demeaned perceived quality 1. 5 .015 Density . 005 . 01 0 0 00 four hundred 600 800 1000 1200 1four hundred Mean advertising expenditures ( one thousand millions of $) 0 600 cd 200 0 200 400 600 Demeaned advertising expenditures (millions of $) Fig. 1 Variance decomposition. Histogram of brand-mean of brand awareness, perceived quality, and advertising expenditures (left panels) and histogram of de-meaned brand awareness, perceived quality, and advertising expenditures (right panels) the decomposition for the overall sample. The left panels show histograms of the brand-mean of brand awareness, perceived quality, and advertising expenditures and the right panels show histograms of the de-meaned variables.Again it is evident that the time-series variation is substantial for both brand awareness and perceived quality. 5 Empirical results In Tables 4 and 5 we present a number of different estimates for the effect of advertising expenditures on brand awareness and perceived quality, .005 Density . 01 . 015 .02 .025 Effect of advertising on brand awareness and perceived quality Table 4 Brand awareness POLS Lagged brand awareness Advertising Advertising2 bare(a) effect of advertising at Mean 2fifth pctl. fiftieth pctl. seventy-fifth pctl. Adve rtising test ? 1 = ? 2 = 0 Speci? ation tests Hansen J Difference-in-Hansen J Arellano & Bond AR(2) Arellano & Bond AR(3) righteousness of ? t measures R2 -within R2 -between R2 obs brands FE DGMM SGMM 221 0. 942*** 0. 223*** 0. 679*** 0. 837*** (0. 00602) (0. 0479) (0. 109) (0. 0266) 0. 00535*** 0. 00687 0. 0152 0. 00627** (0. 00117) (0. 00443) (0. 0139) (0. 00300) ? 0. 00000409*** ? 0. 00000139 ? 0. 0000105 ? 0. 00000524** (0. 000000979) (0. 00000332) (0. 00000745) (0. 00000239) 0. 00481*** (0. 00107) 0. 00527*** (0. 00116) 0. 00514*** (0. 00113) 0. 00470*** (0. 00105) protest*** 0. 00668 (0. 00412) 0. 00684 (0. 00438) 0. 00679 (0. 00430) 0. 00664 (0. 0405) 0. 0138 (0. 0129) 0. 0150 (0. 0138) 0. 0147 (0. 0135) 0. 0136 (0. 00127) 0. 00558** (0. 00269) 0. 00617** (0. 00296) 0. 00600** (0. 00288) 0. 00544** (0. 00263) Do not resist Do not baulk discard* Do not obviate Do not defy fend** disdain** Do not jib Do not pass up 0. 494 0. 940 0. 851 1,148 317 Reject*** 0. 969 1 ,148 317 819 274 1,148 317 measurement errors in parenthesis * p = 0. 10 ** p = 0. 05 *** p = 0. 01 singly. Starting with the simplest case absent competition, we present estimates of ? , ? 1 , and ? 2 (the coef? cients on Qit? 1 or Ait? 1 and Eit? 1 and 2 Eit? 1 ) along with the borderline effect ? 1 + 2? Eit? 1 calculated at the mean and the 25th, 50th, and seventy-fifth percentiles of advertising expenditures. The POLS estimates in the ? rst column of Tables 4 and 5 suggest a signi? cant positive effect of advertising expenditures on both brand awareness and perceived quality. In both cases we also scorn the null supposition that advertising plays no role in determining brand awareness and perceived quality (? 1 = ? 2 = 0). Of course, as mentioned above, POLS accounts for neither unobserved heterogeneity nor endogeneity. In the next columns of Tables 4 and 5 we present FE, DGMM, and SGMM estimates that attend to these issues. 7 7 The stimates use at most 317 out of 348 brand s because we restrict the sample to brands with data for two historic period running but use third and higher lags of brand awareness respectively perceived quality and advertising expenditures as instruments. Different sample sizes are describe for the DGMM and SGMM estimators. Sample size is not a well-de? ned concept in SGMM since this estimator fundamentally runs on two different samples simultaneously. The xtabond2 routine in STATA reports the size of the transformed sample for DGMM and of the untransformed sample for SGMM. 222 Table 5 Perceived quality FE 0. 391*** (0. 0611) 0. 659*** (0. 204) 1. 47*** (0. 0459) 0. 981*** (0. 0431) DGMM SGMM design quality Brand awareness POLS Lagged perceived quality 0. 970*** (0. 0110) Brand awareness Advertising Advertising2 0. 000218** (0. 0000952) ? 0. 000000133 (0. 000000107) 0. 0000822 (0. 000198) 0. 0000000408 (0. 000000162) ?0. 0000195 (0. 000969) 0. 000000108 (0. 000000945) 0. 0000219 (0. 000205) 0. 0000000571 (0. 000000231) 0. 00 00649 (0. 000944) 0. 0000000807 (0. 00000308) 0. 937*** (0. 0413) 0. 00596*** (0. 00165) ? 0. 000298 (0. 000256) 0. 000000319 (0. 000000267) Marginal effect of advertising at Mean 25th pctl. 50th pctl. 75th pctl. 0. 0002** (0. 0000819) 0. 000215** (0. 000933) 0. 000211** (0. 00009) 0. 0001965** (0. 0000793) Do not discard Do not stand Reject*** Do not repudiate Do not reject Do not reject Reject** Reject** Reject*** Do not reject 0. 0000877 (0. 000180) 0. 000083 (0. 000195) 0. 0000844 (0. 000191) 0. 0000887 (0. 000177) ?5. 13e? 06 (0. 000848) ? 0. 0000174 (0. 000952) ? 0. 0000139 (0. 000922) ? 2. 32e? 06 (0. 000825) 0. 0000295 (0. 000176) 0. 0000230 (0. 000201) 0. 0000249 (0. 000194) 0. 0000310 (0. 000170) 0. 0000594 (0. 000740) 0. 0000642 (0. 000917) 0. 0000623 (0. 000847) 0. 0000588 (0. 000714) Do not reject Do not reject Do not reject Reject*** Do not reject ?0. 000256 (0. 000222) ? 0. 00292 (0. 000251) ? 0. 000282 (0. 000242) ? 0. 000248 (0. 000215) Do not reject Reject** Do not reject Reject*** Do not reject Advertising test ? 1 = ? 2 = 0 Speci? cation tests Hansen J Difference-in-Hansen J Arellano & Bond AR(2) Arellano & Bond AR(3) Goodness of ? t measures R2 -within R2 -between R2 obs brands 0. 180 0. 952 0. 909 1,148 317 819 274 1,148 317 Reject** 0. 914 1,148 317 604 178 1,148 317 C. R. Clark et al. beat errors in parenthesis. SGMM estimates in columns labelled object quality and Brand awareness * p = 0. 10 ** p = 0. 05 *** p = 0. 01 Effect of advertising on brand awareness and perceived quality 23 regardless of the class of estimator we ? nd a signi? cant positive effect of advertising expenditures on brand awareness. With the FE estimator we ? nd that the bare(a) effect of advertising on awareness at the mean is 0. 00668. It is borderline signi? cant with a p-value of 0. 105 and implies an snap bean of 0. 00638 (with a standard error of 0. 00392). A one-standard-deviation increase of advertising expenditures increase brand awareness by 0. 0408 standard deviations (with a standard error of 0. 0251). The rate of depreciation of a brands stock of awareness is estimated to be 10. 223 or 78% per year.The FE estimator identi? es the effect of advertising expenditures on brand awareness solely from the within-brand across-time variation. The problem with this estimator is that it does not deal with the endogeneity of the lagged dependent variable on the right-hand side of Eq. 2 and the potential endogeneity of advertising expenditures. We thus turn to the GMM estimators described in Section 3. We revolve around on the more ef? cient SGMM estimator. The coef? cient on the linear term in advertising expenditures is estimated to be 0. 00627 ( p-value 0. 037) and the coef? cient on the quadratic term is estimated to be ? . 00000524 ( p-value 0. 028). These estimates support the supposal that the relationship between advertising and awareness is nonlinear. The marginal effect of advertising on awareness is estimated to be 0. 00558 ( p-value 0. 038) at the mean and implies an elasticity of 0. 00533 (with a standard error of 0. 00257). A one-standard-deviation increase of advertising expenditures increases brand awareness by 0. 0340 standard deviations (with a standard error of 0. 0164). The rate of depreciation decreases substantially after correcting for endogeneity and is estimated to be 1? . 828 or 17% per year, thus indicating that an increase in a brands stock of awareness due to an increase in advertising expenditures persists for years to come. The Hansen J test for overidentifying restrictions indicates that the instruments taken together as a group are valid. guess from Section 3 that we must assume that an otiose condition holds in order for the SGMM estimator to be hold. The difference-in-Hansen J test con? rms that it does, as we cannot reject the null venture that the additional instruments for the level equations are valid.While we reject the hypothesis of no second-order serial correla tion in the error terms, we cannot reject the hypothesis of no thirdorder serial correlation. This result further validates our instrumenting strategy. However, one may still be worried about the SGMM estimates because DGMM uses a strict subset of the orthogonality conditions of SGMM and we reject the Hansen J test for the DGMM estimates (see Table 4). From a formal statistical point of view, rejecting the smaller set of orthogonality conditions in DGMM is not conclusive express that the larger set of orthogonality conditions in SGMM are invalid (Hayashi 2000, pp. 18221). In Fig. 2 we plot the marginal effect of advertising expenditures on brand awareness over the entire range of advertising expenditures for our SGMM estimates along with a histogram of advertising expenditures. For advertising expenditures between $400 million and $800 million per year the marginal effect of advertising on awareness is no womb-to-tomb signi? cantly different from zero 224 C. R. Clark et al. Margi nal effect . 004 0 . 004 0 200 400 600 800 1000 Advertising expenditures (millions of $) 1200 1400 arginal effect of advertising lower 90% potency fix . 015 upper 90% trust limit 0 0 .005 Density . 01 200 400 600 800 1000 Advertising expenditures (millions of $) 1200 1400 Fig. 2 Pointwise con? dence interval for the marginal effect of advertising expenditures on brand awareness (upper panel) and histogram of advertising expenditures (lower panel). SGMM estimates and, statistically, it is actually ostracise for very high advertising expenditures over $800 million per year. The former case covers around 1. 9% of observations and the latter less than 0. 5%.One practicable interpretation is that brands with very high current advertising expenditures are those that are already wellkn experience (perhaps because they have been heavily advertised over the years), so that advertising cannot further boost their awareness. Indeed, average awareness for observations with over $400 million in advertising expenditures is 74. 94 as compared to 69. 35 for the entire sample. number from brand awareness in Table 4 to perceived quality in Table 5, we see that the positive effect of advertising expenditures on perceived quality found by the POLS estimator disappears once unobserved eterogeneity is accounted by the FE, DGMM, and SGMM estimators. In fact, we cannot reject the null hypothesis that advertising plays no role in determining perceived quality. Figure 3 graphically illustrates the absence of an effect of advertising expenditures on perceived quality at the margin for our DGMM estimates. While the effect of advertising expenditures on perceived quality is very imprecisely estimated, it appears to be economically insigni? cant The implied elasticity is ? 0. 0000534 (with a standard error of 0. 00883) and a one-standarddeviation increase of advertising expenditures decrease perceived quality byEffect of advertising on brand awareness and perceived quality 225 Margina l effect . 001 0 . 001 0 200 400 600 800 1000 Advertising expenditures (millions of $) 1200 1400 marginal effect of advertising lower 90% confidence limit . 015 upper 90% confidence limit 0 0 Density . 005 . 01 200 400 600 800 1000 Advertising expenditures (millions of $) 1200 1400 Fig. 3 Pointwise con? dence interval for the marginal effect of advertising expenditures on perceived quality (upper panel) and histogram of advertising expenditures (lower panel). DGMM estimates 0. 000869 standard deviations (with a standard error of 0. 44). Note that the comparable effects for brand awareness are two orders of magnitude larger. Much of the remainder of this paper is concerned with demonstrating the robustness of this negative result. Before proceeding we note that whenever possible we focus on the more ef? cient SGMM estimator. Unfortunately, for perceived quality in many cases, including that in the fourth column of Table 5, the difference-in-Hansen J test rejects the null hypothesis that the extra moments in the SGMM estimator are valid. In these cases we focus on the DGMM estimator. 5. Objective and perceived quality An important component of a brands perceived quality is its impersonal quality. To the extent that mark quality remains constant, it is absorbed into the brand effects. But, even though the time frame of our sample is not very long, it is certainly possible that the intent quality of some brands has changed over the course of our sample. If so, then the lack of an effect of advertising expenditures on perceived quality may be explained if brand managers increase advertising expenditures to compensate for decreases in design 26 C. R. Clark et al. quality. To the extent that increased advertising expenditures and decreased accusative quality natural each other out, their net effect on perceived quality may be zero. The dif? culty with testing this alternative bill is that we do not have data on objective quality. We therefore overleap from the analysis those categories with brands that are likely to put up with changes in objective quality (appliances, automobiles, computers, consumer electronics, spendthrift food, footwear, pharmaceutical OTC, telecommunications, toys, and travel).The resulting estimates are account in Table 5 under the impetus Objective quality. We still ? nd no effect of advertising expenditures on perceived quality. 8 5. 2 Variation in perceived quality Another possible reason for the lack of an effect of advertising expenditures on perceived quality is that perceived quality may not vary much over time. This is not the case in our data. Indeed, the standard deviation of the year-to-year changes in perceived quality is 0. 2154. Even for those products whose objective quality does not change over time there are important changes in perceived quality (standard deviation 0. 130). For example, consider bottled water where we expect picayune change in objective quality over time, both within and across brands. Nonetheless, there is considerable variation in perceived quality. The perceived quality of cobalt blue? na piss ranges across years from 6. 33 to 6. 90 and that of Poland Spring Water from 5. 91 to 6. 43, so the homogeneous of over two standard deviations. Across the brands of bottled water the range is from 5. 88 to 6. 90, or the equivalent of over four standard deviations. Further proof of variation in perceived quality is provided by the automobiles category.Here we have obtained measures of objective quality from Consumer Reports that rate vehicles based on their performance, comfort, convenience, safety, and fuel economy. We can ? nd examples of brands whose objective quality does not change at least for a number of years while their perceived quality ? uctuates considerably. For example, arouse Silverados objective quality does not change between 2000 and 2002, but its perceived quality increases from 6. 08 to 6. 71 over these three years. Similarly, GMC Sierras objective quality does not change between 2001 and 2003, but its perceived quality decreases from 6. 72 to 6. 26. The ? al piece of evidence that we have to offer is the variance decomposition from Section 4 (see again Table 3 and Fig. 1). Recall that the acrossbrands standard deviation of brand awareness is about six times larger than the within-brand standard deviation. In case of perceived quality the ratio is about 4. Hence, while there is more cross-sectional than time-series variation in our sample, the time-series variation is substantial for both brand aware- 8 The marginal effects are calculated at the mean, 25th, 50th, and 75th percentile for advertising for the brands in the categories judged to be stable in terms of objective quality over time.Effect of advertising on brand awareness and perceived quality 227 ness and perceived quality. Also recant from Section 4 that perceived quality with an intertemporal correlation of 0. 95 is somewhat less persistent than b rand awareness with an intertemporal correlation of 0. 98. Given that we are able to detect an effect of advertising expenditures on brand awareness, it seems unlikely that insuf? cient variation within brands can explain the lack of an effect of advertising expenditures on perceived quality instead, our results suggest that the variation in perceived quality is unrelated to advertising expenditures.The question then becomes what besides advertising may puzzle these changes in perceived quality. There are numerous possibilities, including consumer learning and word-of-mouth effects. Unfortunately, give the data available to us, we cannot further explore these possibilities. 5. 3 Brand awareness and perceived quality Another concern is that consumers may confound awareness and preference. That is, consumers may just now prefer more familiar brands over less familiar ones (see Zajonc 1968). To address this issue we proxy for consumers familiarity by adding brand awareness to the r egression for perceived quality.The resulting estimates are reported in Table 5 under the heading Brand awareness. While there is a signi? cant positive relationship between brand awareness and perceived quality, there is still no evidence of a signi? cant positive effect of advertising expenditures on perceived quality. 5. 4 Competitive effects Advertising takes place in a competitive environment. almost of the industries being studied here are indeed oligopolies, which suggests that strategic considerations may in? uence advertising decisions.We next allow a brands stocks of awareness and perceived quality to be affected by the advertising of its competitors as discussed in Section 2. 9 Competitors advertising, in turn, can enter our estimation Eqs. 1 and 2 either relative in the share-of-voice speci? cation or absolute in the total-advertising speci? cation. We report the resulting estimates in Table 6. Somewhat surprisingly, the share-of-voice speci? cation yields an insignifi cant effect of own advertising. We conclude that the share-of-voice speci? cation is simply not an appropriate functional form in our application. The total-advertising speci? ation quickly con? rms our main ? ndings presented above that own advertising affects brand awareness but not perceived quality. This is true even if we allow competitors advertising to enter quadratically in 9 For this analysis we take the subcategory rather than the category as the relevant competitive environment. Consider for instance the beer, wine, liquor category. There is no reason to expect the advertising expenditures of beer brands to affect the perceived quality or awareness of liquor brands. We drop any subcategory in any year where there is just one brand due to the lack of competitors.Table 6 Competitive effects Perceived quality 0. 845*** (0. 0217) 0. 356** (0. 145) Total advertising Brand awareness Perceived quality 228 Share of voice Brand awareness Lagged awareness/quality Relative adverti sing (Relative advertising)2 0. 872*** (0. 0348) 0. 236 (0. 170) ? 0. 00912 (0. 0104) 1. 068*** (0. 0406) 0. 0168 (0. 0164) ? 0. 00102 (0. 00132) Advertising Advertising2 Competitors advertising 0. 00892** (0. 00387) ? 0. 00000602** (0. 00000248) ? 0. 00609* (0. 00363) ?0. 0000180 (0. 000592) ? 0. 0000000303 (0. 000000535) 0. 00128** (0. 000515) Marginal effect of advertising at Mean 5th pctl. 50th pctl. 75th pctl. 0. 00333 (0. 00239) 0. 0164 (0. 01218) 0. 00624 (0. 00448) 0. 00264 (0. 00190) Do not reject Reject* Do not reject Reject*** Do not reject 1,147 317 0. 000225 (0. 000218) 0. 00113 (0. 00110) 0. 00429 (0. 000416) 0. 000179 (0. 000173) 0. 00812** (0. 00355) 0. 00881** (0. 00382) 0. 00861** (0. 00375) 0. 00797** (0. 00349) Reject** Do not reject Do not reject Reject** Do not reject 1,147 317 ?0. 000140 (0. 000524) ? 0. 0000174 (0. 000582) ? 0. 0000164 (0. 000565) ? 0. 0000132 (0. 000510) Do not reject Do not reject Reject*** Do not reject 1,147 317 C. R. Clark et al.Advertis ing test ? 1 = ? 2 = 0 Speci? cation tests Hansen J Difference-in-Hansen J Arellano & Bond AR(2) Arellano & Bond AR(3) obs brands Do not reject Do not reject Do not reject Reject** Do not reject 1,147 317 Standard errors in parenthesis. DGMM estimates in column labeled Total advertising/perceived quality and SGMM estimates otherwise * p = 0. 10 ** p = 0. 05 *** p = 0. 01 Effect of advertising on brand awareness and perceived quality 229 addition to linearly. Competitors advertising has a signi? cant negative effect on brand awareness and a signi? cant positive effect on perceived quality.Repeating the analysis using the sum instead of the average of competitors advertising yields largely similar results except that the share-of-voice speci? cation yields a signi? cant negative effect of advertising on brand awareness, thereby reinforcing our conclusion that this is not an appropriate functional form. 10 Overall, the inclusion of competitors advertising does not seem to in? uence o ur results about the role of own advertising on brand awareness and perceived quality. This justi? es our focus on the simple model without competition. Moreover, it suggests that the next alternative explanation for our main ? dings presented above is unlikely. Suppose awareness depended positively on the total amount of advertising in the brands subcategory or category while perceived quality depended positively on the brands own advertising but negatively on competitors advertising. Then the results from the simple model without competition could be driven by an omitted variables problem If the brands own advertising is highly correlated with competitors advertising, then we would overdo the impact of advertising on awareness and understate the impact on perceived quality.In fact, we might ? nd no impact of advertising on perceived quality at all if the brands own advertising and competitors advertising cancel each other out. 5. 5 Category-speci? c effects Perhaps the ideal dat a for analyzing the effect of advertising are time series of advertising expenditures, brand awareness, and perceived quality for the brands being studied. With long plenteous time series we could then try to identify for each brand in isolation the effect of advertising expenditures on brand awareness and perceived quality.Since such time series are unfortunately not available, we have focused so far on the aggregate effect of advertising expenditures on brand awareness and perceived quality, i. e. , we have constrained the slope parameters in Eqs. 1 and 2 that determine the effect of advertising to be the same across brands. Similarly, we have constrained the carryover parameters in Eqs. 1 and 2 that determine the effect of lagged perceived quality and brand awareness respectively to be the same across brands. As a compromise between the two extremes of brands in isolation versus all brands aggregated, we ? st examine the effect of advertising in different categories. This adds some cross-sectional variation across the brands within a 10 We caution the reader against reading too much into these results The number and identity of the brands within a subcategory or category varies sometimes widely from year to year in the Brandweek Superbrands surveys. Thus, the sum of competitors advertising is an extremely vapourisable measure of the competitive environment. Moreover, the number of brands varies from 3 for some subcategories to 10 for others, thus make the sum of competitors advertising dif? ult to compare across subcategories. 230 Table 7 Category-speci? c effects Brand awareness Marginal effect Carryover rate Appliances Automobiles Beer, wine, liquor Beverages Computers Consumer electronics Cosmetics and fragrances Credit cards Fast food Food Footwear Health and beauty Household Petrol Pharmaceutical OTC Pharmaceutical prescription Telecommunications Toys Travel 0. 0233 (0. 0167) 0. 00526 (0. 0154) ? 0. 0264 (0. 0423) ? 0. 0245 (0. 0554) 0. 0193** (0. 00777) 0. 0210** (0. 0