American Marketing Association

The Long-Term Impact of Promotion and Advertising on Consumer Brand Choice
Author(s): Carl F. Mela, Sunil Gupta and Donald R. Lehmann

Source: Journal of Marketing Research, Vol. 34, No. 2 (May, 1997), pp. 248-261
Published by: American Marketing Association
Stable URL: http://www.jstor.org/stable/3151862 .
Accessed: 04/12/2014 01:36
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
American Marketing Association is collaborating with JSTOR to digitize, preserve and extend access to
Journal of Marketing Research.
http://www.jstor.org
This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

All use subject to JSTOR Terms and Conditions
CARLF. MELA,SUNILGUPTA,and DONALDR. LEHMANN*
The authors examine the long-termeffects of promotionand advertis-

ing on consumers' brand choice behavior. They use 8 1/4 years of panel
data for a frequently purchased packaged good to address two questions:
(1) Do consumers' responses to marketing mix variables, such as price,
change over a long period of time? (2) If yes, are these changes associ-
ated with changes in manufacturers' advertising and retailers' promotional
policies? Using these results, the authors draw implications for manufac-
turers' pricing, advertising, and promotion policies. The authors use a two-
stage approach, which permits them to assess the medium-term (quar-
terly) effects of advertising and promotion as well as their long-term (i.e.,
over an infinite horizon) effects. Their results are consistent with the
hypotheses that consumers become more price and promotion sensitive
over time because of reduced advertising and increased promotions.
The Long-Term Impact of Promotion and

Advertising on Consumer Brand Choice
Tradepromotionsaccountfor 50% of the $70 billion bud- However,supportfor value pricing strategyis not univer-
get of consumer packaged goods manufacturers(Progres- sal. There is an uncertaintyin both industryand academia
sive Grocer 1995). Almost 70% of the firms have increased aboutthe long-termimpactof promotionand advertisingon
their trade promotions between 1990 and 1995. Forbes brand performance.Some companies such as Heinz have
(1991) also reportsan increase in consumer and trade pro- emphasized their continued reliance on promotionsas their
motional spending for consumer packagedgoods manufac- main marketingvehicle (Wall Street Journal 1992). Others
turers-from 50% to 75% of marketing budgets during cite such cases as that of Hi-C, which tried to switch away
1985-90. It suggests that this increase induced a loss of from promotionin the early 1980s and lost significant share
8-15% in the share of the top three brandsin categories as (Business Week1992). The empiricalevidence to date is lim-
diverse as popcom, dishwashingdetergent,cat food, barbe- ited and mixed. Although many previous studies examine
cue sauce, and prepareddishes. The article goes on to cite the short- and long-termeffects of advertising(e.g., Clarke
several marketingmanagerswho blame couponing and pro- 1976), few focus on the long-termeffects of promotions.As
motions for reducedbrandloyalty. Business Week(1991) al- Blattbergand Neslin (1989, p. 93) note, "Almostno research
so contends that a shift of marketingdollars from advertis- has been conductedon the long termeffects of promotions.
ing to promotionsis to be blamed for the drop in the num- Yet, it is critical to a brand'sstrategy."In summary,under-
ber of consumers that buy only well-known brands from standing whether promotionand advertisinghurt or help a
77% in 1975 to 62% in 1990. brandin the long run appearsto be relevant,important,and
In light of this, some companies now believe that promo- underresearched.
tions have made consumersmore price sensitive, which con- We empirically examine the long-termeffects of promo-
sequently has lowered the effective price companies can tion and advertisingon consumers' brandchoice decisions
charge(BrandWeek1993). As a resultof this belief, Colgate- in mature product categories (i.e., we do not focus on
Palmolive, Ralston Purina, Quaker Oats, and Procter & changes due to the productlife cycle factors). Specifically,
Gamble (P&G) recently have curtailed the frequency of we address the following two questions: (1) Does con-
their price promotions(WallStreetJournal 1996). sumers' responsiveness to marketingmix elements change
over time? For example, are consumers becoming more
price sensitive over time? Is the price-sensitivesegment of
*Carl F. Mela is an assistantprofessor,College of Business Administra-
consumers growing over time? (2) If such changes as in-
tion, University of Notre Dame. Sunil Gupta is a professor, Graduate creasing price sensitivity occur, what factors affect these
School of Business, ColumbiaUniversity.Donald R. Lehmannis George E. changes? For example, is the reduction in advertising ex-
WarrenProfessor,GraduateSchool of Business, Columbia University.The pendituresand/or increase in promotions influencing con-
authorsthankInformationResources, Inc. and an anonymouscompany for sumers' price sensitivity to the product?
providing the data and managerialinput for this study, and the editor and
three anonymousJMR reviewers for their valuablecomments. Investigationof these issues not only will enable us to
betterunderstandthe changes in consumerbehaviorover the
Journal of MarketingResearch
Vol. XXXIV (May 1997), 248-261 248

The Long-Term Impact of Promotion and Advertising 249
long run,but also will provideuseful marketingimplications tivity (Comanorand Wilson 1974). The second theory states
for manufacturers'pricing,advertising,and promotionpoli- that advertising increases competition by providing infor-
cies. To ensure consistent interpretation of long term mation to consumers that is likely to make consumers more
throughoutthis article, we begin by defining it. price sensitive (Nelson 1974). Mitraand Lynch (1995) rec-
oncile these conflicting theories by suggesting that advertis-
WhatIs Long Term? ing affects price elasticity throughits influence on two mod-
Similar to Faderand colleagues (1992), we characterize erating constructs:Advertisingcan increase price elasticity
the effect of marketingactions on consumers'choice behav- by increasing the numberof brandsconsidered; it also can
ior as follows: decrease price elasticity by increasing the relative strength
of brandpreference.
1. Short-termeffects:Thisis the immediate(e.g., weekly)effect
of promotionor advertising on the salesor shareof a brand. Kaul and Wittink (1995) provide an excellent review of
Mostrecentstudiesfocuson theseeffects.Thesestudiesgen- the studies examiningthe impactof advertisingon price sen-
erallyfind very strongshort-termeffects of promotionbut sitivity.They also suggest that the apparentlyconflicting re-
very weak or insignificanteffects of advertisingon brand sults of many studies might be indicativeof the differentas-
share(e.g., Guadagniand Little 1983;Gupta 1988;Tellis pects of the problemon which the researchersinadvertently
1988a). focus. For example, Eskin and Baron (1977) study new
2. Medium-termeffects: Some studies attempt to go beyond products,for which more advertisingmight attractaddition-
week-by-weekeffectsof promotions.Most of these studies al consumers who are more price sensitive, and Krishna-
use a 4- to 16-weektimeframe.Forexample,Davis,Inman, murthiand Raj (1985) analyze a constantset of households.
andMcAlister(1992)studytheimpactof promotions on con- Researchersseem to agree that if the advertisingmessage is
sumerattitudes;Ehrenberg,Hammond, andGoodhardt (1994)
examinepromotioneffects on brandshares;and Kim and non-price-orientedand brand building, then it lowers con-
Lehmann(1993) study changes in consumersensitivities due sumers' price sensitivity, whereas a price-orientedmessage
to promotions.Although many of these studies draw implica- could increaseconsumers' price sensitivity.
tions about the long-term effects of promotions, we believe In our application, we use a product category in which
that effects of advertisingand promotionin the following 4- five of the nine brands advertise, all using a non-price ad-
to- 16 weeks qualify as medium-termratherthan long-termef- vertising message. Furthermore,the price of these five
fects. Here, we define the impact of a currentperiod's adver- brandsis, in general, higher than the price of the remaining
tising/promotionon sales, share,or consumers'sensitivities of four brands that do not advertise. According to Boulding,
the subsequent 13 weeks (or quarter)as the medium-termef- Lee, and Staelin (1994), who use price as a proxy for the ad-
fect of advertising/promotion.
3. Long-termeffects: Previousstudies find that advertisinghas a vertisingmessage of a brand,firmscharginghigher-than-av-
substantialcarryovereffect. The long-termeffect of advertis- erage price have a naturalinclination to have a non-price
ing (or promotion) is the cumulative effect on consumers' message in their advertising.Therefore,
brand choice behavior, lasting over several years. One ap- HI:Non-price-orientedadvertisingis expectedto reducecon-
proach to capturethe long-term impact of advertising is the sumers'pricesensitivityin the longrun.
distributedlag model (e.g., Clarke 1976). We use a similarap-
proach here. Few studies, even those using disaggregatedata, involve
4. Effects of changes in marketingstrategy: P&G's move from conductinganalysisat the segmentlevel. A notableexception
high-low pricingto everydaylow pricingis a good example of is Krishnamurthi and Raj (1985), who find thatthe impactof
a change in marketingstrategy.An importanttask for market- advertisingon consumers'price sensitivityis strongerfor the
ing managersis to assess the impactof these strategychanges more price-sensitive(or nonloyal) segment, and this effect is
on brandshare or sales. Notice the distinction between long-
term effects and effects of a policy change. If a company marginalfor the less price-sensitive(or relativelyloyal) seg-
ment. This suggests a "floor"effect: If the loyal segment is
changes its advertisingin one periodonly and evaluatesits cu-
mulativeeffect in futureperiods, it is measuringlong-termef- completely price insensitive, increasedadvertisingwill have
fects of advertising.Conversely, if the company cuts its ad- no effect on this segment's price sensitivity.On the basis of
vertising expenditurein half in all periods and studies its ef- these findings, we expect the following:
fect on consumerchoice, it is evaluatingthe effect of strategy
H2:The long-termimpactof advertisingon consumers'price
change.
sensitivityis likelyto be strongerfor nonloyalconsumers
Our focus here is to study the long-term effects of pro- thanforrelativelyloyal(andless price-sensitive)
consumers.
motions and advertisingon consumers'price and promotion Although several studies address the issue of how adver-
sensitivities. We proceed as follows: We start with a brief tising affects price sensitivity, almost none focus on the
review of the relevant literature and development of effect of advertising on consumers' sensitivity to promo-
hypotheses. We then describe our modeling approach.This tions. To understandadvertising'seffect on consumers' pro-
is followed by data, variables,and results sections. We dis- motion sensitivity,we believe thatpromotionsshould be cat-
cuss managerialimplications of these results and conclude egorized into two groups: price-orientedand non-price-ori-
with the contributionsand limitationsof this study. ented promotions.1Advertising that makes consumers less
REVIEWAND HYPOTHESES
LITERATURE price sensitive also should make these consumers less sensi-
tive to price-oriented promotions that primarily focus on
Long-TermEffectsof Advertising price cues. Conversely, advertising should increase con-
Economists have developed two theories that predict sumers' response to non-price-oriented promotions that
opposite effects of advertisingon consumers'price sensitiv-
ity. The first theory suggests that advertisingleads to prod- IWe use the wordpromotions for all sales and trade promotionssuch as
uct differentiation,therebyreducingconsumers'price sensi- temporaryprice reductions,feature,coupon, display, and so on.

250 JOURNAL OF MARKETINGRESEARCH, MAY 1997
enhance the awareness and visibility of a brand without PIMS data at the business unit level to conclude that adver-
focusing on price. tising decreases and promotion increases consumers' price
H3:Inthe longrun,advertisingwill reduceconsumers' sensitiv- sensitivity for large brands.
ity to price-oriented
promotions. Consumer behavior theories, empirical studies, and the
H4:In the longrun,advertisingwill increaseconsumers'sensi- previous discussion about advertisingsuggest that over the
tivityto non-price-oriented
promotions. long run, price-orientedpromotions will make consumers
H5:Ingeneral,theseeffectsarelikelyto be strongerforthenon- more price sensitive by focusing their attention on price
loyalsegmentthanfor theloyalsegment. cues. Non-price-oriented cues therefore will become less
importantto consumers.The opposite is likely to hold with
As advertising is expected to reduce consumers' price
increasingnon-price-orientedpromotions,which are likely
sensitivity and build loyalty, it is reasonableto expect that it to work like advertisements(consistent with the empirical
also will reduce the size of the nonloyal segment.
generalizationsof Kaul and Wittink 1995). Therefore,
H6:Advertisingwill reduce the size of the nonloyal segment.
H7:Overthe long run,price-oriented promotionswill (a) in-
Long-TermEffectsof Promotion creaseconsumers'pricesensitivity,(b) increaseconsumers'
Several theories suggest a negative long-term effect of sensitivityto price-orientedpromotions,and (c) decrease
consumers'sensitivityto non-price-orientedpromotions.
promotion on consumers' attitude and behavior. Self-per- H8:Overthelongrun,non-price-oriented promotions will(a)de-
ception theory implies that consumers who buy on promo- creaseconsumers'pricesensitivity,(b) decreaseconsumers'
tions are likely to attributetheir behaviorto the presence of sensitivityto price-oriented
promotions,and(c) increasecon-
promotions and not to their personal preference for the sumers'sensitivityto non-price-oriented
promotions.
brand(Dodson, Tybout,and Stemthal 1978). Increasedpro-
motion in a category is likely to lead to a perceptionthatthe Are results likely to be the same across the nonloyal and
key differentiatingfeatureof brandsis the price (Sawyerand loyal segments? Some studies suggest that promotion sig-
Dickson 1983). This simplifying heuristic then can lead to nals that do not carry price information(e.g., displays) can
increasedreliance on sales promotionsfor choosing brands. have a dual effect. According to Inman, McAlister, and
Operantconditioning principles also suggest that frequent Hoyer (1990), these promotion signals have a positive
dealing in a category can condition consumers to look for impact on the choice behaviorof "low need for cognition"
promotions in the future, thereby making them more pro- people. However,high-need-for-cognitionpeople react to a
motion prone. promotionsignal only when it is accompaniedby a substan-
There are some theories that predictan opposite effect of tive price reduction. By definition, loyal consumers are
more habitualbuyers and respond less to price and promo-
promotion.For example, learningtheory suggests that pro-
motions can help a brandthroughincreasedfamiliarityand tions. Therefore,it is reasonableto expect that, like the low
need for cognition consumersexamined by Inman,McAlis-
experience. However,this effect is likely to be small for ma-
ture and stable productcategories (like the one used in our ter, and Hoyer (1990), loyal consumers will be less moti-
vated to process non-price-orientedpromotioninformation
study), in which consumers have been familiarwith almost
all of the brandsfor a long period of time. actively. This suggests that non-price-orientedpromotions
are likely to divert the attention of loyal consumers away
Empiricalresearchin the area of promotionstypically fo-
cuses on identifyingthe short-termeffects of promotions(e.g., from price but in fact might make the nonloyal consumers
focus on price even more. Inmanand McAlister(1993) sup-
Guadagniand Little 1983; Gupta 1988; Kamakuraand Rus-
sell 1989). These studies show that promotionshave a large port this view by suggesting that some consumers think of
short-termeffect on consumers' brand choice. Few recent promotion signals as price-orientedpromotions (probably
studiesexaminethe medium-termeffect (over4-16 weeks) of the nonloyals), whereas others think of them as a non-price
promotionon brandshare,brandevaluations,andconsumers' promotion(probablythe loyals). Therefore,
price sensitivity.Using datafrom four weeks before and four H9:Non-price-oriented promotionswill decreasethepricesen-
weeks after major promotions, Ehrenberg,Hammond, and sitivityof loyalconsumersandincreasethepricesensitivity
Goodhardt(1994) concludethatconsumerpromotionsfor es- of nonloyalconsumers.
tablished brandshave no noticeable effect on either subse- Consistent with our previous discussion, we expect price
quent sales or brand loyalty. Davis, Inman, and McAlister promotions to reduce consumers' loyalty by making them
(1992) use a controlledexperimentover threemonthsto con- more sensitive to price and price promotions.Thereforewe
clude thatpromotionshave no negativeeffect on brandevalu-
expect these promotionsto increase the size of the nonloyal
ations. However,these studiesdo not addresswhetherpromo-
segment. However,the effect of non-price-orientedpromo-
tions make consumersmore sensitive to short-term(weekly) tions on consumers' brand loyalty and price sensitivity is
price and promotionactivities.Kim and Lehmann(1993) ex- expected to be mixed (depending on the loyal or the non-
amine this issue using a four-monthtime frame and suggest
loyal segment), which makes it hard to assess their impact
thatpromotionsdo makeconsumersmoreprice sensitive. on segment sizes. Therefore,
Two studies discuss the long-term effect of advertising
and promotions.Johnson (1984) analyzes 20 productcate- HIO:Price-orientedpromotions are likely to increase nonloyal
gories to examine changes in brandloyalty over the period segment size.
1975-83. He finds no significantchanges in a brand'sshare These hypotheses are summarizedin Table I.2
of requirements. However, if promotions have increased
over time, then it is difficult to say if a brand'sshare is high 2Price sensitivity is generally negative, indicating that higher prices
because of consumer loyalty for the brand or increased reducebrandchoice probability.Thereforea positive sign in this table indi-
brand promotions. Boulding, Lee, and Staelin (1994) use cates reductionin price sensitivity.

Table 1 s
OF HYPOTHESES
SUMMARY (3) Pith = XsPih
s= I
Impacton
ConsumerSensitivityto where
Nonloyal
Long-Term Price Non-price Segment ;s = shareof the sth segment, 0 < cs < 1, l nis= I and
Impactof Price Promotions Promotions Size ph = probabilitythat household h in segment s buys brandi at
Advertising + - + - time t.
Price Promotions - + - +
Non-price Promotions +/-* - + ? The segment-levelchoice probabilityis a logit model:
*This effect is expected to be positive for loyal consumers and negative ph - exp(pOiqs + PqsXh)
for nonloyal consumers. (4) ist
exp(po iqs + PqsXt )
where POjqsand Pqs are segment-specific parametersesti-

MODELINGAPPROACH mated for each quarterq.
We begin by briefly describing the multinomial logit Althoughthe marketingresponseparameters,3 s, are not
model we used to capture the impact of short-term (weekly)
brand specific within a segment, the net effect o? the mar-
keting activity of a brandis a convex combination of seg-
price and promotion activities on consumers' brand choice
behavior. We then extend this model to a segment-level logit
ment-level response parameters.The weights in this convex
combinationdependon within-segmentbrandshares,which
model, which allows for consumer heterogeneity in
response parameters (Kamakura and Russell 1989). Next,
typically differ substantially across brands and segments
we suggest that the slope or first derivative of the logit (Kamakuraand Russell 1989). This approach allows for
model with respect to, for example, price is a better measure brand-specificresponses and consumer heterogeneitywith-
of consumers' sensitivity to short-term marketing activities
out making the numberof model parametersvery large. In
than are the response parameters or elasticities. This is fol-
our application,estimatingbrand-specificresponse parame-
lowed by a discussion of the distributed lag model used to
ters does not provide betterfit or insights.
When the segment-level logit parametersand segment
capture changes in consumers' sensitivities over time as a
function of quarterly marketing activities such as advertis-
sizes are estimated, we can obtain a household's posterior
ing. Finally, we discuss estimation issues.
probability(whs)of belonging to a segment s in quarterq by
Bayes' rule:
Modeling Consumers'Brand Choice Behavior h qs
EsqsLhs
-
(5)
Single-segment logit. The probability that household h s qs qs
buys brand i at occasion t is captured by the multinomial
logit model (Guadagni and Little 1983): where Lhsis the likelihood of household h's purchase history
given that it belongs to segment s in quarter q.
plh - exp(V,h )
(1) "
exp(Vt ) A Measure of Consumers'Sensitivities to Short-Term
MarketingActivities
where the deterministicpartof the utility is The segment specific parameters, [3, represent consumers'
(2) V,h= po + pXh. response to weekly price and promotion activities of brands.
We could use these parameters, which vary over time (quar-
The vector Xh includes short-term (weekly) marketing ters), to capturechanges in consumers'sensitivities to short-
variables such as price and promotion, as well as household- term marketing activities. However, logit models pose a spe-
specific variablessuch as brandloyalty.The parameterPoiis cial problem because the logit parameters are identified only
a brand-specific constant, and P3is a vector of parameters for up to a scale constant (Ben-Akiva and Lerman 1985). This
variables Xht.One of our key objectives here is to test explic- scale constant is inversely proportional to the variance of the
itly whether consumers' sensitivities to short-term market- error in the utility function. Therefore, comparison of ,P
ing activities (which are a function of P parameters) change parameters across time or models is not desirable because
over time and if these changes are correlated with marketing the comparison is confounded by the error variances (Swait
activity such as advertising over the long run. Because some and Louviere 1993).
long-term marketing variables are reported (at least in our It is possible to use elasticities instead of model parame-
data set) on a quarterly basis, we estimate P for each quarter ters as measures of consumers' sensitivities to short-term
q. This is consistent with previous research (e.g., Moore and marketing variables. Unlike the elasticities in multiplicative
Winer 1987) and the managerial judgment of the company models (Krishnamurthi and Raj 1985), the elasticities in log-
that provided us the data. (In the Estimation section, we pro- it models are scaled by brand choice probabilities. This scal-
vide more details about the choice of a quarter as the time ing implies that low-share brands will tend to have high
period of analysis.) elasticities, though consumers are not necessarily more re-
Multiple-segmentlogit. We allow for consumer hetero- sponsive to the marketing actions of small-share brands
geneity in response parameters by using segment-level logit (Guadagni and Little 1983). Scaling also introduces asym-
models. Instead of specifying these segments a priori, we metry into cross-elasticities, even though there might be no
use the latent class or mixture modeling approach advocat- underlying asymmetry in consumer response (Russell,
ed by Kamakura and Russell (1989): Bucklin, and Srinivasan 1993).

252 JOURNAL OF MARKETINGRESEARCH, MAY1997
We thereforeuse firstderivativesor slopes as our measure Yi(q- I)sx= last quarter'sresponse value, that is, lag term;
of consumers' sensitivities. The first derivative represents Ziq= vector of quarterlymarketingvariables,such as
change in brandsharefor one unit change in an independent advertising,for brandi, quarterq;
variable,such as price. In the logit model, for segment s the Cq= controlvariables,such as economy,for quarterq;
first derivativeof marketshare of brandi in quarterq with ais, Xs, Yis, s = parametersto be estimated;and
respect to one unit change in an independentvariableXi is ?iqs= errorterm.
given by This model has several important features. First, the
model parameters(y) representthe medium-term(quarterly)
?htPqsxWhsPihst(1 - Pih )
(6) Yiqsx=- Ehx
zt wqS impact of advertising and promotionon consumers' sensi-
tivities. Second, the decay parameter(k) indicateshow long
For discrete (0,1) independentvariables,such as promo- the effect of advertisingand promotionon consumers'sen-
tion, the discrete analog of the first derivativeis sitivities last. Third, the cumulative long-termeffects (over
infinite time horizon) of marketingactivities, such as adver-
hzt WhsAP,ht tising, are capturedby 0 = y/(1 - A.).When the estimatesand
(7) Yiqsx -
h Yh
zhtWqs
Y' the variance-covarianceof y and X are obtained,the variance
of the long-term effect 0 can be estimated using Cramer's
where APihtis the change in purchase probability due to theorem,which enables us to test for the significanceof 04:
change in the Xi variablefrom 0 to 1 (e.g., from no promo- 2y
tion to promotionscenario). (9) Var(0y)= (9) + Cov(., y)
Note that there are three importantfeatures of this mea- (1I- I) 2 (- k)3
sure. First,it does not sufferfrom the scaling problemsmen-
tioned previously. Second, for a given value of Pqsx,this + Var(X).
(I- -)3
measureis the highest when Pht = .5. This is intuitivelyap-
pealing because it suggests that consumershave the highest An implicit assumptionof the model in Equation8 is that
responseto price and promotionwhen they have the highest the decay parameterk is identical for all quarterlymarket-
level of uncertaintyof whetherto choose a particularbrand. ing variables such as advertising and promotion.Although
Third, even if all brands have the same [Pqsxparameters, this makes the model parsimonious,it potentially is a very
brandswill have different responses to their marketingac- restrictive assumption. In other words, if the long-term
tivities as measuredby the first derivatives. effects of advertisinglast for severalyears but the long-term
In summary,in the first stage of our model, we estimate effects of promotions last for only a few quarters,then this
segment-level logits and then compute first derivativeswith assumption can lead to erroneous conclusions. Following
respect to price, price-orientedpromotion, and non-price- Johnston (1984, p. 347), we test whetheradvertising,price
orientedpromotionfor each quarterof the data. promotion, and non-price promotion have different lag
structures.
Modeling the Long-TermImpactof Advertisingand
Promotion Estimation
The second stage of our model captures the impact of The model is estimated in two stages. In stage one, we
quarterlymarketingvariables,such as advertising,on quar- estimate the segment-level logit model by maximumlikeli-
terly sensitivity (first derivative)estimates using a distrib- hood (Kamakuraand Russell 1989). From this stage of the
uted lag model. These models (e.g., the Koyck model and analysis we obtain parameterestimates for each quarter.We
the partialadjustmentmodel) have been used extensively in chose a quarteras the appropriatewindow or intervallength
marketing to capture the carryover effects of advertising for several reasons. First, we have 8V4years of data, which
(e.g., Clarke 1976). The Koyck model, for example, makes suggests that a larger window (e.g., 1 year) will leave us
the assumption that the advertisingeffect has a geometric with few observationsfor the second stage of the analysis;
decay over time. The partialadjustmentmodel assumes that second, because the median interpurchasetime in our appli-
in the short run consumers adjust their behavior only par- cation is approximatelysix weeks, a shorterwindow (e.g., a
tially to changes in the environment.For example, frequent month) will leave us with few observations and unstable
promotionscan lead consumersto expect promotionsin the coefficients in the first stage of the analysis; third,informa-
future,therebymakingthem more price sensitive. However, tion on some of the marketingvariables(e.g., advertising)is
this shift in price sensitivity is likely to be gradual,as con- available on a quarterlybasis; and fourth, the managersof
sumersslowly adjusttheir behaviorto the new marketenvi-
ronment.These assumptionslead to the following model3:
(8) Yiqsx
= oas + izsYi(q-l)sx +
YisZiq + + ?iqs,
0sCq 4Var(O)= g'ZTwhere g is the vector of first derivativesof the functiong
thatdefines the relationshipbetween 0 and y, x (i.e., 0 = g(y, X) = y/( 1 - .)),
where and I is the variance-covariancematrix for the parameterestimates y and
X. This implies that
Yiqsx= consumer response in quarterq for brand i in
segment s, with respectto short-termmarketing
variableX, such as price;
Var(0) = ay [Var(y) Cov(y, X)1 a0 a0
0- LCov(y, X) Var(X) J -ay a
3Both the Koyck and the partial adjustmentmodels' formulations are
similarto Equation8. However,in the Koyck model, errorsare serially cor- _A3_
related, whereas these errors are independent in the partial adjustment
model (Johnston 1984). We test for errorcorrelationin our estimation. Appropriatesubstitutionthen leads to Equation9.

the company thatprovidedus the data believe that most sig- dent variables(such as price and promotion)as per Equation
nificant marketingdecisions are made on a quarterlybasis. 6 or 7. We then use these first derivativesas dependentvari-
Our initial attemptat estimatingsegment-level logit mod- ables in Equation 8, which is estimated by ordinary least
els on quarterlydata indicatesthateven a quarteris too short squares.
a period to obtain stable parameterestimates. Given this
dilemma (i.e., a desire to have quarterlyparameterestimates DATA
so that we have enough observations for the second stage The data and managerialinput for our application were
analysis, while needing a longer time intervalto obtain more provided by a major consumer packaged goods company
stable parameter estimates in the first stage), we use a and Information Resources, Inc. The data contain panel,
rolling three-quarterwindow for estimating the segment- store, and demographic informationfor a consumer non-
level logit. Specifically, we assume the parameterestimates durablecategory in one marketarea for 8 V4years, from Jan-
for data covering quarters1, 2, and 3 to representquarter2, uary 1984 to March 1992. The sponsoring company also
estimates from quarters2, 3, and 4 to representquarter3, provided quarterlyadvertisingexpendituresfor all brands.
and so on. Although this procedurecould smooth out some Because scannerdata began to proliferateapproximatelyten
of the quarterly fluctuations in coefficients, it also will years ago, it is only now that researcherscan use this infor-
smooth out randomerror.In the spirit of moving averages, mation to make long-term inferences regardingchanges in
this procedurecaptures the overall trend in coefficients in consumer behavior.Our study is one of the first to utilize
the 8V4years of data. More important,it provides us with this long a period of disaggregatedata.
more stable parameterestimates, which are essential for We cannot reveal the productcategory or the brands.The
drawingmeaningfulconclusions.5 category is a household nonfood product.The median inter-
When we obtain quarterlyestimates of [ parameters,we purchase time is 6 weeks, and the mean is 12 weeks. The
then can compute first derivativeswith respect to indepen- category has several characteristicsthat make it particularly
suitable for our application.First, like most other consumer
5We conducted an extensive sensitivity analysis for our choice of the packaged goods, advertising in this category has declined
"window" length of three quarters.Using a sub-sample of the data, we over time relative to promotion(Figure 1). This enables us
comparedthe coefficients of the three-quarterwindow with the coefficients to examine the effects of these changes, if in fact they exist.
from one-quarterand five-quarterwindows. The results were not statisti-
cally significantly different across these scenarios. We acknowledge that
Second, this is a matureproductcategory.Therefore,our re-
using the three-quarterwindow parameterscould inflate the lag parameter sults will not be confounded by changes over the product
in Equation8. life cycle (Sethuramanand Tellis 1991; Tellis 1988b). Final-
Figure 1
CHANGES IN CATEGORYMARKETINGACTIVITY
ofDiscounts
Frequency Advertising
Percent Dollars(000)
40 500
400
30
300
20
200
10
100
0 0 191111111111111111111111111191
1984 1985 1986 1987 1988 1989 1990 1991 19
Time
Time Time
Time

ly, there were no majorbrandentries or exits duringthe pe- (e.g., advertising,feature)are brandorientedratherthande-
riod to complicate the analysis of advertisingand promo- signed specifically for a brand size. Previous studies use
tional effects. brand, instead of brand size, for similar reasons (Krishna-
The category consists of eight majorbrandsthat account murthiand Raj 1991).
for approximately85% of the market.We consider all other
brands,along with private-labeland generic brands,to con- VARIABLES
stitutea ninthbrand.There are four majorsizes, and there is Short-Term(Weekly)Variables
a great deal of switching among sizes. The two medium
sizes representapproximatelythree-quartersof the market We include the following short-term variables X'i (for
share,and the largestand smallest accountfor approximate- household h, brandi, occasion t) in the logit model.
Price. Price for a brandis the actual shelf price in dollars
ly one-eighth each. There are 11 stores in this particular
market. The panel data include 1590 households making per ounce, net of all discounts. The existence of multiple
54,731 productpurchasesover 8 V4years, thatis, 33 quarters. brandsizes poses a problemfor creatinga price variablefor
The demographicsof the panelists roughlymatchthe demo- competing brands.We use minimumshelf price per ounce
across different sizes of a brandas the price for that brand
graphicsof the United States on age, family size,.education,
and children;though the panelists in our study had slightly for several reasons. First, in our data the average unit price
differentialamong the three largestbrandsizes, which con-
higher incomes. Approximately 92% of households pur-
chased more than one brandduringthe period of the study. stitute the bulk of purchases,is approximately3%. Second,
Households did not enter or exit the panel during the study the average price differences across brands are far greater
than are the average price differences across sizes of the
period. The lack of entry or exit could raise questions of
maturationor mortality,however,any disaggregatelong-du- same brand.Third, and most important,a weighted average
ration study will suffer from this limitation.6By having the price formulationis likely to underestimatethe true price
same households, we are uniquely able to monitorexposure variationin the marketplace.For example, if a retaileroffers
to promotionsand trackconcurrentchanges in behavior.As a 15%discount on the largest size of a brandthat accounts
we mentioned previously, conflicting results about the im- for, say, one-eighth of the volume, then the weighted aver-
pact of advertising on consumers' price sensitivity can be age percent discount will be (.15 x 1/8), or less than 2%.
Given size switching in this category, such averaging will
explained partiallyby whetheradvertisingis attractingnew
understateprice effects. Fourth,consumers switch heavily
price-sensitiveconsumers(Eskin and Baron 1977) or reduc-
ing the price sensitivity of existing consumers (Krishna- among sizes, which suggests that minimumprice per ounce
murthiand Raj 1985). By having a fixed set of households, more accuratelyreflects their price search behavior.
we avoid this potentialconfounding.Table 2 provides some Price promotions.We classify three types of promotions
as price promotions:(1) temporaryprice reduction(TPR),
descriptivestatistics of the data.
Although most brandsin our productcategory have four (2) feature,and (3) coupon. Temporaryprice reductionand
sizes, our unit of analysis is a brand, not brand size. This coupons always are accompaniedby price discountsand are
choice was made for the following reasons:First, with nine thereforeclearly price-promotionsignals to consumers.Fea-
brands and four sizes, even a single-segment logit model tures almost always are accompaniedby prominentpricing
will have a large number of parameters.This problem in- informationwith little productinformation.
creases dramaticallyin a multi-segment logit model. Sec- We define a brandto be on featureor TPR, or as offering
ond, a model with 36 brand-sizecombinationsis likely to vi- a coupon if any size of thatbrandis on featureor TPR or of-
olate the IIA propertyof the logit model. Third, manage- fering a coupon. Although the data provide information
ment believes that most marketingactions in this category about coupon redemption,there is no informationabout the
recency of coupon drop. Therefore, we assume a coupon
61nthe spirit of cohort analysis, we checked our data to see if the per- drop for a brandoccurred in a week if the total numberof
centage of purchaseson discount or brandswitching for threedifferentage redemptionsfor thatbrandby all householdsin the panel da-
groups of panelists differedover time. We found no significantdifferences. ta exceed one standarddeviation higher than the average
Table 2
DESCRIPTIVE STATISTICSOF THE DATA
Market Price Advertising

Share ($/ounces) ($000) TPRI Feature Display Coupon
Brand 1984 1991 1984 1991 1984 1991 1984 1991 1984 1991 1984 1991 1984 1991
I .38 .38 .057 .060 393 128 .27 .55 .04 .09 .08 .19 .07 .10
2 .07 .10 .054 .059 142 30 .12 .40 .07 .03 .05 .14 .12 .04
3 .19 .13 .057 .063 225 57 .12 .43 .01 .04 .00 .12 .14 .04
4 .04 .10 .052 .056 144 0 .10 .54 .01 .04 .04 .19 .00 .04
5 .08 .07 .057 .062 278 44 .08 .42 .00 .05 .02 .15 .05 .15
6 .09 .06 .049 .049 0 0 .03 .15 .00 .01 .02 .05 .14 .04
7 .04 .03 .029 .034 0 0 .04 .08 .00 .01 .00 .02 .25 .06
8 .00 .04 .061 .054 0 0 .08 .28 .00 .02 .00 .05 .00 .42
9 .11 .10 .024 .030 0 0 .18 .21 .01 .01 .02 .06 .06 .04
ITPR (temporaryprice reduction), Feature,Display, and Coupon variablesrepresentthe proportionof times the category was discounted, featured,dis-
played, or couponed in that year.

weekly redemptionfor that brandacross the entire period of which we use quarterly marketing variables to explain
the study. In other words, a significant increase in coupon changes in consumers'sensitivities. These results provide a
redemptionis assumed to be an indicatorof a recentcoupon test of our hypotheses.
drop.
Non-price promotions. We classify display as the non- Stage One: EstimatingSegment-LevelLogit Models
price promotionvariable.Typicallyprice is not the dominant Choosing the numberof segments. For each time period,
focus of displays. Similar to features and TPR, we define we calibrated the single-segment choice model and then
display as a binary (0,1) variablethat is I if any size of the proceeded to fit multi-segment models for two and three
brandis on display, 0 otherwise. segments.9 We used the Bayesian Information Criterion
Brand loyalty. To control for observed differences in (BIC) and segment interpretabilityas guidelines to select the
households'purchasingpatterns,we include a brandloyalty appropriatenumberof segments (Bucklin and Gupta 1992).
variable (e.g., Guadagni and Little 1983). A household's For all time periods, the BIC values favoreda two-segment
loyalty for a brandis defined as thatbrand'ssharein the last solution over a single-segment solution. Three-segment
four purchasesof that household (on average four purchas- solutions were occasionally better,and often worse, than the
es representone year of purchasehistory in this category). two-segmentsolutions.Across time periods,the averageand
Previous studies use similar or even fewer purchasesto as- maximumgain in BIC in going from two to three segments
sess brand loyalty (e.g., Chiang 1991 used four weeks of were .30 and 23, respectively.Even the maximum gain in
coffee purchases).Our measure enables a household's loy- going from a two-segment to a three-segmentsolution rep-
alty for a brandto change over time, that is, we do not as- resented a percentagegain of only .5% in the BIC for that
sume that brand loyalty for a household remains constant period, at the expense of 15 additionalparameters.Because
over eight years.7 a constant number of segments over time facilitates inter-
pretation(Moore and Winer 1987), we chose a two-segment
QuarterlyVariables solution for all time periods.
Quarterlyvariables(Ziq)for each brandi and quarterq are The two segmentsappearedto be relativelyloyal and non-
operationalizedas follows: loyal segments, which is consistent with the conceptualiza-
Advertising.Advertising is the total advertisingexpendi- tion of Krishnamurthiand Raj (1991), Bucklin and Gupta
ture (in inflation adjusteddollars) for brandi in quarterq in (1992), and others.
the marketarea.8 Averagesensitivities of consumersacross all time periods
Price promotions.Once again, the intensity of price pro- were as follows:
motion by a brandin a quarteris capturedby three types of
Consumer Loyal Nonloyal
promotions:(1) TPR, (2) feature, and (3) coupon. Specifi- Sensitivityto Segment Segment
cally, we define the feature activity for brandi in quarterq
Price -.28 -1.70
(Fiq)as the proportionof times brandi is on feature across Price promotion .02 .04
all stores and all weeks of that quarter.Similarly,we define Non-price promotion .03 .09
frequency of price discounts as the proportionof times a
brand is on TPR in a quarter.Finally, coupon intensity for These results show that, as expected, consumers in the
brandi in quarterq is defined as the proportionof times that nonloyal segment are more price and promotion sensitive
branduses coupons in that quarter.Quarterlyprice promo- than are consumersin the loyal segment. The price sensitiv-
tion is defined as a simple averageof quarterlyfeature,TPR, ity (or slope) estimates are equivalentto price elasticities of
and couponing activity of a brand. -.51 for the loyal segment and -1.02 for the nonloyal seg-
Non-price promotions.Our non-pricepromotionvariable ment. These estimates are somewhat lower than the average
is the proportionof times a brand is on display across all of-1.76 reportedby Tellis (1988b), who used a meta-analy-
stores and all weeks of that quarter,operationalizedsimilar sis of several productcategories.
to the long-termfeaturevariable. Consumers'sensitivities over time. Over the 29 quarters
RESULTS of the calibration period, in addition to segment size and
brand-specificconstants, we estimated 348 response para-
We first present the results of the first stage analysis, in meters in the first stage of our analysis.l0 Of these, 316 co-
which we calibrate segment-level logit models to obtain efficients were significant and correctly signed (e.g., the
quarterlyparameterestimates and consumers' sensitivities. price parameterwas negative, as expected), 8 were signifi-
Next, we discuss results from the second stage analysis, in cant and incorrectlysigned, and 53 were insignificant. No-
tice that under the null hypothesis of zero parametervalue,
7Results were insensitive to slightly different operationalizationsof
brandloyalty. Also, response parametersin the first stage were similar for
using a 5% significance level would yield 348 x .05 = 17 co-
models with or without the brandloyalty variable.
8Because [ parametersare estimated for a quarter,it could be unreason- 9Wealso compareda single-segmentmodel with brand-specificparame-
able to suggest thatadvertisingfor the entirecurrentquarter(includingend- ters for price, promotion,and so on with a single-segment model in which
of-quarter advertising) affects consumers' sensitivities for that quarter the parametersfor short-termmarketingvariableswere constrained to be
(which include consumerresponse at the beginning of the quarteras well). the same across brands. Using Bayesian Information Criterion (which
Therefore,the quarterlyadvertisingvariablewas created by phase-shifting penalizes a model for estimatingadditionalparameters),a model with the
advertising by half a quarter.In other words, for the current quarterof same parametersfor brandsperformedbetterthandid a model with brand-
April-June, advertisingwas defined as half the advertisingexpenditurefor specific parameters.Moreover, this constrained model was found to be
April-June plus half the advertisingexpenditure for January-March.We more stable over time.
also defined the lag terms accordingly.Once again, this adjustmentdid not 10Foreach quarterwe estimated 12 response parameters:6 parameters
affect the empirical results. We obtained similar results for other quarterly for each of the two segments. Thereforefor the 29 quarters,we get 6 x 2 x
variables. 29 = 348 coefficients.

efficients as significant,by chance. In other words, it is pos- Table 3

sible to obtain approximately8 coefficients as significant OVERTIME
CHANGEIN CONSUMERS'SENSITIVITIES
and correctlysigned, and 8 coefficients as significantand in-
correctlysigned, by chance. Consumers' Changes Significance
Next, we used these estimatesof parametersand segment Sensitivityto Over Time Level
sizes and Equations6 and 7 to compute consumersensitivi- 1. Loyal Segment
ties to price, price promotion,and non-pricepromotionfor Price Increase p < .001
each quarterq. ll We performeda simple trend analysis for Price Promotion Increase p < .001
each brandand segment by regressingquarterlysensitivities Non-price Promotion Decrease p < .05
against time. To control for exogenous macroeconomicfac- 2. Nonloyal Segment
tors, we used recession (0,1) as a covariate because con- Price Increase p < .001
sumerscould become more price sensitiveduringa recession Price Promotion Increase p < .001
period.Consistentwith the economics literature,we define a Non-price Promotion Decrease p < .05
quarterto be in recession if it was a period of three or more Nonloyal Segment Size Increase p < .05
consecutivequarterswith negativegrowth in gross domestic
product.Two interestingresults emerged from this analysis.
First,consumersin the nonloyal segmentbecame more price
sensitive over time, whereas loyal consumers showed little changing over time?"-is a resoundingyes.12Of special in-
change in their price behavior. Second, the nonloyal con- terest is the result that price and price promotionsensitivi-
sumersalso demonstratedincreasingsensitivityto price pro- ties are increasing over time. Our findings have important
motions and reduced sensitivity to non-price promotions, implications for managers and researcherswho attempt to
though these effects were weaker than were the effects for draw empirical generalizations using meta-analysis or to
price.The promotioneffects for the loyal consumerswere in compare the results of two or more studies conducted over
the same directionas for the nonloyalconsumers,thoughthe different time periods. For example, Tellis (1988b) and
brand-levelcoefficients were not significant. Sethuramanand Tellis (1991) find averageprice elasticity to
We also conducteda trendanalysis for the nonloyal seg- be around-1.7. However,our results suggest that these av-
ment size (7) with ln(7/l - n) as the dependentvariable,and erages are likely to change over time. We next examine the
trend and recession as the independentvariables. This re- reasons for these changes.
gression showed a significantincreasein the size of the non-
loyal segment over time, which indicates that an increasing Stage Two:Impact of MarketingActivityon Consumers'
Sensitivities
proportionof consumers have become more price and pro-
motion sensitive over time. The purposeof the second stage analysis is to test explic-
Pooling across brands.It is possible thatconsumersensi- itly if advertisingand promotionaffect consumers'price and
tivities for most brandscould show a positive but insignifi- promotionsensitivities over the long run. We used the dis-
cant trend. However, if we conclude from these results that tributedlag model described in Equation8, in which con-
consumer sensitivities are not changing over time, we may sumers' sensitivities (i.e., first derivatives)t6 weekly price,
be making a Type II errorby not appropriatelypooling in- price promotion,and non-pricepromotionactivities for each
formation across brands (Dutka 1984). Pooling results brandand segment were the dependentvariables,and quar-
across brands also can be thought of as a "meta-analysis" terly advertising,price promotion,and non-pricepromotion
across several brands. of that brandwere the independentvariablesalong with the
One possibility is to run a category-level regression, lagged dependentvariable.We also includedrecession as an
which gives us a single-trend parameterfor the category independent variable to control for exogenous macroeco-
ratherthanseparatetrendparametersfor the nine brands.Al- nomic factors. We first ran Durbin'stest to check for serial
thoughthis approachincreasesthe sample size and pools the correlation of errors in a distributedlag model (Johnston
informationacross brands,it also could lead to erroneousre- 1984, p. 318). Out of 54 regressions(2 segments x 3 depen-
sults if the parametersfor brandsare so far apartthat pool- dent variablesx 9 brands),47 regressionshad no correlation
ing across brandsincreasesthe varianceof the category-lev- in the errors.Therefore,our formulationseems to be consis-
el parameter,making it insignificant. Treating the results tent with the partialadjustmentmodel. Next, we performed
from different brands as independentreplications, we use a test as per Johnston(1984, p. 347) to see if the decay para-
Fisher's pooling test (Fisher 1948; Rosenthal 1978). Pooled meter A is different for advertising, price promotion, and
results that are based on this test are summarizedin Table3. non-price promotion. For all nine brands, we could not
These results suggest that the answer to our first ques- rejectthe null hypothesis thatall of the decay parametersare
tion-"Are consumers' sensitivities to marketingvariables the same.
12Toconfirm furtherwhether the long-term series of consumers' sensi-

tivities to price and promotionis evolving, for all nine brandswe performed
the unit root test suggested by Dekimpe and Hanssens (1995a). We found
strongevidence of evolving sensitivities in the nonloyalsegment and some-
ItToobtain consumers'sensitivity to price promotion,we used a simple what weakerevidence for the loyal segment.We also found the advertising
average of consumers' sensitivity to TPR, feature, and coupons. Results and promotionvariablesto have unit roots. We thankMarnikDeKimpe for
were not sensitive to differentweighting schemes. making the necessary programavailablefor this test.

Using this brand-and segment-specificregression,we es- of this effect is on the nonloyalconsumers.13Furthermore,we

timated the Y parametersassociated with each of the quar- find that less advertisinghelps to increasethe size of the non-
terly variables(see Equation8). These parametersreflect the loyal segment.This dual effect (i.e., a reductionin advertising
quarterly(medium-term)effect of advertising and promo- leading to an increase in the size of the nonloyal segment and
tion on consumers'responsesensitivities. Results from these making consumers more price sensitive) shows a powerful
role of advertising in reinforcing consumer preferences for
brand-specificregressions were pooled according to Fish- brands.Table4 also shows that reduction(increase)in adver-
er's method.We estimatedthe long-termimpactof advertis- tising reduces (increases) consumers' sensitivity to non-price
ing and promotionon consumers'sensitivities [9 = y/(I - Xk) promotionfor the nonloyal consumers.In other words,adver-
and its variancefor each brandand segment.We then pooled tising makes non-pricepromotionssuch as displays more ef-
brand-levelresults using Fisher's method. fective. Moreover,our results indicate that if a markethas a
Finally, we also regressed the size of the nonloyal seg- large "loyal"segment, then using an unsegmentedmodel can
ment (n) against quarterlymarketingvariables.Specifically, lead to insignificant and/or substantiallysmaller advertising
we used log(7/l- n) for each quarteras the dependentvari- effects. This might explain partiallythe inabilityof many re-
able and quarterlyadvertising, price promotion, non-price searchersto find any advertisingeffects in scannerdata.
2. Price promotions.As expected, in the long run price promo-
promotion, and recession as independent variables along tions make consumers more price sensitive in both the loyal
with the lagged dependent variable. Because the segment and nonloyal segments.They also trainnonloyalconsumersto
sizes were not brand specific, we used the mean category look for promotions,thereby making them more sensitive to
level variables(e.g., averagecategory advertising)as our in- price promotions.
dependent variables.The purpose of this regression was to 3. Non-pricepromotions.Non-pricepromotionshave significant
see if, for example, decreasedadvertisingexpendituremade effects only on consumers'price sensitivities. Consistentwith
consumers less brandloyal. our hypotheses, these promotionsmake loyal consumersless
price sensitive and make the nonloyal consumers much more
The Long-Term Effects of Advertising and Promotions price sensitive. These results are consistent with prior litera-
ture that suggests opposite effects of certain promotionsfor
Results of both the medium-term(over one quarter)and differentconsumers.
long-term (over infinite horizon) impact of advertisingand
promotionon consumers' price and promotionsensitivities Several additional insights emerge from our analyses.
are given in Table 4. To facilitate comparison and discus- First, the results are identical for the medium-term and long-
sion, we have also includedthe hypothesizedsigns for these term effects of advertising and promotion (except for the
effects. Overall, we found supportfor many of our hypothe-
ses. Although some of the results were insignificant, none 13Althoughour data show advertisingexpendituresto be declining over
were significant in the wrong direction. time, we expect our results to be bidirectional (i.e., we expect an
increase/decreasein advertisingexpenditureto reduce/increaseconsumers'
1. Advertising. Consistent with several previous studies (e.g., price sensitivity). Also, when the price parameter(which is negative) is
Krishnamurthiand Raj 1985), we find that a reductionin ad- used as a dependentvariable,a positive coefficient for advertisingimplies
vertising makes consumersmore price sensitive and that most that advertisingreduces price sensitivity.
Table 4
HYPOTHESES AND RESULTS
1. HypothesizedSigns
Loyal Segment Nonloyal Segment
Impacton ConsumerSensitivityto Impacton ConsumerSensitivityto
Nonloyal
Long-Term Price Non-price Long-Term Price Non-price Segment
Impactof Price Promotions Promotions Impactof Price Promotions Promotions Size
Advertising + - + Advertising + - +
Price Promotions - + - Price Promotions - + - +
Non-price Promotions + - + Non-price Promotions - - + ?
2. Resultsfor Medium-TermEffects
Advertising ns ns ns Advertising + ns + -
Price Promotions - ns ns Price Promotions - + ns ns
Non-price Promotions + ns ns Non-price Promotions - ns ns ns
3. Resultsfor Long-TermEffects
Advertising ns ns ns Advertising + ns + -
Price Promotions - ns ns Price Promotions - + ns ns
Non-price Promotions + ns ns Non-price Promotions - ns ns ns
Note: Price sensitivity is generally negative. Therefore, a positive sign (e.g., ad-price) indicates a decrease in price sensitivity. Significance is based on
pooled Fisher test with p < .05.

Table 5
ANDPROMOTION
AVERAGEEFFECTOF ADVERTISING ON CONSUMERS'
SENSITIVITIES
REGRESSIONRESULTSFROMSECOND-STAGE
ANALYSIS1

Price- Non-price- Price- Non-price- Nonloyal
Price Promotion Promotion Price Promotion Promotion Segment
Impactof Sensitivity Sensitivity Sensitivity Sensitivity Sensitivity Sensitivity Size
Lag (x) .384* .461* .410* .516* .285* .389* .639*
Advertising .0007 .000 .000 .008* .000 .0003* -.0001 *
Price Promotion -.374* -.023 -.006 -1.207* .080* .011 .059
Non-price Promotion .349* .041 .011 -4.368* .020 -.029 -.047
Recession -.106* -.003 -.012* -.015 .005 -.001 -.001
R2 .29 .32 .39 .56 .35 .27 .74
'Numbers in this table are weighted averages of the parametersfrom brand-levelregressions, where weights are proportionalto inverseof the parameter
variances(R2 is a simple, not weighted average across brands).Therefore,they representmedium-termeffects of advertisingand promotionon consumers'
sensitivities. Parameterswith p < .05 based on pooled Fishertest are representedby an asterisk.
"effect size" which we discuss subsequently).Second, the size in the mediumtermfor each of the two segments.15The
average R2 for the second stage analysis is .37 (R2 = .26). long-termeffect size is then simply the medium-termeffect
The model fit is slightly poorerfor the loyal segment (aver- size divided by (1 - X), where X is the decay parameter.
age R2= .34, R2= .21) than for the nonloyal segment (aver- We give the effect sizes in Table 6. Three key results
age R2 = .39, R2 = .30). This is an intuitively appealing emerge from this table: First, consistent with previous re-
result, which suggests that long-term marketingvariables search, advertising effects are generally smaller than pro-
can explain better the changes in the behavior of nonloyal motion effects; second, long-term effect sizes are approxi-
consumers, who by definition are affected by marketing mately 1.5 to 2 times largerthan medium-termeffect sizes;
variables more than are loyal consumers. This is also and third, the changes in price sensitivities are greaterthan
reflected in more significant signs for the nonloyal segment the changes in promotionsensitivities.
than for the loyal segment. Third, the decay parameterX
varies from .29 to .52 with an average of .41 (see Table 5). Summaryof Results
This indicatesthaton averagethe long-termeffects of adver- In summary, our results suggest the following:
tising and promotion are 1/(1 - .41), or 1.7, times their
medium-term(quarterly)effects. The decay parameteralso *Advertisinghelps a brandin the long runby makingconsumers
(especially nonloyal ones) less price sensitive as well as by re-
suggests that 90% of the cumulative effect of advertising ducing the size of the nonloyal segment.
and promotion on consumers' sensitivities occurs within *In the long run, price promotionsmake both loyal and nonloy-
three quarters, which is similar to the result obtained by al consumers more sensitive to price. An increaseduse of such
Clarke (1976). Finally, though recession had no impact on promotionsalso trainsconsumers(especially nonloyal ones) to
consumers' sensitivities in the nonloyal segment, it look for deals in the marketplace.On average,we find these ef-
increased consumers' price sensitivity and reduced their fects to be almost four times largerfor nonloyal consumersthan
non-pricepromotionsensitivity in the loyal segment. for relatively loyal consumers.
*As expected, we found that non-pricepromotionshave different
How LargeAre the Long-TermEffectsof Advertisingand effects for loyal and nonloyal consumers.Although these pro-
motions act like advertisingfor loyal consumers,making them
Promotions?
less price sensitive, they make the nonloyal consumers focus
Our previous discussion indicates the directionaleffects even moreon prices.In otherwords,these promotionsreducethe
of advertisingand promotions.It is also useful to assess the price sensitivityof loyal consumersbut significantlyincreasethe
magnitudeof these effects. We estimated the medium- and price sensitivity of nonloyal consumers.The effects for nonloy-
long-termeffect sizes as follows. als are almost 12 times largerthanthe effects for loyals.
To estimate the medium-termimpact, y, we note from To the extent that increasedprice and promotionsensitiv-
Equation 8 that the parametery = aY/aZ represents the ity of consumers can be considered undesirableoutcomes,
change in consumers' sensitivity (Y) for one unit change in our resultsconfirmthe conventionalwisdom thatin the long
the quarterlyvariableZ. Therefore, (y/Y) x 100 is the per- run, advertising has "good" effects and promotions have
cent change in consumers'sensitivity for one unit change in "bad"effects on consumers' brand choice behavior. Note,
Z. 14We use the averageY and Z values to arriveat the effect however, that these results are based on the analysis of only
one category in one market.
15Strictlyspeaking,we should evaluatechanges in Y for each brand,seg-

(Share Price Price ment, and quarter.However,our "aggregate"approachshould provide us
4Becauseprice elasticity r1 = = Y -- it is easy to with a good approximation.For example, in the context of discrete choice
aPrice Share Share models, Ben-Akiva and Lerman(1985) indicatethat errorsdue to aggrega-
show that under certain conditions, (y/Y) x 100 also representspercent tion across consumers are relatively small if all consumers have the same
change in Trfor one unit change in Z. choice set.

Issues of Causality b) for the following reasons. Increasing promotionand re-

This study is correlationalin nature. It is possible that duced advertisingmakes consumersmore price and promo-
tion sensitive.This leads firms to offer even more price pro-
changes in consumer behaviorhave led to changes in mar-
motions to obtain short-termgains. However, all firms en-
keting activityratherthanchanges in marketingactivitylead-
ing to changes in behavior.We empiricallytested for causal- gage in this action. This form of competitive action has two
effects: First, it makes the overall promotionalintensity in
ity by runninga simultaneousequationmodel as follows:
the category higher, thereby making consumers even more
= f(Lagpricesensitivity,
Pricesensitivity Ad,PP,NPP,Recession) price sensitive over time (as we find); and second, competi-
PPsensitivity= f(LagPPsensitivity, Ad, PP,NPP,Recession) tive matching of promotional spending leaves the relative
NPPsensitivity = f(LagNPPsensitivity, Ad,PP,NPP,Recession) promotionacross firms to be largely the same, therebyhav-
Ad = f(Lag Ad of brand,CompetitiveAd, Price ing little effect on their marketshare. In other words, it is
sensitivity,PP sensitivity,NPPsensitivity) conceivablethatthoughconsumersare becoming more price
PP= f(Lag PP of brand,CompetitivePP, Price sensitive because of increasedpromotionalactivities of var-
sensitivity,PP sensitivity,NPPsensitivity) ious firms, "theoffsetting competitiveactivity plays a role in
NPP= f(LagNPPof brand,Competitive NPP,Price the maintenanceof what appearsto be stationaryand zero
sensitivity,PPsensitivity,NPPsensitivity),
order behavior"(Bass et al. 1984, p. 284), leading to sta-
where PP is price promotionand NPP is non-price promo- tionary market shares. This highlights the importance of
tion. The first three equations represent our second-stage studyingthe long-termeffects of promotionand advertising
analysis and the last three equations capture the reverse on not only the changes in marketshareor sales, but also on
causality (i.e., the effect of changes in consumers'sensitivi- changes in consumerbehavior.
ties on marketingactivities). We estimated these equations
at a brandand segment level using two-stage least squares. CONCLUSION
We then pooled the resultsacross brandsusing Fisher's pro- Our purposewas to develop and empirically test a model
cedure, as indicatedpreviously. to understandthe long-term impact of advertisingand pro-
Of the seven significant results obtained in Table 4, four motion on consumers' brandchoice behavior.Specifically,
show causalityonly in the proposeddirectionat p < .05, and we set out to find if consumers'responsivenessto marketing
three show bidirectionaleffects (and two of them have p- mix variableschanges over time and, if so, why. To address
values that are almost ten times smaller in the proposeddi- these issues we used a unique data set that includes store
rection than in the reversedirection).Although these results environmentand purchasehistoryof more than 1500 house-
do not rule out reverse causality completely, they do lend hold for 8V4years for one frequently purchasedconsumer
supportfor the proposedcausal effects. packagedgood in one market.
At least in our data set, consumers have become more
CompetitiveReaction price and promotionsensitive over time. We found two seg-
Recent articles by Lal and Padmanabhan(1995) and ments of consumers-loyal, or relativelyless price-sensitive
DeKimpe and Hanssens (1995a, b) suggest that market consumers,and nonloyal, or price-sensitiveconsumers.Our
sharesare stationary(i.e., not evolving over the long run)for resultsshow thatthe size of the nonloyal segmenthas grown
a majority of product categories. Lal and Padmanabhan over time. In other words, a larger number of consumers
(1995) also find the long-termeffect of promotionson mar- have become increasingly more price and promotionsensi-
ket shares to be largely insignificant. tive over time.
In contrast, we find long-term effects of advertisingand We used quarterlyadvertisingand promotionpolicies of
promotionson consumers'sensitivities to price and promo- brandsto help explain these changes in consumerbehavior.
tions, not on the marketshares of various brands.We con- Ourresultsconfirmthe conventionalwisdom thatin the long
jecture thatour findings are consistent with those of Lal and run, advertising reduces consumers' price sensitivity and
Padmanabhan(1995) and DeKimpe and Hanssens (1995a, promotionsincrease consumers'price and promotionsensi-
Table 6
MEDIUM- ANDPROMOTION
EFFECTSOF ADVERTISING
ANDLONG-TERM ON CONSUMERS'SENSITIVITIES1

% Change in Consumers'Sensitivityto % Change in Consumers'Sensitivityto
Price Non-price Price Non-price
1% Increase in Price Promotion Promotion Price Promotion Promotion
Medium-TermEffects
Advertising ns ns ns .15 ns .10
Price Promotion -.37 ns ns -1.20 .08 ns
Non-price Promotion .35 ns ns -4.37 ns ns
Long-TermEffects
Advertising ns ns ns .30 ns .15
Price Promotion -.61 ns ns -2.50 .11 ns
Non-price Promotion .57 ns ns -9.02 ns ns
IWe focus on only significanteffects (p < .05). Note that negative numbersfor price imply increase in price sensitivity.

tivity. These effects were found to be significantly larger for Clarke,DarralG. (1976), "EconometricMeasurementof the Dura-
nonloyal consumers than for loyal consumers. In general, tion of AdvertisingEffect on Sales," Journal of MarketingRe-
compared with the "good" effects of advertising, promo- search, 13 (November),345-57.
tions were found to have significantly larger "bad" effects Comanor,WilliamS. andThomasA. Wilson (1974), "TheEffect of
on consumers' price and promotion sensitivities. We also Advertisingon Competition,"Journalof EconometricResearch,
18, 453-76.
conjecture that though increased promotional spending
Davis, Scott, JeffreyInman,and Leigh McAlister(1992), "Promo-
could make consumers more price sensitive over time, mar- tion Has a Negative Effect on BrandEvaluations-Or Does It?
ket shares of brands may not see any long-term trends be- AdditionalDisconfirmingEvidence,"Journal of MarketingRe-
caue of offsetting competitive activities. search, 29 (February),143-48.
Although this study provides some managerial insights DeKimpe, Mamik G. and Dominique M. Hanssens (1995a), "The
about the relative roles of advertising and promotions, addi- Persistenceof MarketingEffects on Sales," MarketingScience,
tional research is required before we can make such general 14 (Winter), 1-21.
recommendations to managers such as to increase advertis- and (1995b), "Empirical Generalizations about
ing and reduce promotions. Specifically, the results of this Market Evolution and Stationarity,"MarketingScience, 14 (3,
study are limited by the fact that we only studied one prod- part2 of 2), G109-G121.
uct category in one market. Generalization of these results Dodson, Joe A., Alice M. Tybout,and BrianStemthal(1978), "Im-
across several markets and categories will be a useful area of pact of Deals and Deal Retractionon BrandSwitching,"Journal
of MarketingResearch, 15 (February),72-81.
further research. Moreover, it will be useful to compare the Dutka,Solomon (1984), "CombiningTests of Significance in Field
short- and long-term changes in brand shares due to differ- Marketing Research Experiments,"Journal of Marketing Re-
ent advertising and promotion policies. It is possible that the search, 21 (February),118-19.
large, positive short-term benefits of promotions outweigh Ehrenberg,A.S.C., K. Hammond, and G.J. Goodhardt (1994),
their relatively small, negative long-term effects. The study "The After-Effects of Price-Related Consumer Promotions,"
also can be extended by focusing on sales and profits instead Journal of AdvertisingResearch,34 (July/August), 11-21.
of shares. It is conceivable that negative long-term effects of Eskin, GeraldJ. and Penny H. Baron(1977), "Effectsof Price and
price promotion on consumers' brand choice are significant- Advertisingin Test-MarketExperiments,"Journal of Marketing
ly mitigated or even reversed by promotions' positive long- Research, 14 (November),499-508.
term effects on consumers' purchase timing or purchase Fader,Peter S., Bruce G.S. Hardie,John D.C. Little, and Makoto
Abe (1992), "MarketResponse with PurchaseFeedback,"work-
quantity decisions. Therefore, understanding the long-term
ing paper,WhartonSchool, Universityof Pennsylvania.
effects of advertising and promotion on other components of
Fisher, R.A. (1948), "Combining IndependentTests of Signifi-
consumers purchase behavior, such as purchase incidence cance,"AmericanStatistician, 2, 30.
and quantity, is a fruitful area of further research. Here we Forbes (1991), "TheTrendis Not Their Friend,"(September).
have speculated about the competitive reactions; however, a Guadagni, Peter M. and John D.C. Little (1983), "A Logit Model
better understanding of the competitive dynamics over time of Brand Choice Calibratedon Scanner Data,"MarketingSci-
is certainly needed. In summary, we believe that under- ence, 2, 203-8.
standing the long-term effects of marketing variables on Gupta,Sunil (1988), "Impactof Sales Promotionson When, What,
brand share, sales, and competition is a useful and rich area and How Muchto Buy,"Journalof MarketingResearch,25 (No-
of academic research and managerial relevance. We hope vember), 342-55.
our study sparks interest in this direction. Inman,J. Jeffrey and Leigh McAlister (1993), "A RetailerPromo-
tion Policy Model Considering Promotion Signal Sensitivity,"
MarketingScience, 12 (Fall), 339-56.
REFERENCES
, and Wayne D. Hoyer (1990), "PromotionSignal:
Bass, Frank M., Moshe M. Givon, Manohar U. Kalwani, David Proxy for a Price Cut?" Journal of Consumer Research, 17
Reibstein, and Gordon P. Wright (1984), "An Investigationinto (June), 74-8 1.
the Orderof the Brand Choice Process,"MarketingScience, 3 Johnson, Tod (1984), "The Myth of Declining Brand Loyalty,"
(Fall), 267-87. Journal of AdvertisingResearch,24, 9-17.
Ben-Akiva,Moshe and Steven R. Lerman(1985), Discrete Choice Johnston, J. (1984), Econometric Methods. New York: McGraw
Models: Theory and Applications to Travel Demand. Cam- Hill.
bridge, MA: MIT Press. Kamakura,WagnerA. and Gary J. Russell (1989), "A Probabilistic
Blattberg,RobertC. and Scott A. Neslin (1989), "Sales Promotion: Choice Model for Market Segmentation and Elasticity Struc-
The Long and The Short of It,"MarketingLetters, 1, 81-97. ture,"Journal of MarketingResearch,26 (November), 379-90.
Boulding, William, Eunkyn Lee, and Richard Staelin (1994), Kaul,Anil and Dick R. Wittink(1995), "EmpiricalGeneralizations
"Mastering the Mix: Do Advertising, Promotion, and Sales About the Impactof Advertisingon Price Sensitivity and Price,"
Force Activities Lead to Differentiation?"Journal of Marketing MarketingScience, 14 (3, part2 of 2), G 151-G 160.
Research, 31 (May), 159-72. Kim, DongHoon and Donald R. Lehmann(1993), "The Long and
BrandWeek(1993), "Sleepless in Cincinnati" 34 (July 26), 18-24. Short Run Impactsof Advertising,Price, and Promotion,"work-
Bucklin, Randolphand Sunil Gupta (1992), "BrandChoice, Pur- ing paper,Columbia University.
chase Incidence,and Segmentation:An IntegratedModelingAp- Krishnamurthi,Lakshmanand SethuramanP. Raj (1985), "The Ef-
proach,"Journal of MarketingResearch, 29 (May), 201-15. fect of Advertising on Consumer Price Sensitivity,"Journal of
Business Week(1991), "What Happenedto Adertising,"(Septem- MarketingResearch, 22 (May), 119-29.
ber 23), 66-72. and (1991), "An EmpiricalAnalysis of the Rela-
(1992), "Not Everyone Loves a SupermarketSpecial," tionship Between Brand Loyalty and Consumer Price Elastici-
(February17), 64. ty,"MarketingScience, 10 (Spring), 119-29.
Chiang, Jeongwen (1991), "A Simultaneous Approach to the Lal, Rajiv and V. Padmanabhan(1995), "Competitive Response
Whether,What, and How Much to Buy Questions,"Marketing and Equilibria," Marketing Science, 14 (3, part 2 of 2),
Science, 10 (Fall), 297-315. G101-G108.

The Long-TermImpactof Promotionand Advertising 261
Mitra,Anusree and John G. Lynch, Jr. (1995), "Towarda Recon- paper,Ohio State University.
ciliation of MarketPower and InformationTheoriesof Advertis- Sethuraman,Raj and GerardJ. Tellis (1991), "An Analysis of the
ing Effects on Price Elasticity,"Journal of ConsumerResearch, Tradeoff Between Advertisingand Price Discounting,"Journal
21 (March),644-59. of MarketingResearch,28 (May), 160-74.
Moore, William and Russell S. Winer (1987), "A Panel Based Swait,JoffreandJordanLouviere(1993), "TheRole of the Scale Pa-
Methodfor MergingJoint Space and MarketResponse Function rameterin the Estimationand Comparisonof MultinomialLogit
Estimation,"MarketingScience, 6, 25-47. Models,"Journalof MarketingResearch,30 (August), 305-14.
Nelson, P. (1974), "Advertisingas Information,"Journal of Politi- Tellis, Gerard J. (1988a), "Advertising Exposure, Loyalty, and
cal Economy,78 (March/April),311-29. Brand Purchase:A Two Stage Model of Choice," Journal of
ProgressiveGrocer(1995), 'The Prisoner'sDilemma,"74 (May),99. MarketingResearch,25 (May), 134-44.
Rosenthal, Robert (1978), "Combining Results of Independent (1988b), "The Price Elasticity of Selective Demand: A
Studies,"Psychological Bulletin, 85, 185-93. Meta-Analysis of Econometric Models of Sales," Journal of
Russell, Gary J., Randolph E. Bucklin, and V. Srinivasan(1993), MarketingResearch,25 (November), 331-41.
"Identifying Multiple Preference Segments from Own- and WallStreetJournal (1992), "MajorFood CompaniesCut Advertis-
Cross-PriceElasticities,"MarketingLetters,4 (January),5-18. ing Budgets,"(April 2), B 10.
Sawyer,Alan G. and Peter R. Dickson (1983), "PsychologicalPer- (1996), "BrandManagersGet Old-time Religion," (April
spectives on ConsumerResponse to Sale Promotions,"working 23), A19.
*~~~~~~~~~ W'II R!
|
reA
AMA's newest books will keep you current.
Dictionary Terms D,,i...R

ofMarketing AMA
Marketing
Encyclopedia
Issues
andTrends
PeterD.Bennett,
editor 1LR[EI11 theFuture
Shaping
StateUniversity
Pennsylvania T "I I
. _ .m EO.._ editor
/ JeffreyHeilbrunn,
__ .. .~~. . 3 IMM,
P"r his completely revised and
expandededition is an essential ith an emphasison recent
Jl referencefor business and emergingissues, a !J!_I
professionalsand studentsalike. Fully 5 panel of leading marketing - [
cross-referencedfor ease of use, this s experts surveys the key areasof marketing,
comprehensive resource lists more than .::::...- identifies currenttrends,and providesrealisticprescriptions
2500 up-to-datedefinitionsof today's most important for the future.Contributorsinclude industryleadersand top-
marketingterms.Coveringboth the day-to-dayterminology name academics,as well as pioneersand inventorsof key
and the specialized vocabularyin corporateand academicuse, marketingstrategiesand techniques:TonyAlessandra,John
the Dictionaryhelps everyone from newcomersto senior-level Hauser,Philip Crosby,Philip Kotler,JackTrout,Al Ries, Don
marketingexecutivesgain a more thoroughunderstandingof E. Schultz, JagdishSheth, and many others.The past, present,
criticalmarketingconcepts. and futureof marketingis shapedin this renownedbook.
$32.95 AMAMembers/$39.95Nonmembers $42 AMAMember/$47.95 Nonmembers

1995. 1995.
-
- b -l1.. .l4.1---e. - -
CALL TOLL-FREE 1 -800-AMA- 1 150 TO PLACE YOUR ORDER.

HAVE YOURCREDITCARD AND AMA MEMBERSHIPNUMBERHANDY.REFER TO JR597 WHENYOU CALL.
AMERICAN M ARKETING ASSOCIATION

250 S. W ac ke r Drive, Suite 200, Chicago, IL 60606-5819


American Marketing Association

Uploaded by

Copyright:

Available Formats

American Marketing Association

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

American Marketing Association

Uploaded by

Copyright:

Available Formats

The Long-Term Impact of Promotion and Advertising on Consumer Brand Choice

Author(s): Carl F. Mela, Sunil Gupta and Donald R. Lehmann

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

The authors examine the long-termeffects of promotionand advertis-

The Long-Term Impact of Promotion and

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

where POjqsand Pqs are segment-specific parametersesti-

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

Market Price Advertising

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

efficients as significant,by chance. In other words, it is pos- Table 3

12Toconfirm furtherwhether the long-term series of consumers' sensi-

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

Using this brand-and segment-specificregression,we es- of this effect is on the nonloyalconsumers.13Furthermore,we

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

Loyal Segment Nonloyal Segment

15Strictlyspeaking,we should evaluatechanges in Y for each brand,seg-

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

Issues of Causality b) for the following reasons. Increasing promotionand re-

Loyal Segment Nonloyal Segment

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

AMA's newest books will keep you current.

Dictionary Terms D,,i...R

$32.95 AMAMembers/$39.95Nonmembers $42 AMAMember/$47.95 Nonmembers

CALL TOLL-FREE 1 -800-AMA- 1 150 TO PLACE YOUR ORDER.

AMERICAN M ARKETING ASSOCIATION

This content downloaded from 192.231.202.205 on Thu, 4 Dec 2014 01:36:10 AM

You might also like