Assessing long-term promotional influences on market structure

Publish in

Business & Economics


Please download to get full document.

View again

of 19
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Intern. J. of Research in Marketing 15 Ž1998. 89–107 Assessing long-term promotional influences on market structure Carl F. Mela a a,) , Sunil Gupta b, Kamel Jedidi b College of Business Administration, UniÕersity of Notre Dame, Notre Dame, IN 46556-5641, USA b Columbia UniÕersity, Graduate School of Business, New York, NY 10027, USA Received 1 November 1996; accepted 6 November 1997 Abstract The allocation of marketing budgets across advertising and sales promotions has changed over the p
  Ž . Intern. J. of Research in Marketing 15 1998 89–107 Assessing long-term promotional influences on market structure Carl F. Mela a, ) , Sunil Gupta b , Kamel Jedidi b a College of Business Administration, Uni Õ ersity of Notre Dame, Notre Dame, IN 46556-5641, USA b Columbia Uni Õ ersity, Graduate School of Business, New York, NY 10027, USA Received 1 November 1996; accepted 6 November 1997 Abstract The allocation of marketing budgets across advertising and sales promotions has changed over the past decade with amarked decrease in the percentage of budgets directed towards advertising. Moreover, there has been much speculationregarding how these changes have affected a brand’s positioning vis-a-vis its competitors. In spite of this speculation, ` previous research has not examined the impact of changes in promotion and advertising on market structure. The purpose of this paper is therefore to ascertain how changes in promotional and advertising policy affect market structure over thelong-term. The eight and one quarter years of scanner panel data used for our analysis indicate that brands in the analyzed Ž . product category tend to fall into premium r non-premium and attribute-based mildness tiers. Furthermore, the data suggestthat the differentiation between non-premium and premium brands has diminished during the period of our study Ž . 1984–1992 . The data also suggest that increases in price promotions and reductions in advertising have led to decreaseddifferentiation between brands. These findings suggest that shifts in marketing dollars from advertising to promotions havemade national brands more vulnerable to store brands’ marketing activity. q 1998 Elsevier Science B.V. Keywords: Market structure; Marketing mix; Promotion; Advertising; Econometric models 1. Introduction 1.1. Changing managerial tactics A key brand management decision pertains toallocating promotional budgets across advertising,trade promotions and consumer promotions. Re-cently, brand management’s disposition has lent it-self toward trade promotions. According to Progres- Ž . sive Grocer 1995 , 70% of manufacturers have in-creased promotions between 1990 and 1995. Indeed,manufacturers spend over $70 billion annually ontrade promotions and a large portion of those trade ) Corresponding author. Tel.: q 1-219-6318117; fax: q 1-219-6315255; e-mail: promotions were passed through to consumers in theform of discounts. Recently, Nielsen found that priceincreases in 71 out of 100 categories have not Ž matched inflation, in part due to discounting Wall . Street Journal, 1997 .The increasing levels of promotions have ledmanagers to be increasingly concerned about theinfluence of these promotions on brand differentia- Ž . tion. While certain promotions e.g. sampling ap-pear to have a positive effect on brand differentiation Ž Blattberg and Neslin, 1989; Gedenk and Neslin, . Ž . 1997 , others e.g. discounting may have negativelong-term consequences for brands even though theycan induce dramatic sales increases in the short-term. Ž . For example, Forbes 1991 suggests that an increasein trade promotional spending over a five year period 0167-8116 r 98 r $19.00 q 1998 Elsevier Science B.V. All rights reserved. Ž . PII  S0167-8116 97 00041-4  ( )C.F. Mela et al. r  Intern. J. of Research in Marketing 15 1998 89–107  90 led to a decline in share for the top three brands incategories ranging from popcorn to cat food. The Ž . Forbes 1991 article asserts that competition be-tween premium and non-premium tiers of brandsmay have somehow increased over time as a resultof an increase in promotions. Thus, the analysis of the long-term effects of promotions on competitionand market structure has considerable interest tomarketing managers. Yet much of the voluminousmarketing research to date has focused on the short-term effect of these instruments. Ž . Since Blattberg and Neslin 1989 and Gupta Ž . 1993 called for more research into the long-termeffects of this increase in promotions, several studieshave begun to appear. The literature on these effects Ž has covered sales Boulding et al., 1994; Dekimpe . Ž and Hanssens, 1995a,b , choice Mela et al., 1997; . Ž Papatla and Krishnamurthi, 1996 and share Lal and . Padmanabhan, 1995 . Dekimpe and Hanssens Ž . Ž . 1995b and Lal and Padmanabhan 1995 both ana-lyze how promotions impact brand market share over Ž . long periods of time e.g. years . They conclude that,although promotions increase share in the short run,they have little impact on the long run share. Yet it ismisleading to conclude from these studies that pricepromotions do not affect brand profits in the long-term. The positive short-term impact of one brand’spromotions on brand share is likely to increase theretaliatory use of promotions by other brands. This,in turn, can increase brand switching and price sensi-tivity. In this scenario, even if the long run brandshares remain the same, consumer behavior and prof-itability can change significantly.Some studies have begun to acknowledge this Ž . possibility. Boulding et al. 1994 find that higherlevels of promotion tend to induce greater sales pricesensitivity thereby reducing brand differentiation. Pa- Ž . patla and Krishnamurthi 1996 and Mela et al. Ž . Ž 1997 find that price oriented promotions e.g. . coupons and temporary price reductions induce Ž . higher price sensitivity. Mela et al. 1997 also findevidence that some types of non-price oriented pro- Ž . motions e.g. displays , may benefit brands overtime.Many other studies have also suggested that ad-vertising has an analogous, but opposite, effect topromotions by affecting price sensitivity over thelong run even in the face of relatively constant Ž . market shares. Dekimpe and Hanssens 1995a findthat market shares are largely unaffected by advertis-ing over the long-term. However, Kaul and Wittink  Ž . Ž . 1995 and Mitra and Lynch 1995 overview theextensive literature regarding the effect of advertis-ing on brands and conclude that unique, brand-ori-ented advertising in mature categories does tend toreduce sales price sensitivity. This type of advertis-ing has the opposite long-term effect of price promo-tions and serves to better differentiate brands. Con-versely, these studies suggest that advertising mes-sages that do not stress unique features may expandconsumers’ consideration sets, lead to diminishedbrand differentiation and greater competition. Priceoriented advertising is one such example of this.Although this research is beginning to shed muchlight upon how promotions and advertising affectconsumer choice, market share, brand sales, andprice sensitivity in the long run, the long-term effectof promotions and advertising on market structureremains an unanswered question. The purpose of thispaper is to fill that gap. 1.2. Dynamic market structure Ž Market structure the representation of interrela-tionships of a set of products or brands in a way thatreflects consumers’ evaluations of the brands in the . set has been an area of extensive research since it isa critical component affecting brand planning and Ž category strategy for a complete review see Elrod, . 1991; Elrod and Keane, 1995 . The importance of market structure in brand management has been re-flected in the more than 100 articles published in the Ž . area by 1987 Grover and Srinivasan, 1987 and 99 Ž published on the subject since then Business Ab- . stracts Database, 1997 . Although both market struc-ture and the long-term effects of promotion areimportant areas of research, virtually no research hasbeen done at the confluence of these research streams.Most previous market structure research has focused Ž on a static or ‘snapshot’ analysis see Blattberg andNeslin, 1989; Bemmaor and Schmittlein, 1991; . Gupta, 1993; Blattberg et al., 1995 . Bemmaor and Ž . Schmittlein 1991 recently wrote, ‘‘While there hasbeen some investigation of longer-run effects andcompetitive market strategies, these latter issues have Ž not been explored in as much depth as opposed to . shorter-run effects to date.’’  ( )C.F. Mela et al. r  Intern. J. of Research in Marketing 15 1998 89–107  91 Ž . Moore and Winer 1987 , in a pioneering studyon dynamic market structure, used ‘pick any’ data toassess competitive maps over time. Their resultssuggest that advertising moves brands closer to ideal Ž . points. Erdem 1996 accommodates dynamics in thechoice process when developing maps of marketstructure, yet the representation of market structure Ž . itself is fixed. Cooper 1988 used three-mode factoranalysis to assess changes in competitive structure Ž . over a 52 week period. The Cooper 1988 work presages ours. Compared to these previous studies, Ž . our emphasis is on longer-term many years of datachanges in market structure. In addition, our studyexplicitly examines the long-term effects of promo-tions and advertising on brand positions. Ž . Day et al. 1979, p. 13 offer one explanationregarding why studying shifts in market structuremay have remained an existing gap in the literature. Ž . Behavioral measures of market structure suffer froman endemic weakness...actual switching is affectedby current market factors such as...promotional mes-sage and expenditures.... If data are developed overlong periods of time...sufficient variability may havetaken place in the determinants of demand to revealsuch potential substitutability.With the advent of single source scanner dataover ten years ago, data streams have recently be-come available that enable researchers to look over‘long periods of time’ and erase the limits that Day Ž . et al. 1979 noted.Given the importance of market structure and thepaucity of research regarding how it develops andchanges, this paper seeks to understand how advertis-ing and promotions affect market structure over time.The paper is organized as follows. We first discussour approach. Next, we describe our data and presentour results. We conclude by outlining some manage-rial implications and some directions for future re-search. Our contribution is both substantive andmethodological. Our substantive contribution is toascertain the impact of many years of marketingactivity on market structure. Our methodological Fig. 1. Analytical procedure.  ( )C.F. Mela et al. r  Intern. J. of Research in Marketing 15 1998 89–107  92 contribution is to provide a parsimonious and reli-able approach to calibrating a dynamic model of market structure. 2. Analyzing changes in market structure 2.1. A longitudinal procedure for inferring market structure We outline our procedure for inferring changes inmarket structure and creating a dynamic brand mapin Fig. 1.This procedure consists of five steps. These stepsare overviewed below and are subsequently dis-cussed in greater detail. Ž . 1 Divide a long period of data into successive, Ž . discrete intervals for example, quarters . Use short- Ž . term weekly marketing activities to assess own andcross price sensitivities for each quarter. These seriesof price sensitivities contain information about howcompetition among brands changes over time. Ž . 2 Use the own and cross price sensitivity mea- Ž sures to partial out brand specific effects the degreeto which a brand is unique from the market as a . Ž whole and brands’ substitutability the degree to . which particular brand pairs are substitutable . Ž . 3 Apply MDS techniques on the longitudinalseries of brands’ substitutability to construct a dy-namic competitive map that portrays changes in sub-stitutability over time. Ž . 4 Assess how long-term changes in promotionand advertising affect brand specific effects. Ž . 5 Assess how long-term changes in promotionand advertising affect brands’ positions in the dy-namic competitive map. 2.2. Step 1: Estimate similarity matrices Ž . Shocker et al. 1990 suggest that own and crossprice responses 1 are particularly well suited for theanalysis of market structure as they offer a good 1 Cross price responses are defined as the change in the shareof brand j with respect to a change in the price of brand i . Ownprice response is defined as the change in share of brand i withrespect to a change in its price. measure of brand similarity as revealed by con-sumers’ actual purchase behavior. We therefore mea-sure changes in market structure via changes in cross Ž price responses see Elrod and Keane, 1995 for anextensive list of studies that use cross price matricesfor analyzing market structure in time invariant set- . tings .Using disaggregate household-level purchase data,we fit the clusterwise logit model of brand choice Ž . Kamakura and Russell, 1989 to each successive,discrete period of the data. The clusterwise logitmodel has two benefits. First, it accounts for hetero-geneity in households’ price sensitivities. Second,the clusterwise logit allows for a very general patternof cross price response and is therefore especiallywell suited for capturing changes in market structure.Specifically, the probability, P , of household h qt  choosing brand i at time t  during quarter q is givenby 2 S S h h w x P s p  P ; p  g 0, 1 ; p  s 1 Ý Ý iqt sq isqt sq sq s s 1 s s 1 1 Ž . where s indexes the market segment and p  is the sq size of segment s in period q . The probability of household h in segment s choosing alternative i attime t  is then given by the logit choice probabilityfor that segment,exp b q b X h Ž . 0 isq sq it h P s 2 Ž . isqt  I h Ý exp b q b X . Ž . i s 1 0 isq sq it  Using the series of estimated price sensitivity param-eters, b , we then simulate the effect of a one unit sq 2 We use a quarterly measure for several reasons. First, the Ž . specification is consistent with Moore and Winer 1987 and theadvice of sponsoring managers. Second, given the six week  Ž interpurchase times for this product, shorter periods e.g. monthly . or weekly could indicate little meaningful change and longer Ž . periods semi-annual or yearly could fail to capture some of the Ž . dynamics. The Moore and Winer 1987 panel data based analysisindicated monthly, quarterly and eight-month windows yieldedrelatively stable results.
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks