You are on page 1of 22

Order of Entry as a Moderator of the Effect of the Marketing Mix on Market Share

Author(s): Douglas Bowman and Hubert Gatignon
Source: Marketing Science, Vol. 15, No. 3 (1996), pp. 222-242
Published by: INFORMS
Stable URL:
Accessed: 29-05-2016 01:28 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

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

INFORMS is collaborating with JSTOR to digitize, preserve and extend access to Marketing Science

This content downloaded from on Sun, 29 May 2016 01:28:47 UTC
All use subject to

Order of Entry as a Moderator of the Effect
of the Marketing Mix on Market Share

Douglas Bowman * Hubert Gatignon
Purdue University

Order of entry has been demonstrated to have a significant effect on market share. A number of explanations for this effect have been suggested in the
marketing and strategy literatures. To date, the market share advantage gained by pioneers has typically
been treated as a main effect-an automatic regularity. Treating order-of-entry as a main effect implies
that there is no penalty on the effectiveness of a
brand's marketing instruments for late entry and
that a late entrant can compensate for being late by
dedicating sufficient marketing resources to their
In this study, we investigate the influence of order-

of-entry into a market on the effectiveness of a firm's
marketing mix decisions by asking the question,
"Can followers compensate for not being first by
their marketing mix decisions?" Also, even if they
can compensate for being late, does this effort become increasingly more difficult with later entry?
That is, are there asymmetries in the effectiveness of
a brand's marketing mix variables that relate to its
order of entry into the market, or as has been typically assumed to date, is order of entry strictly a
main effect? An asymmetry exists, for example, if the
market response to advertising is different for the
first entrant versus the second or third entrant. An
asymmetry also exists if the effects of, say, a price
change by the first entrant on the second entrant are
different than the effects on the third entrant. We develop a market share attraction model where the parameters vary as a function of order-of-entry. Our
main contribution is in modeling the sources of
order-of-entry advantage as asymmetrics in the

effectiveness of a brand's marketing instruments.
Hence, distinct from previous research we explain
why there are inherent order-of-entry effects. This
paper is potentially of interest to researchers developing market share models and studying the effectiveness of marketing-mix variables. The substantive
implication of our results concern directly academics
interested in marketing strategy as well as the practicing marketing strategists.
We model asymmetries in the market response of
early entrants versus late entrants using data from
two durables and three nondurables categories. With
one exception, all data sets are established from the
inception of the category and hence do not suffer
from the possible bias of excluding pioneers who
have failed. Results show that asymmetries in the ef*fectiveness of a brand's marketing mix variables are
an essential source of order-of-entry effects; we find
that the main effects of order of entry are minimal.
Order-of-entry effects do not necessarily lead to
lower shares, but overcoming these effects is not
without substantial cost to the late entrant.

Our results support previous research that has
demonstrated advantages to early entry. In addition,
we provide guidelines for how late entrants should
compete. Later entry tends to reduce a competitor's
price sensitivity, suggesting that they not instigate in
a price war with earlier entrants in order to gain
share. Order-of-entry tends to decrease response to
quality and to promotion. To achieve the same impact
on market share, later entrants need a bigger change
in quality and need to spend more on promotion. Our
data did not support an asymmetric effect on advertising.

(Marketing Mix; Competitive Strategy)


MARKETING SCIENCE/VO1. 15, No. 3, 1996

Copyright ? 1996, Institute for Operations Research

pp. 222-242

and the Management Sciences

This content downloaded from on Sun, 29 May 2016 01:28:47 UTC
All use subject to

ever. Moore.. Here. are there asymmetries in the effectiveness of a brand's mar- keting mix variables that relate to its order of entry into 1992) advantage. AT&T obtained 10. we examine asymmetries in the effec- achieved by the pioneer. Empirical evidence suggests that. (1986) find a significant response to advertising is different for the first entrant order of entry effect on the market shares of brands that versus the second or third entrant. Robinson 1988. Introduced in March 1990. though General Electric now dominates the ously in the literature to model sources of asymmetries market. or as has been typically assumed to date. if the market and advertising. it is not clear what these firms had to do to development. Carpenter et al. EMI.45. is general. we tiveness of the marketing instruments and with suffi- examine the relationship between order of entry and cient resources. The modeling of have been in an industry for at least three years. 29 May 2016 01:28:47 UTC All use subject to http://about. introduction strategies. does this effort become in- creasingly more difficult with later entry? That is. Bond and Lean (1977) results of our empirical analysis. We then elaborate on the moderating role share superior to the pioneer. 1986. Kalyanaram and Urban tiated products supported by heavy promotions can overcome the handicaps of being late.ORDER OF ENTRY AS A MODERATOR OF THE EFFECT OF THE MARKETING MIX ON MARKET SHARE Order of Entry as a Moderator of the Effect of the Marketing Mix on Market Share Considerable theoretical and empirical research has been directed towards the competitive advantage gained by the pioneer into a market (Weitz 1985). This raises two fundamental questions.jstor. No. the first entrant into the ulcer therapy market. and market For example. Blattberg and Wisniewski 1989) indicates that asym- Whitten 1979. For these differences across brands by the brand's order of example. it is argued that the rewards translate to higher market shares and larger profits. Treat- asymmetries is particularly important for the timing ing order of entry as a main effect implies that a late and the marketing plan of an entry if the asymmetries entrant can still compensate for being late by dedicating are systematically related to order of entry (Kalyanaram sufficient marketing resources to their product. Cooper and Nakanishi 1988). even if they can compensate for being late.5 million customers within a year following introduction. How- and Urban 1992). and Goodstein 1991) or orderof-entry (Urban et al. Urban et al. This content downloaded from 196. MARKETING SCIENCE/VOL 15. the market share advantage achieved by pi- competitive phenomenon. We develop a model to test our hypotheses. early followers can expect to achieve at most a order of entry strictly a main effect? market share that is no greater than 60% of that In this paper. after controlling for differences in positioning entry. Research on pected to achieve at most a market share equal to 40% market response modeling (e. while late followers can be ex- tiveness of a brand's marketing mix efforts. An asymmetry exists. the late entrant can achieve a market performance. 3. was able to capture a substantial share of the credit card market by entering with heavy advertising and low price. we intend to explain oneers has typically been treated as a main effect. in the market. We conclude with a in the pharmaceutical market and Whitten (1979) in the discussion of the implications of our findings and sug- cigarette market found that late entrants with differen- gest some directions for future research. First. in these models there is no penalty on the effec- The paper proceeds as follows. taking fourth rank among the most pop- ular bank cards (ABA Banking Journal 1991. AT&T for ex- ample. for example. 1988. The Marlboro in the plain filter cigarette market (Whitten model is based on a market share attraction model 1979).48. Boulding. Wall Street Journal 1991). keting mix efforts. A number of recent studies provide empirical evidence for a pioneering (Biggadike 1979. 1986).org/terms 1996 223 . While it is expensive (Urban and Hauser 1980) and risky (Lieberman and Montgomery 1988) to be a pioneering brand. a British firm was first in the CT scanner (multiplicative competitive interaction) used previ- market. metries between brands exist and can be an important To date. Urban et al. we present our data and then discuss the circumvent their early handicap.g. Following the model However. Such cases where the first of order of entry on the effectiveness of a firm's mar- entrant loses its leadership position do indeed occur.130 on Sun. Viceroy lost its position to Winston and share. can followers compensate for not being first by their marketing mix decisions? Second. In the next section. 1988. gamet. and Glaxo's Zantac surpassed Smith-Klein's Ta- and explore their strategic implications (Carpenter et al. Robinson and Fornell 1985. of that achieved by the pioneer (Bond and Lean 1977.

g. found that the likelihood of success for the third Empirically. That is. they found tendencies in market offerings exist.jstor. a later entrant's product may diffuse faster. in a study of 112 new industrial products from 52 French MS = . cost of advertising. and / or lower price unless it costs more for late categories examined. In 17 higher and why its price is lower.130 on Sun. and more / different marketing mix variables and to examine 224 MARKETING SCIENCE/VOL. Lilien and Yoon (1990). or entrants should offer better quality. the literature suggests that the advantages of early entry dominate the disadvantages. and early entrants must quality and advertising. higher than that of the first and second entrants. must trade off the possibility of high effect approach provides only a partial explanation for return if successful. Urban et al. for No. but was also pioneers and followers through its influence on price. the existence of a pioneering advantage-it does not. For example. if. these failure rates differ from nonpioneers in the line. When a product to vary systematically with a brand's order of entry into first appears on the market. a broader product how. pioneering advantage has typically been mod- Later entrants' products may diffuse faster than the pi- eled as a main effect (e. However. Rob- Empirical Evidence of Order inson 1988). some consumers may be the market. if such regular (47%) of 36 product categories examined. it is not clear why that the brand with the first instance of sales no longer followers fail to act on these differences. Golder and Tellis (1993) question the conclusions from for example.i3Quality + . the main effect model suggests an automatic regularity without explanation. in the most profitable segment and can initially realize monopolistic profits. have been estimated using cross-sectional data. uncertain about how to evaluate the product. Pio- was defined as whether or not a new product grew into neers have better product quality. The studies discussed above that model order of entry as a main effect or model pioneering as a recursive sys- tem.45. (1) While. 15.60 + .BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share the cost of production (Robinson and Fornell 1985. a recursive influence has been pro- of Entry Effects In general. market share has been found incur the cost of educating consumers. That is. ing at a later date when consumer uncertainty about category benefits is lower and consumer confidence in the evaluation process is higher (Gatignon and Robert- son 1985). By entering the market first. say. Kalyanaram and Urban (1992) examined order A number of studies using data from the PIMS database of entry effects using time series and cross sectional data has suggested that the market share advantage for pi- for frequently purchased consumer products. uct knowledge (Sujan 1985). They ex- oneering firms is due in part to order of entry effects on tend Urban et al. (1986) to control for the effects of factors such as product quality. late had sales in the category. indicate why the quality of the pioneer is empirical studies that exclude nonsurvivors. This content downloaded from 196. Biggadike 1979. the pioneer can choose to position itself posed whereby pioneering influences. the recursive trant. On the other hand. late entrants face less uncertainty of market demand and less consumer resistance towards adopting a new innovation (Gatignon and Bansal 1990). and lower relative prices. The first en- lines. entrants to do so. and relative product line breadth. as discussed below. 1996 . therefore. Also. Indeed. Main Effect of Order of Entry Order of Entry and the Diffusion Process To date. after controlling for differences in. broader product a product group for the introducing firm. 29 May 2016 01:28:47 UTC All use subject to http://about. 3. a number of competing explanations for an order of entry effect have been proposed. Re- Recursive Effect of Order of Entry cently. Success relative quality.. against the risks of premature entry. Robinson and Fornell (1985) find that order and fourth entrants into a market was higher than for of entry affects the difference in market shares between the fifth and sixth entrants as expected. It is not clear however. oneer's product since consumers have little initial prod- 1986). By enter- MS = OrderolQualityO2Advertising3 . relative product quality and the relative prices that brands charge.48.62Price + (2) where: Quality = a0 + alPioneer + Price = 60 + 61Pioneer + firms.

More promising directions focus on con- neering advantage consists of two components: (a) an sumers and the issues of preference formation and con- order of entry effect or permanent (constant over time) sumer learning mechanisms (Carpenter and Nakamoto share advantage which is greatest for the first entrant 1987. this model demonstrates Order of Entry Effects and Marketing the importance of controlling for diffusion. and may be different when the brands have been in the mar- have been challenged as to whether or not they are sus- ket. Mahajan. They could be more critical for durable goods. quality advan- effect has been raised (Fershtman. In fact. barriers such as brand two-month period and the followers within one month. as indi- atively small for frequently purchased consumer prod- cated earlier.. 3. and average. these are developed in the learn about and adjust to the presence of all brands in context of each marketing mix variable. an approach of entry has no main effect on long term market shares. diffuse faster) than earlier entrants. if the pioneer initially enjoys advantages require data on many competitors in a market. their model does not guarantee that the market of entry on the effectiveness of a brand's marketing de- share of the follower is always lower than that of earlier cision variables as an explanation for differences in the entrants-it may depend solely on the lag between en- market performance of the brands. Brown and Lattin 1994) or by selecting a sample to explain why order of entry effects exist. 15. then the final steady state their existing shares). Brown and Lattin (1994) argue that the pio- Muller 1990).. and entrants are lower. that have been in the market for at least three years (Ur- These explanations. the effect pirical support (Lieberman and Montgomery 1988).130 on Sun. This content downloaded from 196. No. This study represents an important step towards un- advertising costs. and (b) a com- sumer decision processes (Kardes et al. the stability of con- ple with the first brand achieving its potential within a sumer preferences (Bain 1956). 29 May 2016 01:28:47 UTC All use subject to http://about. versus 15 or 20 years on tainable in the long term (Fershtman.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share the influence of order of entry on a firm's market share the market. a recursive process has been advanced ucts. whereby pioneering firms are shown to have achieved The question of whether order of entry is a long run their position through cost advantages. and product position- Therefore. and Muller (1990) growth rate. MARKETING SCIENCE/VOl. The key ponent which is related to a brand's time in the market theoretical mechanisms leading to order of entry effects and whose effect dissipates over time as consumers are described briefly. The lag between entries could be the sole reason for the follower having a different market share than the earlier entrant. Their results indicate that all else market shares do not depend on the magnitude of these being equal. This control Mix Effectiveness can be done either by explicitly modeling the diffusion A number of theoretical arguments have been advanced (e. 1993). and Muller tages.45. the effect may depend on the time the lems in definition (Montgomery 1988). however. but one of a higher level of goodwill. say. but instead it is the effect of order of entry on price elasticity. production costs. order of entry alone does not influence and they confirm that the asymptotic values for later long run market shares. five years on average. 1989. that allows the use of cross category data and does not In particular. Robinson 1988). Then. They examine market share and marketing model a duopoly and identify conditions where order mix variables relative to the first entrant. How- Our interest is in investigating the influence of order ever.e. diffusion benefits or handicaps might be rel- ing (Hauser and Shugan 1983). Mahajan. Mahajan. derstanding the dynamic effects of order of entry. 1986). them. For example. and lower advertising costs draws share from existing competitors in proportion to which dissipate over time.g. brand loyalty. try. from weak em- brands have been in the 1996 225 . and on con- and smaller for each subsequent entrant. Fershtman. Early (main where diffusion effects are minimal.jstor. More recently. That is. Kardes and Kalyanaram 1992). Nevertheless.48. Kalyanaram effects) explanations centered around the risk of trying and Urban (1992) found a quick diffusion for their sam- new brands (Schmalensee 1982). that implies a constant ratio model (a new entrant lower production costs. later entrants approach their eventual share advantages or on the length of time it took to overcome levels faster (i. names and patents (Porter 1980). lower price. While such an effect has been shown for brands factors (Robinson and Fornell 1985. and consumer information or customer switching 1990). have suffered from prob- ban et al.

Brands that later entrants are likely to be ineffective (Carpenter and are preferred or prototypical of the product class tend Nakamoto 1989). Instead. While research on decision making and choice has typ- Two factors favor the prototypical brand. the pio- position of various brands and the order in which the neer is much more likely to become prototypical of the brands are presented. greater opportunity to contribute to the initial learning mechanism.45. exposure to brands many overlapping features. Therefore. and and attention-drawing for the earlier entrants. these features. learning process is influenced by both the attribute com- In addition to its role in preference formation. shares the most features with its category and the fewest In most decisions. First. when a ically focused on the limited capacity of short term new instance becomes available.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share Theoretical Mechanisms for Order of Entry Effects eration sets are being formed (Alba and Hutchinson Two behavioral mechanisms are relevant to explain 1987). Learning (Kardes and Kalyanaram 1992).jstor. we would expect a dif- product category (Carpenter and Nakamoto 1987). and acted upon by consumers. earlier entrants have a dently held judgements.48. such Sujan and Bettman (1989) discuss the impact of new brands enjoy a memory based advantage relative to information on the general product category schema their competitors (Alba and Hutchinson 1987). while novel presents consumers with an opportunity to learn. Hence. This suggests that "me-too" strategies by ered and the heuristics used to process it.1996 . they discuss familiar. In particular. because brands in a category have son 1987. Since brands enter the mar- and the pioneer benefits from more extreme and confi- ket in a sequential manner. (Kahneman and Snell 1988). prototypicality implies that the pi- and fundamental as to cast doubt on any purely oneer is that brand most closely associated with the stimulus-based decision making. Familiar brands are more likely to be percep- the schema and tag model-an appropriate model tually enhanced. suggests that later entrants are at a disadvantage. an individual com- No. which gives them a competitive ad- when the features of the new brand entering the market vantage in the "race" for consumer attention as consid- are moderately discrepant to the generic schema. Hence. memory factors play a crucial role. and a brand cannot be selected if it is not consid- order-of-entry effects: learning mechanisms in memory ered. Secondly. features with other categories (Medin and Smith 1984). This content downloaded from 196. as they are updated through the learning mechanism learning about brands decreases with order of entry. a pioneer does not react to estab- Categorization. Individ- by shifting the taste distribution towards its position uals tend to organize their belief systems about prod- and by influencing the attribute weights buyers use to ucts around category-related factors rather than a set of evaluate brands (Carpenter and Nakamoto 1989). 3. They 226 MARKETING SCIENCE/VOL 15. but also the type of information consid- prominent. The ferent preference structure to have evolved if the entry prototype for a category is that one instance which orders of the various brands were altered. and Lynch (1990) argue pares it with the prototype to see if it belongs to the that the effects of long term memory are so pervasive category. The single brand attitudes or beliefs (Cohen and Basu 1987). This when a new brand enters with a differentiation strategy. Categorization involves compari- lished preferences. later entrants positioned only the amount of information that enters into a deci- "close" to the pioneer may make the pioneer more sion process. Pioneers tend to be more Drawing on O'Sullivan and Durso (1984). Information search and information integration also and categorization. Hutchinson. Sujan 1985). They The emphasis is on how consumers cope with infor- have to "work harder" to have their messages heard mation incongruent to their present mental schema. Memory affects not product category. Hence. are re- consumer preferences are expected to evolve over time dundant and uninteresting for later entrants. Sequential exposure to about a new product category is often difficult for con- information about different brands produces differen- sumers. Alba. to be recalled more frequently and more quickly. 29 May 2016 01:28:47 UTC All use subject to http://about. it develops the best position son between a target and category knowledge. consumers tial learning about brands as a function of their order of likely have little initial knowledge (Alba and Hutchin- entry. If the category is particularly novel.130 on Sun. However. influence the magnitude of the pioneering advantage Memory and Learning Mechanisms.

They found that new ready-to- butes to the brand. Later entrants Srivastava 1979. then later entrants are at a disadvantage. They argue that the consideration set of a rational consumer will represent tradeoffs between decision costs and the incremental benefit of choosing from a larger set of brands. A strongly associated with the organizing schema. The decision to evaluate a brand for inclusion in a consideration set is modeled as different from the decision to consider an evaluated alternative. Shocker. b. adver- tising effectiveness. evoked set composition. we now describe how order of entry influences lationship in Equation (3). 15. and distribution effectiveness through its effect on the priming of brands. Therefore.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share predict that the new brand will be seen as having many lower than that for earlier entrants. that brand has the advantage of being perceptually distinct.2 A number of empirical studies have shown that the size of the considera- tion set tends to be small relative to the total number of available brands (Hauser and Wernerfelt 1990). memory will decline faster over time for tive) price elasticities.. or differentiating features will decline faster over time will be moderated by its order of entry. This content downloaded from 196. similar finding would be expected if later entrants were if the pioneer is more closely associated with the cate- successful in achieving a subtyped position where they gory schema. in general. is likely to be viewed as prototypical. b). but will also consistent with the empirical results of Ghosh. ket.6p(i). MARKETING SCIENCE/VOl. Carpenter and Naka- are strongly discrepant from the general category (Day. This argument is features common to the generic schema. Gatignon. A brand that contributes greatly to the process by which a consumer becomes aware of and learns about a product category. and moto provide experimental results to support this argument. while the prototype becomes more distinct. retical base to understand this phenomenon.45. Hence. appeal to less price sensitive consumer segments. Neslin. No. In partic- I Oft/0O I < 0. i.e. product quality is typically measured in terms of are at a disadvantage when competing on price since attributes or benefits (Lambin 1970a.130 on Sun. price effectiveness. and Bultez market response to a price change by a later entrant is 1975.48. Weitz. then. Carpenter and Nakamoto (1989) argue that later entrants positioned close to the prototypical brand become less distinct. memory for its unique Therefore.jstor. Since consumers of- ten organize their product knowledge based on the pro- totype. Also. any price reduction by the later entrant has less impact than a price reduction ' A subtyped position is one where perceptions about the competitor by the (more distinct) prototype.. Lambin 1970a. . 3. The decision to less distinct.g.(0) (3) of entry on the effectiveness of a brand's marketing de- cision variables. Hauser and Wernerfelt's (1990) evaluation cost model of consideration sets suggests that later entrants must Order of Entry and Price Bond and Lean (1977) and Carpenter and Nakamoto (1989) find that later entrants often fail to gain market share even with substantial price cuts. Naert. all else equal. That is. Sujan and Bettman 1989). Our interest is in investigating the influence of order . there should be asymmetries with re- spect to the market's response to price. Lambin. Oi. Hence. the prototype (perceptually) overshadows later entrants. Because the later entrant is provide a higher quality product. 29 May 2016 01:28:47 UTC All use subject to http://about. 2 Product quality typically refers to the characteristics or attributes of the product which correspond to consumers' needs (e. they predict that for a differ- eat breakfast cereals tended to have lower (less nega- entiated brand. where the function fp represents the re- ular. The decision to evaluate a brand entails a trade-off between the cost of search and the expected benefit from including the brand in the consideration set. brand i's price sensitivity. This means that late entrants need the discrepant (tagged) brand attributes than for con- to decrease their price to a greater extent than earlier sistent brand attributes because the former are not as entrants to achieve the same effect on market share.' For a differentiated late entrant.. Concepts from the memory and cate- and we suggest Hypothesis 1 that a brand's price sen- gorization literatures introduced above provide a theo- sitivity is a decreasing function of its order of entry. contain a unique tag linking the differentiating attri- and Shoemaker (1983). and cognitive processing effort.6P(i) = f. and Bansal 1990).org/terms 1996 227 . Order of Entry and Product Quality A consideration set is those brands that the consumer seriously considers when making a purchase decision. market response to quality. into the mar- than those associated with pioneer (category schema).

29 May 2016 01:28:47 UTC All use subject to http://about. and the degree of competitive Priming a brand. The objectives of advertising include position- ing (Nedungadi 1990). 6q(i). i. For a brand to be selected in memory-based 228 MARKETING SCIENCE/VOl. perience little indirect benefit from the earlier pioneer's In general. The Hauser and Wernerfelt (1990) model hence the benchmark upon which new brands entering predicts an order of entry penalty to late entrants. Direct cal) (Popkowski Leszczyc and Rao 1990). More importantly. When the market are evaluated for inclusion as a member of a new brand is evaluated or enters the consideration set. Such a retrieval is most likely for ing and creating awareness. Cohen and in the consideration set. No.g. not only a particular identification of a product. Recent research suggests that recalling one brand in a . Memory theories on choice also provide a rationale for an order of entry effect on advertising effectiveness. (4) product category can facilitate or dampen the recall of We propose Hypothesis 2.g. This influence on the other has a moderating effect on a brand's advertising effect- brands in the subcategory is referred to as indirect prim- iveness.e. Ofq/ O < 0. brands are evoked and how brands are evaluated. will be moderated by its order of entry.130 on Sun. 15. For example. low versus high involvement) (Krish- increase the probability that the brand is retrieved. that the market response to other brands and that the effect is contingent upon how quality improvement efforts decreases with order of en- the brands are grouped in memory (Alba and Hutch- try. more prominent or accessible brands in memory can be evoked both through direct Order of Entry and Advertising Effectiveness priming and indirectly through the priming of another A number of previous studies have proposed factors brand in the subcategory (Nedungadi 1990). the product category. Later entrants must provide be at a disadvantage.48. national versus lo- function of brand and subcategory activation. The Nedungadi (1990) studies memory-based choice sit- categorization and memory literatures. into the market: advertising. the nature of priming refers to providing an external cue intended to the product (e.g. Pioneers may indirectly benefit larger incremental benefits in order to be fully evaluated from later entrants' advertising. we propose that order of entry brands in the subcategory.. This suggests that later entrants may their knowledge base. namurthi and Raj 1985). category level advertising) and ticular. the off between the expected benefits at consumption and pioneer is likely to become prototypical of brands in the the incremental decision costs of choosing from a larger category (Carpenter and Nakamoto 1987. inson 1987). by increasing subcategory activation. The market is more demanding Basu (1987) argue that marketing efforts to create and of later entrants. pro- uations where changes in a brand's accessibility may vide a useful framework for studying how order of en- affect the probability that it is retrieved and considered try influences advertising effectiveness.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share include a brand in the consideration set entails a trade- Through its role in category preference 3. later entrants must provide increasingly larger later entrants would benefit only from brand specific additional incremental benefits to ensure their inclusion cues (Sujan and Dekleva 1987). therefore.6q(i) = fq(Oi). In par- product type cues (e.. while later entrants ex- by consumers. for choice. has implications for which ing of that particular brand. 1989) and set of brands. and later entrants must provide higher communicate specialized product characteristics create levels of product quality. in turn. from a may also increase the probability of retrieving other strategic point of view. but also The information integration model proposed by the increased salience of information relevant to the Kardes and Kalyanaram (1992) argues that consumers product category together with category based infer- fail to integrate information about later entrants into ences that result. This content downloaded from 196. This.. Minor spective. The prob- which may moderate advertising effectiveness. Oi. reactivity (Gatignon 1984). This would suggest that the pio- its perceived incremental utility must exceed the (dis- neer would benefit most from any product class or counted) evaluative search and / or decision cost. From the consumer's per- the more prominent brands in the subcategory. 1996 ..jstor. Similarly. These ability of retrieving a particular brand is a multiplicative include the type of advertising (e. advertising influences perceptions and facili- brands in a category benefit primarily from direct prim- tates recall.45. the market response to brand i's product quality.

1992. that advertising Data used in previous studies on order of entry effects effectiveness is a decreasing function of order of entry has been criticized for including only surviving brands into the market. Golder and Tellis 1993). 1992). the consumer must recall that brand and fail to venience goods industries characterized by low prices recall other brands that might otherwise be preferred. it is necessary to control for the first influence- sons 1975).e. Hence. time series of cross sections are required as order of entry increases. In this case. it is suggested that advertising is most effective under less cognitive processing (Gatignon and Robertson 1985). pioneers may be able to Much of advertising serves to prime a particular brand. dominate available retail space. ing elasticity declines over the product life cycle (Par- Hence. but instead typically continue to ing brands.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share choice.jstor. multiple brands are necessary to investigate order- of-entry effects as the number of competitors in a mar- amounts of information consumers must process when faced with a ket changes and multiple markets are desirable for gen- large number of differentiated products from which to choose.48. In fact. indications are that primary demand advertis- rable categories examined. This would be consistent with decreasing differences in available retail outlets. (5) Data In particular./ 090 < ? (Lieberman and Montgomery 1988. This content downloaded from 196. Oi. 15. level of distribution coverage which. Mahajan. 3. for considering only cross sec- Order of Entry and Distribution Robinson and Fornell (1985) argue that pioneers may have access to more efficient distribution channels or be able to prevent competitors from using the distribution channel entirely. f. could based effects.130 on Sun.45. and high purchase frequency. affects pa(i). achieve lower market share solely due to their limited The influence of order of entry on advertising effect- distribution. the brand elasticities as new brands enter the market. we propose Hypothesis 3. 1996 229 . the diffusion literature suggests that pio- to test alternative or complementary explanations for neers often bear the expense of category development (Rogers 1983).(t) = f(Oi). order of entry effect on distribution is a recursive effect These arguments suggest that the effectiveness of in the sense that order of entry is a determinant of the brand i's advertising expenditures on its market share. For the du- Finally. then the consumer is biased in favor of later entry. Later entrants compete in increasingly capture or dominate the most effective channels of dis- crowded markets and thus their message must be heard tribution. i. pioneering distribution efforts may be lower than that of pioneer- brands are not passive. Later entrants are at a disadvan- tribution channel entirely. Appropriate datasets should contain both new entrants and brands withdrawing from the market. the effectiveness of a later entrant's over increasing amounts of "noise. and for (with few exceptions) study- ing only a single product category (Kerin et al. and for late entrants by virtue of the vast elasticities could not be estimated without time series data. Cognitive processing tends to be higher for pioneers by virtue of their entering when category uncertainty is high. they have access to fewer buyers. familiarity or wholesalers) buyer attitudes become less favorable with prototypicality. This is supported by If some brands are more prominent or accessible in Alpert et al. particularly when the brand is the second or including those brands in their consideration set. due either to their positioning." Also. the channel is a captive one. and they should have both cross sectional and time series obser- : In contrast with this rationale for decreasing advertising elasticities vations. if and indirect brand priming. they suggest that in con- tional effects (Fershtman. pioneers. In this study. the market: p. Kerin et al. and Muller 1990. into market share. 29 May 2016 01:28:47 UTC All use subject to http://about. the later me-too follower brand. ten) well established distribution channels. Sec- iveness is not limited to categorization and memory- ondly. in turn. eralizability purposes. (1992) who find that reseller (retailers and memory.. products are sold through (of- advertise and improve their product and positioning. Kerin et al. early entrant benefits from both direct brand priming Two influences of distribution are important. then late entrants could tage due to their advertising being less effective. In addition. No. changes in elasticity over time and across brands: brand In addition.. since our interest is MARKETING SCIENCE/Vol. 1992). In particular. will be moderated by its order of entry. This implies asymmetric pioneers are able to lock out later entrants from a dis- effects of advertising. by virtue of their earlier entry.

for tomotive News. and five withdrew. This content downloaded from 196. Wildt 1974). some pricing much below existing com- petition. Five product categories satisfying these A brand's entry period into the market was defined criteria are examined-two categories of durable prod- as the first period of sales. the Ford Ex- Following Urban et al. the attraction model also represents the fact that configurations were available. considerable competitive the market and hence its order of entry. in practice. 230 MARKETING SCIENCE/Vol. order is determined using plorer in 1990. important role. For each model year. For example. 29 May 2016 01:28:47 UTC All use subject to http://about. more for each brand. The two durables categories are the sport utility final period of sales. peal to a wider range of consumers. and as the vehicles began to ap- servation period. and many additional luxury options were offered.. 16 brands entered and three brands exited. years prior to 1983 and brand descriptions found in Au- Durables (Automotive) Data Automotive Yearbook. variety of optional equipment ordered across custom- Market Share and Order of Entry Data. (1986). When our obser- issues prior to 1980. likely offered uniformly across brands or customers. it is impossible to control for the the pioneer. Our observation period includes vehicle market and the minivan market from 3. Lambin 1970a. To de- We also analyze data from three nondurable categories termine the order of entry for brands in the sport utility that have helped establish the relationship between or- vehicle category that entered prior to January 1983. model specification. Ward's Automotive sions. However. 1972a. prices may not be for comparably lowed two months later by the Chrysler Caravan and equipped vehicles. any new product introductions with each manufacturer. Ward's Automotive collects monthly unit sales data for all brands of automobiles and trucks sold within the United States.g. During the period 1983-92. Table 1 presents the order of entry. the inception of the minivan product category. b. which was replaced by the Toyota Previa. a brand's entry pe- the market (three of which were line extensions). ers. some brands became available in 4-door ver- Price Data. the base price is used as in the majority of aggregate econometric models (e. competitive rivalry average unit sales per month. Ward's did not provide detailed vation period began in 1983. rivals entered.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share in testing hypotheses about the influence of order of The market share for each brand in any particular entry on the effectiveness of a brand's marketing mix month was calculated as the sales for the brand during variables. and average market share increased (Wall Street Journal 1990). Table 1 presents the price data approaches saturation. For these years. 17 new brands entered. While the product category sport utility vehicles was they were used to determine a brand's entry period into established over 20 years ago. Therefore. publishes the suggested base price of each vehicle. and in the appropriate issue of Ward's activity took place during the 1980s. we der of entry and market share. the midpoint of the range the marketing mix elasticities of a brand decrease as the market share of base price offerings. As the market got only those brands that survived to the start of our ob- increasingly crowded. The price of a brand for a particular model year was taken as the base price.4 means that customers seldom pay this price. including However.48. 15.130 on Sun. Lambin and Dor 1989. When sales data were available. including the current market share leader. 1996 . or when a number of base model 4As discussed below.jstor. A brand's exit period (if ap- ucts and three categories of consumer nondurable prod- plicable) from the market was defined as the brand's ucts. Dur- riod was determined from the Yearbook's description of ing the 10-year period 1983-92.45. The These changes in competitive offering and the conse- wide variety of optional equipment and proliferation of quences on market share are captured by an attraction manufacturer and dealer pricing incentives. nine brands competed in monthly sales data. for each brand. b. Also. any pricing incentives are not Plymouth Voyager brands. Since The minivan category was established in 1983 with equipment that is optional on one brand may be stan- the introduction in September of the Toyota Van fol- dard on another. however. No. the markets under study should be ones the month divided by the total sales of all the brands in where a brand's marketing decision variables play an the category during the month. These data were used examined both monthly sales data (when available) for previously by Kalyanaram and Urban (1992).

. The substantive conclusions from our analyses do not change if we establish order of entry based on the first usage of the brand name within any automotive category.7% 18.419 Bronco-Il3 9 . and Previa replaced the Toyota Van.590 Mitsubishi 7 11/86 3/92 262 0.16 658 1.11 753 4.8% 13.569 9.72 0 4.25 565 2.235 1.237 2. 3Blazer-S.705 Montero 10 3/83 - 467 0.4.90 1.510 . MARKETING SCIENCE/VOl.00 1.128 1.117 3.458 Ramcharger 5 .939 Silhoutte 12 10/89 .8% 12.2% 14.429 26.3% 12.389 3.3.408 Aerostar 6 7/85 . 4Wagoneer is a brand extension-the Wagoneer brand name was also used for a truck introduced at an earlier date.1.9% 25.7% 16.19 312 1.924 2.521 2.016 2.908 Jimmy-S3 8 .2. for the Blazer.6% 7.126 2.537 5.518 3.634 3.458 3.1% 13.2.7% 7.755 4. This content downloaded from 196.) Share Price ($) Rating ($k/mo.988 Lumina 11 7/89 .94 1996 231 .063 Yukon 26 1/922 - 504 0.12/86 2.00 107 4.59 165 1.48 602 4.081 1.304 2. The mix of sales within the transition year was not available.0% 12.jstor.5% 19.2% 17.678 2..824 6.2.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share Table 1 Sport Utility Vehicle and Passenger Minivan Data 1983-92 Avg.3% 19.751 2.48.644 Blazer 2 .6% 15.3.854 Safari 4 11/84 .020 15.2% 14.3.2% 18. and $646/mo.624 0.450 2.976 5.298 .800 586 Rocky 21 11/89 . Sales Market Median List Avg.) (No.13.03 496 4.75 882 2.649 Quest 15 7/92 .334 Toyota Van 1 9/83 12/892 1.961 2. 29 May 2016 01:28:47 UTC All use subject to http://about.548 12.931 2.7% 16.154 223 Rodeo 22 7/90 .041 4.078 3.324 2.07 70 333 Nissan Van 8 1/87 9/91 604 1.995 Astro 4 11/84 .162 3.909 1.659 3.641 2.1% 14.965 Cherokee 6 .029 20.1.828 22.970 1.1% 11.025 Pathfinder 17 12/86 .93 674 301 Tracker 18 9/88 . Advertising for the related.085 2.566 31.01 1.3.042 478 4-Runner 13 5/84 .024 3.78 1.665 2.151 .3% 14.1.961 3.27 0 872 Town & Country 10 6/89 . for the Jimmy-S.00 731 895 Bravada 25 11/90 .2..039 3.583 2.978 28.50 11 3.108 Sidekick 18 9/88 .721 3.141 3.511 Blazer-S3 7 .2% 11.9% 16.3% 24.3% 13. Quality Advert'g Avg.993 4.2.3% 19. We attributed all sales to the new brand beginning in the quarter that sales of the new brand were discussed by the popular press.9% 13.532 Raider 16 8/86 6/90 909 1.6% 11.429 CJ 4 .9% 19.22.098 1 Brand introductions and exits that take place during our observation period are shown as month/year.364 2.696 Trans Sport 12 10/89 .13.2.28 341 260 Wagoneer 11 9/834 12/90 948 1.704 5.444 Navajo 23 9/90 .397 3.453 3.1% 13.45.613 2.1% 13.691 3.142 4.04 84 1.122 Samurai 14 1/86 .369 3.3% 18.222 3.55 78 2.3% 12.60 350 4.917 1.32 231 1.00 1.93 1.112 5.083 Previa 14 1/902 .2.1.45 617 246 Wrangler 15 4/86 .008 14.775 3.3.915 .977 Voyager 2 11/83 .755 6.39 1. Average Order Entry Exit Avg.143 Amigo 20 2/89 .1% - 2.532 2.822 1.445 2.8/902 9.284 .00 924 593 Explorer 23 9/902 ..5% 14.50 179 2. and Bronco-Il are line extensions.83 17 2.1.096 Caravan 2 11/83 .05 999 1.364 3.89 71 1.256 0. 15.175 2.6% 12.459 Trooper 12 2/84 .9.908 Jimmy 3 .1.0% 12.8% 17.202 2.29 604 3..12.454 3.405 2. earlier entrant (not shown in this table) averaged $347 k/mo.2.12.6. of Dealers) Bronco 1 .820 3.9% 14.192 2.938 2.22 347 4. 2 Yukon replaced Jimmy. NO.2% 15. $140 k/mo.130 on Sun.4% 15..626 3.186 Villager 15 7/92 .331 2.2. Jimmy-S.25 128 2.416 1.110 Mazda MPV 9 8/88 . 3.444 4. Dist'n Brand of Entry Period' Period' (units/mo.5% 12. Explorer replaced Bronco-Il.690 1. for the Bronco-li.3.12/912 596 1.621 2.

and the audience efforts to improve its objective quality. For this study. the simple two-point average was used as an es- timate of the number of retail outlets for each manufacturer throughout a calendar year. tributors was no greater than 10% above the minimum. the number of outlets remained relatively stable over the ten year observation period. With few excep- tions.. These measures reflect product quality information available to consumers. Curry and Faulds 1986.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share Product Quality Data. For 11 of the manufacturers. in all cases.5 quality. In the subsequent analyses. readers. Table 1 presents the average monthly number of new truck dealer outlets for each brand. Lich- hicle category-Bronco II (Bronco). we assume that product quality perceptions are influenced by published ratings of the prior year's model. Objective product quality re- scanned for advertisements. the broadcast station. Each item advertising for the brand under study. and Suzuki were the only manufacturers to ever have fewer than 750 outlets dur- ing our observation period. in addition to quality ratings of motor vehicles on 17 items. perceived quality is Broadcast Advertising Reports (BAR) collects advertising the consumer's judgement about the superiority or ex- data for a number of broadcast media including local spot cellence of the product (Zeithaml 1988). The two factors with the largest ei- genvalues were similar in both markets: a measure of body / engine quality and a measure of transmission quality. Retail outlets include dealerships.130 on Sun. Curry and Faulds 1986. body 3. Perceived qual- television. we examine product quality based 1 presents the average monthly advertising expendi- on data reported in Consumer Reports. No. BAR estimates brand specific factors such as price (e. Quality is measured by the items body exterior (paint). this value was between 10% and 20%. A factor analysis was done for sport utility vehicles and for minivans. Faulds 1986). and were the only instances where the print media including magazines and newspapers. An exploratory factor anal- ufacturers. data have been used previously in research as a measure There are three line extensions in the sport utility ve- of product quality (e. This content downloaded from 196. the maximum number of dis- ysis revealed a single factor structure. For this study. However. and factory-owned outlets. ity tends to be an overall judgement (Olshavsky 1985. We assumed that when a manufacturer such as Jeep produced multiple brands. 1996 . ner 1985) and advertising. tenstein and Burton 1989). Since quarterly data is available. Leading National Advertisers (LNA) collects advertising data covering a number of was between 20% and 30%. Blazer-S (Blazer). only the measure of body/ engine quality was significant. all the brands were available at all Jeep outlets. On the other hand. Hence. 232 MARKETING SCIENCE/VOl. 29 May 2016 01:28:47 UTC All use subject to http://about. Consumer Reports publishes and Jimmy-S (Jimmy). and radio. we also included is rated on a 5-point scale.g. Automotive News publishes the number of retail truck outlets by manufacturer as of January 1 of each year. Isuzu. national network television. a uniform tween perceived and objective price quality relation- monthly distribution within each quarter is used. for 2 manu- facturers. Based on the time of day an advertisement is Zeithaml 1988) that is influenced not only by a firm's aired.jstor. cable television. and engine mechanical. we present re- sults that include only this single dimension for product Distribution Data. It white). LNA uses advertising rate cards supplied by is that which is measurable and verifiable (Curry and publishers to estimate expenditures for each brand. Monroe and Krishnan 1985). Consumer Reports tures for each brand. We use the most re- cently published ratings on the newest model. All maximum number of distributors exceeded the minimum by greater major and most minor newspapers and magazines are than 50% over the 10-year period. Based on the size of the fers to the actual technical superiority of a product advertisement and its format (color or black-and- (Hjorth-Anderson 1984.45. Daihatsu.g. we examined including multiple measures of product quality. For these brands. body integrity.48. but also other covered (local or national). Table ships. 15. Lichtenstein and Burton The total advertising expenditure for each brand is (1989) (using Consumer Reports data as a measure of ob- the total of the broadcast media and print media expen- jective product quality) find a positive correlation be- ditures. Mitsubishi. electrical sys- 5 The sport utility vehicle brands were produced by 18 different man- tem. distributorships. The ratings represent fre- a measure to capture the influence of advertising ex- quency of repair estimates based on a survey of their penditures for the related entrant. its duration. and for 1 this value Advertising Data. Gerst- broadcast advertising expenditures. For parsimony.

This content downloaded from 196. The market share of brand i is then simply the ratio of the attraction of brand i to the sum of the attraction of all the competitors in the industry: A MSi the total of the broadcast media and print media expenditures. As described above for the durables categories. and cally related to order of entry into the market. Crest toothpaste. promotional expenditures. over-the-counter ibuprofen pain reliever. in the tartar control toothpaste cat- egory.7 and. tered during the period of no observation. Asymmetries in market provided their data. For example. 29 May 2016 01:28:47 UTC All use subject to http://about. 2. data. over the vertising and no barrier created due to advertising or period October 1983 to January 1988. and "related brand" advertising as any advertising for other Crest toothpaste products (e.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share Nondurables Data (1986) assume that the effectiveness of advertising is Kalyanaram and Urban (1992) examined penalties as- identical for all brands in the market. 6 The influence of order-of-entry on promotional expenditures was not explicitly discussed above. and and time series data and allows for asymmetries in mar- monthly distribution were available. Consequently.48. I) and MSi be its market share. excluding advertising expendi- response cannot be estimated with only cross-sectional tures. a uni- form distribution within each quarter is used. 3. and all other advertising expenditures for the brand name under study within the category. since quarterly data is available. A third competitor en- phenomenon is needed. Let Ai be the attraction of to the national roll-out date discussed in Advertis- brand i (i = 1.. advertising expenditures for the brand (A). Therefore. a model that is sensitive to this two noncontiguous time series. In addition. a model that uses both cross-sectional suggested price. The were useable for testing our model based on the criteria multiplicative competitive interaction (MCI) model that complete data was available for all major compet- (Nakanishi and Cooper 1974.. The attraction of a particular brand is a function of its ing Age. and Shoe- study entered during the observation period (required maker (1983) also suggest that the MCI model is appro- to determine a "time in market" measure discussed sub- priate for studying the influence of order of entry on the sequently). for seven categories.jstor. in eight cities. we control for differences in distribution (D). MARKETING SCIENCE/Vol..130 on Sun. No. for line extensions. We also estimated our model using threeyear cumulative expenditures (motivated by Urban et al. Measures for weekly sales. total advertising expenditure for each brand was determined using data collected by LNA as marketing efforts. Two of the nondurable categories are line extensions. in sociated with late entry using data from eight categories their models. However.g. Two measures of advertising were developed: expenditures that include a message about the specific line extension under study. promotion (PO) for non-durables. Again.45. Advertising Data. many of the mechanisms that contribute to a moderating role of order of entry on price and advertising sensitivity should also be applicable for sensitivity to promo- Model Specification tional expenditures. They generously other marketing mix variables. The ibuprofen data was in ket share and hence.6 product quality (Q) for durables. When both Crest tartar control toothpaste). Our interest is in testing whether asymmetries in re- sponse to competitors' marketing efforts are systemati- 'Similar to Gurumurthy and Urban (1992). Models cumulative expenditures with decay over the previous 36 months. and such as Kalyanaram and Urban (1992) and Urban et 1996 233 . (1986)). there is no differential advantage of ad- of consumer packaged goods.e. Crest for kids). we used instantaneous advertising expenditures. The results of Ghosh. Three categories keting mix response across competitors is required. Neslin. find that price elasticity decreases with increasing mar- and frozen pineapple drink. did not find any improvement in model fit.. and that the first entrant in the markets under characteristics. and its entry A multiplicative form for each competitor's sales re- date into the observed markets was assumed identical sponse function is specified. .. we measured "brand" advertising as expenditures for the specific line extension under study (i. The categories used were: tartar-control effectiveness of a firm's marketing mix variables. - Aj=j = A' (6) We have identified five marketing mix variables- price (P). advertising expenditures of related brands carrying the same brand name (AR) -whose effectiveness may be influenced by a brand's order of entry into the market. They toothpaste. 1982) embodies these itors. 15.

However.45. the marginal change in marketing mix effectiveness due to order of entry may be lower for later entry. Nonlinear relation- ships are. 4. Nakani- shi and Cooper 1982). I. Al (i) is the price sen- number of competitors in the market may influence the sitivity of brand i. also examined. /33(i) is the market fect is likely to be most evident for large variations in response to quality for brand i.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share cross-sectional and time series data are available. the where 60 (i) is the constant term. 3. 29 May 2016 01:28:47 UTC All use subject to http://about. 64(i) is the advertising the number of competitors over the observation period. the effectiveness of each of its marketing mix variables (ex- cept for distribution) is influenced by its order of entry into the market: fkJO) = fk(Oi) Vk (E 1. + a110i ln Pit + a12NCt ln Pit + a20 ln POit + a210i ln POit + a22NCt ln POit + a30 ln Qit + a310i ln Qit + a32NCt ln Qit + a40 ln Ait + a410i ln Ait + a42NCt ln Ait + a50 ln ARit + a510i ln ARit + a52NCt ln ARit is less than the marginal change in response for a com- petitor entering as the third brand versus entering as the second brand. 2. Such an ef- motional expenditures for brand i. That is. therefore. Equations We have hypothesized that. for a particular brand. the 3.130 on Sun. A linear relationship is typically robust. /65(i) is the advertising effect- and we need to control for that possible effect. Hence. related to increased competition over time and related to nonobservable factors such as changes in consumer 234 MARKETING 8 A similar equation results for the nonlinear process equation (8b). and 66 iS the market response to that includes the order-of-entry term and a term to cap- distribution. the expected marginal change in response to quality for a competitor entering as the ninth brand versus entering as the eighth brand (8a) and (8b) show respectively a linear and a nonlinear specification: f3k(it) = akO + aklO? + ak2NC. No. (8a) fk(it) = akO + akl ln(Oj) + ak2 ln(NCt). This relationship can also be specified for the constant term 60 (i) to represent an advantage for early entry beyond asymmetries in response parameters. Traditional economics predicts that a market will become more competitive when the number of market share for brand i is given by firms in the market increases-Hauser and Wernerfelt (1990) argue that the average number of brands consid- MS Dee(7(i)p)(i)po82(i)Q3()A64(i)AR85()D6e3 ePit LI= ep jPt ipjt ''jt ti)A$4(i)ARjt j)Djt eP. the BDj's are brand specific dummy variables that take on a value of 1 if j = i (O otherwise).36 ln Dit + bjBDj + E XUITDu + pit (9) j=2 u=2 where NCt is the number of competitors in the market at time t. 1996 . The process function (8a) leads to the following model to be estimated:8 ln MSit = /0 + alo ln Pi. This content downloaded from 196.t ered by consumers is the relevant construct. The specific functional form of these relationships is not clearly derived from theory. effectiveness of brand i. and the TDu's are time specific dummy variables that take on a value of 1 if u = t (O otherwise).Lk(i) l.jstor. Such a relationship would be justified if increasingly crowded markets serve to attenuate the incremental penalties for later entry. a linear approximation is likely to be most valid in markets with few competitors where the order of entry variable does not take very large values. ture the effects of the number of competitors. /2(i) is the market response to pro- marketing mix elasticities of all the brands. iveness of advertising for "related brands" when brand The above discussion leads to the process function i is a line extension. 15. Elasticities may vary over time for reasons I T + . for example. and order of entry is decreasing. (8b) An MCI model is a special case of a log-linear model and can be estimated using log-linear regression techniques: with appropriate dummy variables defined be- low. A number of competing arguments for variation in the response to marketing mix efforts have been alluded to earlier. the ordinary least squares estimator is BLUE (best linear unbiased) (Cooper and Nakanishi 1988. More particularly.48. In markets with a large number of competitors. SCIENCE/VOl. 5}. we hypothesized that the relationship between market response.

31a -3.20 Sport Utility Vehicles 1.12a Sport -2.45.57 Related Advertising constant 0. B.82 Order 0.94 Category AC Category Bc -0.92 Order .06 -1.41 Category Category AC Bc -58. which is chang- lation does not reduce to simple differences of loga- ing over time.61a -14.30 ln(TimeMktNonD) 0.05 (based on two-tail tests).49 Category Bc -0. Category B (1-4).79 Utility Vehicles 0.35a 5.71 Category cc Quality constant -0. Category A (2-3). 3. the log-odds specification assumes log-odds equation because of the ease of interpretation. Cooper and Nakanishi (1988) favor this rithms of the dependent variables relative to a base model specification and estimation approach over the brand.01 (based on two-tail tests).91 RF2 NA Sport Utility Vehicles -1.07a -7. c A.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share Table 2 The Influence of Order of Entry on Marketing Mix Effectiveness Weighted Least Square Estimates Parameter Parameter Variable Process Function Estimates t Stat Variable Process Function Estimates t Stat Intercept* constant 63.97 Distribution constant 0. Category C (2-4).04a 8. Minivans: NumComp < 5.17a 2.01 -1. Minivan (1-13).44 n 3729 -3.01a 9. actual) 0. This The dummy variable specification of Nakanishi and interpretation is especially easier in the case of Cooper (1982) allows for the pioneer to exit the market MARKETING SCIENCE/VOL. of Competitors# -0. based on the log-odds specification (Theil 1969).05 0. the pioneer exists the market.30 Category CC -0. of Competitors# -0.31a 10.02 -0. of Competitors# 0.15 -0.90 Category cc 5. in a number of cases.75 ln(TimeMktDur) 0.05 No.81 No. % Quality data missing for the Sport Utility: NumComp < 13.72 Category AC 586a 8.26 -0.36a 4.45 Order -0.48.55 Category Bc 5.30 -0.85a -15.36b 0. 1996 235 This content downloaded from 196.06 Category cc 0. 29 May 2016 01:28:47 UTC All use subject to http://about.05b 2.69 Vehicles Category 0.39 Sport Utility Vehicle -37.40a 12.02 Category Cc -71.23 -14.35a -5. In estimates than would be obtained using Theil's method fact. of Competitors# 0.42 -2.16a 6.81 Sport Utility -0.04a 0.02b 2.061 1.jstor. of Order Competitors#% -0.59 Category Bc 0.03 -1.59a -10.14 No.84 Category Bc -0.32 Advertising constant 0.32 Category AC 0.77 Order Price constant _7.07a 3. b significant at 0.041 Ac 2.96 ln(Order) -0.80 p (predicted.18a 6.37a 2.17a 13. of Competitors: Sport Utility (9-21).03a -4.54 Promotion constant 0.01 -0. In addition. No. 15.01 a 11. C correspond to the consumer packaged goods categories. The time-dummy variables account for the logarithm differential-effects models where the log-odds formu- of the denominator of the MCI model.77a 9.32a -4. that the pioneer (or any base/reference brand) remains the estimation of equation (9) yields identical parameter in the market throughout the observation period. a significant at 0.05a -70.80 No.02 No.08 * time specific constants not shown.01 -0.02 -1.130 on Sun. Range for No.93a -9.

+ I E XUMTDII"l + Wimt (10) mn=1 u=2 where the process function for the intercept term is specified to include the main effects of order as .. has no interval properties. That is. a con- differences in the observation interval (monthly versus strained version of Equation (9). and there is no need treat the variable this way is motivated by parsimony. Because of al.130 on Sun. 15. for a U-shaped influence of time in market-an investigation motivated by the discussion of brand retrieval processes in Kardes et al. could be a function of order of entry (e.9 While Equations (6) to (9) are expressed in terms of Urban (1992).48. we pool our data across markets. (1986). No. Empirical Analysis and Results As discussed earlier. A Hausman test (Maddala 1988) Urban (1992). 29 May 2016 01:28:47 UTC All use subject to http://about. the decision to eled in the attraction . (1986). where the brand spe- weekly). this leads to a recursive system ln MSimt = Oo(imt) + /31(imt) ln Pimt + 62(imt) ln POimt + 63(imt) ln Qimt + f84(imt) In Aimt + 635(imt) In ARimt + 66(m) In Dimt + I 6jBDj j C Non-Durable so that Equation (10) can be estimated separately with- out biasing the parameter estimates (Dhrymes 1974).60(imt) = aoo(m) + a0o In Oim + a02 In TimeInMarketimtDDmi + a03 In TimeInMarketimtDDm2.. the model should that it would have a different parameter for each market capture the main effects of order of entry (e. Consistent with previous studies. Kalyanaram and Urban 1992).jstor. Further.g.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share while observations would have to be deleted in the log- has been in the market plus one.g. (1993). per se. To test for a diffusion effect in these data. Rob- kets which are represented by the subscript m: inson and Fornell 1986). and Brown and Lattin (1994). Most brands that entered during our observation period achieved sales at least equal to their average sales within the first few periods following entry. The dummy main effects of order of entry. Consistent with Urban et al. Thus. While order of tors has an impact on market shares is implicitly mod- entry. the dataset still had at least two-thirds of the observations following the period in which the brand first exceeded its average 9We also tested. 1996 This content downloaded from 196. following Kalyanaram and is a single equation. (11) Here. we also include a would cancel out with the same term which would ap- time in market variable to capture the influence of ad- pear at the denominator of the attraction equation vantages that accrue from consumer learning over (Equation 6). 3. and TimeInMarketimt is the number of periods the brand market. a separate time-in-market parameter is esti- cific dummies are replaced with a variable capturing the mated for durables and for nondurables. The intercept in the process function (aoo(m)) is indexed-by m to indicate odds specification. we examined the number of periods it took for a new entrant to attain at least their average unit sales for the observation period. including distri- term and base response parameters to vary across mar- bution. but did not find support. sales.. We es- indicated that distribution is exogenous. even timate a fixed effects model that allows the intercept though some marketing mix variables. Kalyanaram and time. The func- variables DDmi and DDm2 reflect the two types of market tional form of the main effects of order is consistent with in Equation (11). The fact that the number of competi- that proposed by Urban et al. A concern for our study is that a diffusion effect would depress the shares of later entrants during our observation period without corresponding to a real order of entry effect. Because the automobile industry typically introduces new models (under existing and new brand names) at 236 MARKETING SCIENCE/VOL. For those brands that took longer. 1986. ?im is the order of entry of brand i into market m. to add this factor in Equation (11) because the term Following Brown and Lattin (1994). initial sales are relatively low due solely to the fact that the brand is a recent entry into the M T. Urban et estimated using dummy market variables. is estimated. our model a single product category.45. in some markets sales of new brands experience a diffusion effect (Parker and Gatignon 1994). Such an effect should diminish over time.

although the pattern applying to each competitor's had similar values within each market. 1996 237 This content downloaded from 196.3. A test of the functional form of the process function was performed. brands within each category does not significantly im- Because the product quality data is based on a survey pact the sum of squared residuals (F = 1. Because of the nonnested na- ture of the two models. The linear process function (8a) model was estimated as well as the model with the logarithmic process function (8b) specification.130 on Sun. This indicates no significant differences in the each year. of Consumer Reports readers and is published in April of p < 0. Period has no effect on market share. the R2 obtained from OLS tual market shares and the predicted market shares us- The OLS estimators are consistent and unbiased. therefore. The data tends to support the hypotheses.01). Therefore. Also.69. annual seasonality patterns may oc- brand dummy coefficients for the nondurable markets cur. based on the estimated logarithmic specification of the process func- "1 The procedure involves two tests. None of the signs of the significant variables is applied where the null hypothesis assumes that the logarithmic pro- changed and the magnitude of the coefficients remained similar. MARKETING SCIENCE/ .3. specification. However. the process functions also include terms to capture (potential) variation in response parameters due to the number of competitors in the market. An analysis of residu- quality data was available. We report the results of dummies. These two statistics lead to cation) model is estimated. 29 May 2016 01:28:47 UTC All use subject to http://about.13). rejected the logarithmic process function ation in marketing mix variables due to order of entry effects. In a second step.45. the quality data were trans- of the Weighted Least Squares regression is not inter- formed using Cooper and Nakanishi's (1983) zeta- pretable in terms of percentage of explained variance squared transformation. 1985). We hypothesized that the relationship between market response. As discussed ear- tors are more efficient and were used to estimate the lier. indicating that the null is rejected. The coefficient of the predicted variable using the alternate model was insignificant (t = 1. the correlation between the ac- transformations.3132. although differ- market share data is less subject to seasonality than sales ences appeared larger across markets. Consequently. An examination of the autocorrelations and partial of coefficients (Chow 1960) within nondurable markets autocorrelations (especially of the twelfth order) indi- indicate that the restriction of equal coefficients for the cated that seasonality is not present. cess function specification is true. in our empirical context. Weighted Least Squares estima- brand's marketing decision variables.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share regular intervals. we only report the estimation with the largest degrees of additional variable corresponding to the predicted dependent variable freedom. quality data were not available during the brand specific constant after controlling for order of en- periods immediately following entry and for some try and number of competitors. the results indicated that the which is estimated. The R2 cative model specification. 15. Our hypotheses predict the direction (increasing or brands) heteroscedasticity (Greene 1993) rejects equal decreasing) of the influence of order of entry on a variances. A test of equality data. 10Ok(i) I. the linear model is assumed tion. the J test was used (Davidson and MacKinnon 1981. rep- a model that includes an indicator of whether or not resent the competitive activities. Therefore. egory dummy variables are presented in Table 2.jstor. However. No. indicating that 10 We also estimated our model on a data set that excludes observa- the linear model is not rejected. The missing data information on product quality. in spite of the lack of brands with very small market shares. and the predicted dependent variable is the conclusion that the linear process function specification is statisti- added to the linear model independent variables in a new model cally superior to the logarithmic specification. p < 0. First. Greene 1993) and. the re- corresponds to lack of information about the variable sults of the more parsimonious model with market cat- for the brand at a given time period.10 Because the quality is mea- als did not reveal outliers and the fit of the model to the sured on an interval scale and because of the multipli- data is similar to previously published models. The alternative (logarithmic specifi- < 0.55. the reverse procedure tions with missing data.11 In addition. in addition to including terms to capture the vari- parameters of equations (10) and (11).48. 3. (Judge et al. The coefficient of this last term is strongly significant (t = 5. df = 8. and order of entry would be decreasing. the variable Qimt in equation (10) is expressed in terms of zeta-squared regression is 0. This model is estimated with the Therefore. a likelihood ratio test of group-wise (across ing the WLS estimates is 0. which.81. p to be true (the null hypothesis). which contribute to the fit of the model.3).

Category B 0. n. 0. consistent with our hypotheses.044 Advertising Minivan 0.000 0.079 n.582 0. n.a.a.a.a. Consistent with the findings of Brown and vehicle market. Category B 0.48.790 -3.547 0. 0.038 0.183 n.000 n. n.021 0. MARKETING SCIENCE/VO1. n.a.573 -2.061 n. n. Similarly.036 0. has on the implied elasticities. n.a.141 0. Category C 0.a.343 No.063 0. na = Extrapolations outside the range of the data used for estimation are not shown. n.a. Category Category B C n. n. This content downloaded from 196. the sensitivity of market share to mar- keting mix is not only different across brands due to p < 0. n.130 on Sun.a. but becomes minivan category and decreases 90% in the sport utility insignificant. promotion ef- ble) twelfth entrant.04. the promotion elas- for brand and related advertising.a.a.a. Category C 0.057 0. and (where applica- 2.01 respectively).a. the main effect of order remains creases 24% from the first to the twelfth entrant in the negative.a.036 n.302 Advertising (Related) SUV n.034 0. as found in previous research.017 n. the effectiveness of quality is decreasing plied elasticities (at the mean market share for the cat- with order (a = 0.a.0. Although the coefficients are of the right sign pending on the product category. Table 3 shows the im- As expected.01 and a = 0.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share Table 3 Implied Influence of Order of Entry on Elasticities Implied Elasticities at the Mean Market Share for Brands in the Category At N = 3 Competitors At N = 12 Competitors 1 st 2nd 3rd 1 st 2nd 3rd 1 2th Parameter Market Entrant Entrant Entrant Entrant Entrant Entrant Entrant Price Minivan SUV -6.a.100 n. Promotion Quality -6.451 1. p and a market with twelve competitors. third.a.a.318 0. n. n.a.a. n.583 -2. These results clearly demonstrate the Lattin (1994) and Huff and Robinson (1994). n. decreases with order (a = 0. 0. n. n. n.a. although market variable is positive and significant both for dur- they are much smaller for price and brand advertising. n. n.a. 0.a. n. In addition to modeling the sources of order of entry a significantly higher promotion level to obtain a similar advantage as influencing the effectiveness of a brand's share.062 0.378 n.010 0.377 n. second.a.a.a.a.005 0. supporting Hypothesis egory) for the first. 0.814 -3.01. n.616 0.a. n.145 n. 1.095 0. the elasticity for product quality de- marketing mix efforts.35.035 0.034 0. p < 0.191 -1. de- <. B. indicating that the later entrant must have insignificant.03. for a market with four competitors fectiveness decreases with order of entry (a = 0. n. n.796 -3.a.a.a.111 0. 3.a. 238 Furthermore. 1996 .a.414 1.023 0.a. For example.a.a. 1.064 0. n.a.a.a.092 0. n.a.a. 0. 15.019 n.01).308 n.093 0.062 0.45.a. n.484 -2.jstor. n. the time in managerial significance of these differences.a.086 (Brand) SUV n.a.a.01).120 0. * A. C correspond to the consumer packaged goods categories.01). n.a.022 0. p < 0. -6.207 -1. n. Minivan SUV 0. Category B -1. n.200 n.a. 0. n. 0.234 0.063 n.719 Category A -3.a.079 n. Category A 0. they are statistically ticity decreases between 29 and 67% from the first to the third entrant. Category C -1.a.028 Category A 0. Price sensitivity is negative.287 n. n. ables and nondurables (a = 0.022 0.a.a. p < 0.037 0.214 -1.089 n.a.a.070 n.825 -3. 29 May 2016 01:28:47 UTC All use subject to http://about.40. and consistent with Hy- It is interesting to ask what influence order of entry pothesis 1.a.784 n. Also.593 -2.092 0.835 -3. -3. n.a.199 -1.

3. Our observation period begins with entry of the pioneer in four of the five categories we examine. this category is negative. First.01. p sponse to price is lower (less negative) for later entrants. This result suggests that later entrants MARKETING SCIENCE/Vol.48.. No. petition from a larger number of brands.01 to quality. but also ing. Much of the analysis of order of entry effects has been based on either PIMS data or the Assessor database. Secondly. Thirdly. when added to the constant distribution pa- That is. order of entry and the market response number of brands (a = 0.45. and they provide guidelines for how later entrants should compete.01). In addition. our results indicate that distribution penetration is a significant predictor of market share in all markets except one of the nondurable categories. distinct from previous re- Discussion search. Given that these recursive effects have been reported in the literature and are independent. however. the price sen- Neslin. Hence. we introduce new data which allow us to analyze the effect of order of entry on durables. and order of entry and promotion effective- and a = 0. 29 May 2016 01:28:47 UTC All use subject to http://about.g. to product quality (for durables) and to adver- the relationship between order of entry and market re- tising increases when facing competition from a larger sponse to price. We hypothesized that the relationship between market response and order of entry would be decreas- Strategic Implications Our results support previous research that demonstrate the advantages of early entry. atively affected. We also find that market response to pro- with caution. they need to change price by a larger amount than ear- 12 Although the coefficient of the dummy variable for distribution of lier entrants to attain the same change in market share. Market share sensitivity to promotion (for nondur- entry effect is found supporting our hypotheses about ables). which results in higher price elasticity distribution appears as a significant predictor of market (which is the price sensitivity coefficient multiplied by share which should be included in the model in order one minus the market share). Robinson and Fornell 1985. as demonstrated above. later entry tends to decrease a competitor's price rameter (0. ables. A significant order of ing. Cooper and Nakanishi 1988) on of the changes in the competitive environment it is fac- data from five product categories. When facing com- ness. p < 0. 1986) and studies.16. differences in the effectiveness of marketing mix efforts across brands are systematically related to order of en- try.01. we can explain why there are inherent order of We have examined the influence of order of entry on entry effects and the special role of marketing mix ac- the effectiveness of a brand's marketing decision vari- tivities.765). First. a = 0.12 Our model does not exclude the possibility of order-of-entry effects due to the greater difficulty to obtain distribution for late entrants. This result should.. Consistent with the empirical results of Ghosh. Finally. later entrants have a disadvantage in competing on price. Contributions We offer improvements over existing models (e. the effect of the to avoid a misspecification bias of the other model pa- number of competitors on an elasticity depends on the rameters. 1996 239 This content downloaded from 196. distinct from previous research our data sets permit the use of an absolute market share measure-we are not limited to examining the market share relative to the first entrant. complementary to those investigated. the sum is not statistically significant from zero. sensitivity. < 0. Therefore. a recursive effect is indeed significant when distribution penetration is modeled as a function of order of entry among a set of other explanatory variables.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share order-of-entry .g. and Shoemaker (1983) we find that market re- sitivity parameter is significantly lower (a = 0. Urban et al. Controlling for differences in distribution.jstor. Indeed. Our main contribution is in modeling the sources of order of entry advantage as asymmetries in response function parameters. market share impact of the entry as well as on the im- pact of the number of competitors on the sensitivity coefficient. be interpreted ceteris paribus.02. In fact. 15. p < 0. respectively). p < 0. The hypotheses are tested using a multiplicative the sensitivities for any brand change over time because interaction model (e.01.31. we acknowledge these recursive effects without reporting the results. we have attempted to deal with survivor bias.130 on Sun. the market share of the incumbent firms is neg- quality for durables are lower for later entrants. when more competitors enter the motional expenditures for nondurables and product market.

However." in Thomas S. For the five markets we examined. References ABA Banking Journal (1991). and John L. the number of durables is limited to two categories of automobile products-sport utility vehicles and minivans sold during 1983 through 1992. their optimal price should be higher than the here. That is. Also. This suggests that Finally. Michael A. if the firm already has an entry in a similar or related Secondly. a general substantive on market share. may also depend on a number of factors not considered therefore.130 on Sun. Len Lodish. Englewood Cliffs. This they exited the market. "Dimensions of Consumer Expertise. Future research should examine other durable markets. Our analysis is not subject to sur- Much previous research has concentrated on model- vival bias since we include data on all entrants.). they need a bigger berman and Montgomery 1988). examining alternative measures of firm per- later entrants have a disadvantage in competing on formance such as economic profits could be useful (Lie- quality/positioning attributes. and Tom Rob- Limitations and Future Research Our results are subject to limitations. and Pete Fader. We conclude that order of entry effects are not insur- keting mix variables across competitors is due in part to mountable. Although the results suggest similar issue not yet addressed in the literature concerns the tendencies for advertising. tion of Glazer (1985). "Memory and Decision Making. 13 (March). are more likely to benefit from tial discounting offered by different manufacturers or a price increase than pioneers." Journal of Marketing Research. the advertising effort measured in this study concerns brand level advertising by the manufacturer."3 effects is not without considerable cost to the late en1 The authors wish to thank Gurumurthy Kalyanaram and Glen Ur- trant.. 1996 . even if ing and explaining a main effect of order of entry. later entrants must spend more on promotion than recently been questioned (Jacobson and Aaker 1985. at a comparable distributors." September. the disadvantage may not be as great earlier entrants' price. later entrants. They are also endebted to the area editor and the reviewers.jstor. This content downloaded from 196. Bill Ross. For example. asymmetric responses in a brand's hard to support for some companies and which lead marketing mix efforts are a critical source of order of them to exit the market. Lynch (1990). Alpert.45. al- though the variables in our analysis explain a significant proportion of the variance in market share. the pricing data corre- from pioneers. . In fact. Simi- lationship between market share and profitability has larly. Empirical analy- asymmetric impact of advertising: late entrants do not sis to date provides results only for pioneers who sur- have a significantly different market share sensitivity to vived and were successfully imitated. this study brings strong support that asymmetries of the effectiveness of mar- entry effects. and J. earlier entrants in order to achieve the same influence Boulding and Staelin 1990). "Uneasy at the 3. particularly if they spond to base prices and do not consider any differen- occupy a market niche. NJ: Prentice Hall. No. Wesley Hutchinson (1987). response to quality and to promotion. 15. Although the analysis offers generalizability over multiple product cate- gories. their price elasticity is lower and. in ad- sarily lead to lower shares: however. 87-90. 56 (July). Alba. The disadvantage of being a late entrant level of market share. Frank H. 411-454. Joseph W. and John G. Handbook of Consumer Behavior. Also. the failed. Also. especially since the re- change in product quality than earlier entrants. It is possible that the asymmetries identified in main effects of order of entry are lowered to insignifi- this study lead to costs for late entrants which might be cance. ban for granting the permission to use the data from their previously published study. with the excep- advertising than pioneers.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share should not instigate a price war in an effort to gain share individual dealerships. For the two automobile markets. Robertson and Harold H.48. the data do not support an characteristics of pioneers who failed. order of entry tends to decrease the market product category." Journal of Consumer Research. The measures used in this study are partial measures. Graham (1992). Indeed. "An Examination of Reseller Buyer Attitudes Toward Order of Brand Entry. surmounting these dition to the recursive effects found in previous work. 29 May 2016 01:28:47 UTC All use subject to http://about. order of entry effects do not neces- the order in which the brand entered the market. However. SCIENCE/VOl. Kassarjian (Eds. Kamins. It does not include any local advertising by 240 MARKETING ertson for their feedback on an earlier draft of this paper. 25-37. future research study suggests that the main effects of order of entry should focus on the characteristics of pioneers who are minimal.

. 134-145. Boulding. Ralph E." Marketing Science. MA: Harvard University Press. 8 (Fall). "Investigating the Relationship Between Time in Market and Pioneering Advantage. 30 (May). UT: Association for Consumer Research." Management Science.. Federal Trade Commission. 1370-1377. Insights in Decision Making. 27 (November). "Order-ofEntry Effects on Consumer Memory and Judgement: An Information Integration Perspective. "Customer-Oriented Approaches to Identifying Product- and Repeat Purchases for Frequently Purchased Consumer Markets. William H. 226-230. Phoebus J." Marketing Science." Journal of Consumer Research." The American Economic Review. Allen D. Hauser." Journal of Marketing. Jacobson. Hanssens. 1370-1377. and Gerard J. 22. George S. "Are Share and Market Power. 591-605. "Modeling Asymmetric Competi- keting Logic or Marketing Legend?. and Eitan Muller (1990). 291-309. acteristics. Hubert (1984). "An Evaluation Cost Model of Con- ing Advantages. "Tests of Equality Between Sets of Coefficients in Two Linear Regressions. Peter N." Journal of Marketing. "Market Pioneering Advantage: A Theoretical Approach. Golder. 2 (Fall). Joseph (1956). February. "Price-Induced Patterns of Competition. (1974). Curry. 473-480. Learning. Kardes. and Gurumurthy Kalyanaram (1992). 13 (June). "Pioneer Advantage: Mar- David F. "The Impact of Leadtime and Years of Competitive Rivalry on Pioneer Market Share Advantage.jstor. F. Chicago IL: Brown. 21 (November). MacKinnon (1981). and Robert A. Market Barton A. and Kenneth J. Econometric Analysis. "Environment. "Alternative Models of Categorization: Toward a Contingent Processing Framework. Theory. Gregory S. Blattberg.).130 on Sun. 235-250. Lean (1977). and Rajendra K. Peterson (1992). "Standardizing Variables in Multiplicative Choice Models. Corporate Diversification: Entry. "The Concept of Quality and the Efficiency of Markets for Consumer Products. Joel B. Faulds (1986). Gurumurthy and Glen L. "Do Higher Prices Signal Higher Quality?.." Journal of Consumer Research. 390-401. and J." Journal of Marketing." in R. Carter Hill.. New York. William and Richard Staelin (1990). 3. MARKETING SCIENCE/VOl. 15. Eitan (1985). and Steven M. Cambridge. Huff. Lee G. Frank R. P. CT: JAI Press Inc. 56 (October). Inc. Midgley (1988)." Journal of Consumer Research. 13 (March). Urban (1992).48. and Ronald J." Journal of Marketing Research." Management Science. Kalyanaram. "A Propositional Inventory for New Diffusion Research. Cooper." Journal of Marketing Research. 158-170.). 393-412. Dhrymes. Carpenter. 900-918. Cambridge." Journal of Marketing Research. Scott A. 33-52. Inc. and American Marketing Association. Conceptual Framework. search. and Thomas S. "Several Tests for Model Specification in the Presence of Alternative Hypotheses. 40." Journal of Marketing. "Consumer Preference Formation and Pioneer- and Birger Wemerfelt (1990). E." in Michael Houston (Ed. 11. Greenwich. Robert C. Robertson (1985). "Sales Promotion and Product Dif- ferentiation in Two Prescription Drug Markets. 16 (March). Lattin (1994). 36 (October). 408." in The Interface of Marketing and Strategy.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share Bain. "Brand Introduc- tion Strategies and Competitive Environments. Greene. 10 (October). 12 (March). Glazer. and (1989). 781-793. fects of the Order of Entry on Market Share. No. Aaker (1985). and Masao Nakanishi (1983). Judge. Vol. "Defensive Marketing Strategies. "The Advantages of Being First. Daniel and Jackie Snell (1988). Tellis (1993). Advances in Consumer Research. NY: Macmillan Publishing Company. Rajan Varadarajan. and D. sideration Sets. Gregory (1960). The Theory and Practice of Econometrics. and Robin Wensley (Eds. Chaim. "Market Pioneering. 36 (8) August. 49." Marketing Science. Kahneman. Shocker. NY: John Wiley & Sons. Bond. . and Re- and Pradeep Bansal (1990). Chow. Dornoff (1993). George Day. Robinson (1994). W. Econometrics: Statistical Foundations and Applications. Roger A. Shoemaker (1983). Murali Chandrashekaran.849867. (1993). 1996 241 This content downloaded from 196. . 15.). New York. and Kent Nakamoto (1987). 209-215. and James M. (1979). Kerin. and Pradeep Bansal (1990). and (1988). Wisniewski (1989). R." Journal of Consumer Research. 8-19. Robert and David A." Journal of Marketing Research. R. New York. There Associations Between Price Elasticity and Brand Char- 1160-1177. Shugan (1983). (1985). "Is Market Share Really All That It's Cracked Up to Be?. "Predicting Utility. 2nd ed." Management Science. 43 (Fall)." Journal of Consumer Research." Econometrica. Srivastava (1979)." Econometrica. Cooper. 11- 10. 29 (August). Hogarth (Ed. A." Journal of Marketing Research. 393- 285-298." Journal of Marketing Re- tion. IL: University of Chicago. 96-108. Neslin. and Tsoung-Chao Lee (1985). "First- Mover Advantage: A Synthesis. 11 (Summer). MA: Harvard University Press. Barriers to Competition. Griffiths." Economic Report." in 1983 AMA Educator's Proceedings. Vijay Mahajan. Trial Penetration. Chicago. 387-398. 7 (4) Fall. and Preference. Hjorth-Anderson." Management Science.. Journal of Consumer Research. NY: Springer-Verlag. 343-357. David J. 26 (3) August. (1984). Gatignon. 29 May 2016 01:28:47 UTC All use subject to http://about. George G. Barton Weitz." Marketing Science. "Indexing Product Quality: Issues. New York. John R. Biggadike. and David J. 275-279.45. S. 62-75. 446-462. 22 (May). Lenard C. Goods. Gerstner. Boston: Kluwer Academic Publishers. 40. 28. Ghosh. Weitz. 319-360. and William T. and Robert W. Cohen. Lee G. egies. 455-472. "Competition as a Moderator of the Effects of Advertising on Sales. Market-Share Analysis. 20 (June). Avijit. Dominique M. Christina L. Chr. Strat- egy and Performance. and Kunal Basu (1987). 75 (3). "Market Entry and Defensive Strat- search Propositions. June. Provo. Helmut Lutkepohl. 10 (October). Fershtman. Davidson. "Dynamic Ef- Day. R. and Results.

"Consumer Knowledge: Effects on Evaluation Market Entry: An Exploratory Study of New Industrial Prod- Strategies Mediating Consumer Judgements. (1972b). 231-251. G. "The Product Life Cycle and Time-Varying Advertising Elasticities." Journal of Marketing. and Edward E. Douglas L. (1975). "Guest Editor's Introduction to the Special Issue on Research in the Content of Strategy." Journal of Marketing Research.jstor. Elasticity. (1970b). 35. 113-138. 11 (August). O'Sullivan. William T. Theresa Carter. 25 (February). 12 (June). An Empirical Study. 27 (June). B8. and Claes Fornell (1985). 305-317. Hauser (1980). and Scot Burton (1989).). "Effect of Schema- Incongruent Information on Memory for Stereotypical Attri- Some Insights From Schema Research. "Parameter Estimation for a Multiplicative Interaction Model-Least Squares Approach. 645-659. 97-104. Lilien. and Ronald C. B6. (1988). 15.. Steven Gaskin. Urban. 429-444. No. Perceived Quality. 9 (Special Issue).45. Lambin." Marketing Science." The Journal of Business. Lakshman and S. Mita (1985). Dubbed the 'Family Sedans' of the '90s. Smith (1984). Schmalensee. MA: Lexington Books. Tioneerng and Market Share: Is Entry Time Endogenous and Does It Matter?" Journal of Marketing Research. Philip M.). Perceived Quality. 303-311. Raj (1985). 349-365. New York: The Free Press. Michael E. Goodstein (1991). "Advertising and Competitive Behavior: A Case Study." Journal of Consumer Research." Journal of Marketing Research. "A Multinomial Extension of the Linear Logit Model. 31-46. 454-467. Lexington. 568-585. Michael J." in J. 32 (June). Chris S. 11. "Product Differentiation Advantages of Pi- oneering Brands. 439-452. Nakanishi. and (1982). butes. "The Timing of Competitive Sujan. L. Durso (1984)." Strategic Management Journal. Rogers. 29 May 2016 01:28:47 UTC All use subject to http://about. Montgomery (1988)." Journal of Marketing Research. 17-39. Montgomery. 22 (August)." Annual Review of Psychology." European Economic Review. 251-259. 259-267. 242 MARKETING SCIENCE/Vol." Journal ical Analysis of National and Local Advertising Effect on Price of Marketing Research. and Francis T. Gary L. Weitz. 16 (November). Liebernan. Ira T. Olshavsky. 45 (October). "The Effects of Brand Positioning Strategies on Consumers' Brand and Category Perceptions: Medin. 22 (May). Quality." Journal of Con- ucts." Management Science. "Rivalry Accelerates for Sport Trucks. New York. sumer Research. Albert R. and 3. 1 (Summer). and Zofia Mucha (1986). Monroe." Strategic Management Journal. Popkowski Laszczyc. Valarie A. Everett M. Moore. Maddala. "Perceived Quality in Consumer Decision Making: An Integrated Theoretical Perspective. Englewood Cliffs. Krishnan (1985). Donald R. (1980). 1996 ." Journal of Marketing tive Effects in Diffusion Models: An Empirical Analysis. (1985). "The Relationship Be- tween Perceived and Objective Price-Quality. 119-126. Naert and Alain Bultez (1975). P. Glen L. national Journal of Research in Marketing. "AT&T Announcement.48. Leonard J. 372-378. and Hubert Gatignon (1994). 41-58. "The Effect of Price on Sub- jective Product Evaluations. 28 (February). 17 (December). New York. "Brand Performance in the Cigarette Industry and the Advantage of Early Entry. "Sources of Market Pioneer Advantages in Consumer Goods Industries. 14 (December). and James R. 468-484. 585-619. Bureau of Economics. 4. "Specifying Competi- vertising on Consumer Price Sensitivity. B1." Journal of Marketing Research. "Market Share Rewards to Pioneering Brands: An Empirical Analysis and Strategic Implications. This paper was received April 7." Journal of Consumer Research. Prakash (1990). Olson (Eds. American Economic Revaiew. 6. "Sources of Market Pioneer Advantages: une strategie de modelisation." Federal Trade Commission. Introduction to Econometrics." Journal of Business. 21 (August). 9 (May). 105-204. "Consumer Perceptions of Price. and John R. "Is Gasoline Advertising Justified. Jean-Jacques (1970a). Cynthia (1988). Porter. Richard (1982). 263-276. and Eunsang Yoon (1990). William Boulding. 4. Kent B. Theil. MA: Lexington Books." June 6. Peter T. 9 (Special Issue). (1985). Richard W. and Ram C." in J." Journal of Marketing Research. "First-Mover Advantages. Barton A." July 18. "Recall and Consumer Consideration Sets: Influencing Choice without Altering Brand Evaluations. Design and Marketing of New Products. "Introduction to Special Issue on Competition in Marketing. 10 (October). Philippe A. "A Computer On-Line Marketing Mix Model. (1979). 1994. and has been with the authors 241 days for 2 revisions. "An Empir- (1972a). "Multifirm Analysis of Competitive Decision Variables. Henri (1969). Research. 2-22. 1(2)." Inter- Research. Competitive Strategy. 55- and Value: A Means-End Model and Synthesis of Evidence. and David B." Recherche et Applications en Mar- The Case of Industrial Goods Industries. and Christine Dekleva (1987). "The Effects of Ad- Parker." Marketing Letters. "Product Categorization and Inference Making: Some Implications for Comparative Advertising. Whitten. 2. 314-322.130 on Sun. Bettman (1989). Wildt." Journal of Marketing Research." Journal of Personality and Social Psychology. Lexington. Journal of Marketing. Jacoby and J. "Part de marche et pression marketing: vers Robinson." International Economics Review. Jacoby and J. 87-94. and Eric Dor (1989). 26 (November). (1991). 26 (November). Masao and Lee G. Marvin B. Diffusion of Innovations. Nedungadi. 1913-1974. (1983). (1988). 149-160. 4. 36 (5). (1974). Cooper (1974). 3-24. 22 (August). "Optimal Allocation of Competitive Marketing Efforts: Parsons. Zeithaml. "Optimal Marketing Behavior in Oligopoly. Lichtenstein. 229-238. NY: Macmillan Publishing Company. NJ: Prentice-Hall. "Concepts and Con- cept Formation. Rao (1990)." Management Science.BOWMAN AND GATIGNON Effect of Marketing Mix on Market Share Krishnamurthi.. 3-8." Applied Economics. (1988). This content downloaded from 196." 70. S. 17 (October). 119-129. Olson (Eds. 52 (July). Wall Street Journal (1990)." Journal of Marketing keting. "Simplified Estimation Procedures for MCI Models. 47 (July). NY: The Free Press.