Analyzing the effects of a brand introduction on competitive structure using a market share attraction model

https://doi.org/10.1016/j.ijresmar.2003.09.001Get rights and content

Abstract

The introduction of a new brand may change the competitive structure, which concerns the market shares and the marketing instrument elasticities. For example, the new brand can cause intensified price competition and possibly also influence the relative effectiveness of price promotions for the incumbent brands.

In this paper, we develop a methodology to statistically test for the various possible changes in the competitive structure, where we focus on weekly scanner data. We develop a market share attraction model with which we can handle a changing number of brands.

The usefulness of our approach is illustrated in a simulation experiment where its implementation is compared with various alternative methods. We further develop tests to infer whether there are competitive reactions to an entry. Our illustration to scanner data for the detergent category emphasizes the usefulness of our approach.

Introduction

The introduction of a new brand in an existing market can have a large impact on the competitive structure, which concerns the market shares, the marketing instrument elasticities (as established by consumer behavior) and the use of marketing mix instruments by brand managers. For example, such an introduction may trigger intensified price competition, thereby possibly also affecting the relative effectiveness of price promotions for the incumbent brands. There may also be consumer reactions which can lead to a change in the relative effectiveness of pricing strategies. Note that similar changes could occur if a brand is removed from the market. Naturally, the effects of a changing number of brands is not confined to price competition, as the same arguments would hold for any other marketing instrument like, e.g., display or distribution. Furthermore, after correcting for the effects of the marketing mix variables, the new brand may turn out to win share at the cost of only a few competitors, instead of drawing share proportionally from all incumbent brands.

There are several studies of the effects of the entry of a new brand on the competitive structure. These studies can be broadly divided in two types of approaches. One approach takes a noncooperative game-theoretic view, while the other is predominantly based on empirical research using time series or panel data models.

The first type of studies take a normative viewpoint; that is, there is a focus on how one should respond to an entry in an optimal way. An example is Basuroy and Nguyen (1998), who derive the theoretical conditions under which a market share attraction model is appropriate for equilibrium analysis. Within the context of this model, these authors demonstrate that the entry of a new brand would establish price decreases, which would hold true for fixed and expanding markets. For fixed markets, they further show that incumbent brands would be inclined to lower marketing expenditures, while in expanding markets these expenditures would be set at lower levels. Other examples of similar approaches can be found in Cubbin and Domberger (1988), Gruca, Kumar, and Sudharshan (1992), Gruca, Sudharshan, and Kumar (2001) and Karnani (1985). A key feature of these studies is that there is usually no focus on empirical data. A notable exception is Shankar (1997) who studies the marketing mix reactions by pioneers to entry. In the studied market, the entry changes the competition from a monopoly to a duopoly, with as a consequence that in the duopoly market, different competitive games can be played. Shankar (1997) derives optimal responses to entry for each case and then uses empirical data to find the actual game played in a pharmaceutical market. Shankar (1997) concludes that the results found in, e.g., Gruca et al. (1992), only hold under certain competitive games, while it also has to be assumed that marketing mix elasticities are constant.

Good examples of the second type of research, which is more data based, are Bowman and Gatignon (1996) and Chintagunta (1999). This research is explicitly based on observed market situations. Bowman and Gatignon (1996) study the effect of the order of entry on market shares and the effectiveness of marketing instruments. They show that the order of entry negatively influences the effectiveness of promotion and that it lowers price sensitivity. The main effects of the order of entry on own market share are found to be small, while in contrast, there are strong effects of the order of entry on the effectiveness of marketing efforts. Bowman and Gatignon (1996) do not consider the effects of an entrant on the incumbent brands, while it would not be unlikely that a brand introduction also affects the effectiveness of marketing mix variables of these other brands. Note that Bowman and Gatignon (1996) do assume dependence between marketing mix elasticities and the number of competitors, but they abstain from testing for changes in these elasticities, nor do they consider cross effects.

An example of a study that does consider changes in the marketing mix effectiveness of incumbent brands is Chintagunta (1999), where the effects of entry are studied in the context of an individual choice model. A random effect multinomial logit model is used where brand intercepts are modeled by brand locations in attribute space with household-specific importance weights. A new brand introduces an additional brand position in the attribute space. As a consequence of this entry, several changes to the competitive structure may occur. First of all, locations of extant brands or importance weights may change, or both. It is suggested that a brand introduction has a substantial impact on the importance weights assigned to attributes. Only minor changes occur in brand positions and in the sensitivity to marketing activities. Chintagunta (1999) documents that, due to a new brand introduction, price sensitivity tends to increase while promotional sensitivity tends to decrease.

As can be understood from the discussion above, Bowman and Gatignon (1996) and Chintagunta (1999) only describe the demand side of the market. That is, these studies focus on the effects on elasticities, where these elasticities may change due to brand repositioning or changes in preference. The behavior on the supply side, by retailers and manufacturers, is not studied. An example of an empirical study, which does consider the effects of new entry on this side of the market is Robinson (1988). He presents an analysis of the reactions to an entrant in 115 different cases, where he shows that the most common reaction pattern to entry is no reaction or only a reaction with a single marketing instrument. However, as Basuroy and Nguyen (1998) suggest in a response to these findings, there is a need for further empirical analysis to support the theoretical results.

This brings us to the contribution of our paper to the literature on the effects of market entry. We put forward several empirical methods to examine both sides of the market. We suggest statistical methods to validate the various predictions from normative studies. Our techniques are designed for weekly scanner time series data on market shares. We focus on market shares as we are interested in the relative performance of brands when a new brand is introduced, even when this introduction would lead to an increase in category sales. Naturally, our methods can be redesigned if one intends to focus on sales only. Such an approach could turn out to be useful in a category in which expansion effects play an important role. However, in a sales model, one needs to model seasonal effects and the category expansion or category contraction explicitly.

In order to analyze changes in actual behavior of a brand manager, i.e., changes in the use of instruments like price and display, one needs methods to test for structural changes in time series variables. Upon doing so, we build on the findings in Srinivasan, Popkowski Leszczyc, and Bass (2000), who document that, except for possible level shifts, marketing time series data seem to be stationary. Hence, we also assume stationarity of all time series under scrutiny. Next, to examine possible changes in the effects of marketing mix efforts, within the context of a market share attraction model, we propose new methods. First of all, we introduce a new estimation method that can handle a changing number of brands in the observation period. Next, we propose a method to test whether parameter values differ across the subsamples. To demonstrate the value-added of our approach, we perform a simulation study where we compare our method with various alternatives.

To summarize, our paper contributes to the literature by the development of measurement tools to examine the effects of the entry and exit of brands, given the availability of a sample of the relevant time series data. As the exit of a brand mirrors an entry, we only focus on the latter for brevity. Naturally, our methods can readily be adapted to the exit case.

It should be mentioned that we condition our analysis on the observed entrant's strategy. Hence, we study the effects of this strategy on a specific existing competitive structure, and we therefore cannot consider the effects of different strategies. There are papers where the entrant's strategy is explicitly considered; see, e.g., Gatignon, Weitz, and Bansal (1990), Shankar (1999) and Shankar, Carpenter, and Krishnamurthi (1999). However, when one were to apply our methods to a range of data sets, one might make generalizing statements about the observed behavior and its consequences.

We apply our methods to a set of weekly time series observations concerning the detergent category. For this market, we observe the introduction of a new brand at about one-third of the sample. Upon application of our methods, we do not find supportive evidence for the hypothesis that prices are lowered after the introduction of a new brand. Next, we find that not many marketing efforts are increased, after the introduction of a new brand, where we should bear in mind of course that we consider only one market. We further find that part of the competitive structure changes after an introduction, thereby providing an incentive to further examine consumer response to brand introduction.

The outline of this paper is as follows. In Section 2, we discuss the testing approach verbally, i.e., without explicit formulas. In Section 3, we briefly discuss the attraction model, and we discuss parameter estimation in the attraction model, while taking into account the introduction of the new brand. Technical details are relegated to Appendix A. In Section 4, we present a testing procedure to assess whether the introduction of the new brand results in a different competitive structure among the incumbent brands, where we put the technicalities in Appendix B. In Section 5, we present the results of a simulation experiment in which we compare the approach suggested in Section 4 with various alternatives. In Section 6, we discuss the testing procedure for breaks in the level of marketing instruments. The testing and estimation procedures are illustrated in Section 7 for the detergent category. We conclude our paper in Section 8 with some remarks.

Section snippets

Testing approach

In this section, we outline the ideas behind our empirical approach, without laying out any technical issues. Our analysis of the effect of a brand introduction on the competitive structure is guided by the notion that part of the changes in market shares might directly be attributed to the marketing efforts of the new brand. This is in contrast to standard structural break analysis, where there is no endogenous effect at the time of the break. Hence, the question at hand is whether all the

Attraction model

The competitive structure can be studied using either sales or shares. We choose to analyze market shares and to measure the effectiveness of marketing instruments of different brands using the familiar market share attraction model Bronnenberg et al., 2000, Cooper & Nakanishi, 1988, Leeflang & Reuyl, 1984, Naert & Weverbergh, 1981. An advantage of using market shares is that these are in general not influenced by category expansion or contraction. Unlike sales, market shares are therefore

Testing for shifts

In this section, we discuss how various kinds of shifts in the competitive structure can be translated into the context of the attraction model. We discuss how one can test for these shifts. This section focuses on changes in aggregate consumer behavior. In Section 6, we will discuss testing for changes in the use of marketing instruments.

The previous discussion of the attraction model considered a stable competitive structure among incumbent brands. More technically stated, the parameters Σ*, μ

Simulation study

As an alternative to the method we suggest, one could choose to ignore some of the technicalities and test for changes in the competitive structure using other techniques. One of the possibilities is to ignore the newly introduced brand and estimate a competitive model for the incumbent brands only. Dependent on the chosen functional form of the attraction model, it may matter if the researcher does or does not take into account the marketing mix of the new brand. For example, if in the

Testing for breaks in marketing efforts

In the previous section we have discussed the testing of changes in the competitive structure. Such changes can mainly be attributed to the consumers. As discussed earlier, brand managers oftentimes also react to an entry. Next to testing whether the competitive structure has changed as a consequence of a brand introduction, one may also be interested in testing whether the incumbent brands adapt their marketing strategy as a response to the new competitor. For example, one may want to test

Illustration

The use of an attraction model in case the sample contains a brand introduction is illustrated in this section using a data set on detergent. We also illustrate the use of some statistical tests for changes in the use of marketing instruments or for changes in competitive structure. The data set concerns 12 brands of liquid detergent, covering 134 weeks. One of these 12 brands (“Surf”) is introduced in week 38.

As explanatory variables for the market shares of the brands in this market, we have

Conclusion and discussion

In this paper, we proposed methods to empirically analyze the effects of a brand introduction on the competitive structure, where we focus on weekly observed market shares and marketing instruments. We suggested a number of statistical tests that can be used to judge whether or not the brand introduction affects the competitive structure among incumbent brands. Tests for the constancy of the marketing strategies themselves were also presented. If incumbent brands respond to the introduction by

Acknowledgements

We thank the Editor, two anonymous referees and participants of the ART Forum 2001 (Amelia Island, FL, June 24–27, 2001) for many constructive comments.

References (23)

  • J. Cubbin et al.

    Advertising and post-entry oligopoly behavior

    Journal of Industrial Economics

    (1988)
  • Cited by (0)

    View full text