PRICE SENSITIVITY AND BRAND SWITCHING

(1)

PRICE SENSITIVITY AND BRAND

SWITCHING

A COMPARISON OF BRAND TIERS ACROSS PRODUCT

PRICE SENSITIVITY AND BRAND

SWITCHING

A COMPARISON OF BRAND TIERS ACROSS PRODUCT

MANAGEMENT SUMMARY

Marketing accountability has become increasingly important for marketers in the recent years. Firms that are able to justify their marketing expenditures and have a better market understanding perform better (Verhoef & Leeflang, 2009). One step to achieve this accountability is by getting insights in specific sales promotion effects. This report focuses on the secondary sales promotion effects. These secondary promotion effects increase the demand for the promoted brand by taking away demand from non-promoted brands. Since these consumers will not become loyal to the promoted brand there is no gain in this promotion effect. Research found that over the years consumers have become more price sensitive (Bijmolt et al., 2005). This raises the question: did the increasing price sensitivity reach a point where even loyal customers are considering switching to a competing brand when promoted? These considerations lead to the following research question:

Is the magnitude of the brand switching effect on the sales bump affected by the level of price sensitivity of the category? And does this differ for perceived brand tiers?

Sales promotions are used to trigger consumers to purchase the promoted product at that moment. After distribution, price promotion is the most effective element of the promotion mix in short term (Ataman et al., 2010). Other price promotion effects are: stimulation of deal-to-deal purchasing, and a decrease of brand equity over time (Nijs et al., 2001; Blattberg et al., 1995). Hence, price promotion increases (price) promotion effectiveness and leads to brand switching.

Non-price promotions are effective because it generates extra attention to the promoted brand (Dhar et al. 2001). When a promotion is presented separate from its alternatives, prices are less easily compared. Lemon & Nowlis (2002) found that high tier brands benefit more from these types of separation promotion than low tier brands do. Separation promotion instruments can be more effective in specific categories (Dhar et al. 2001). Advertising has negative effects on the price sensitivity because it stimulates consumers to become loyal (Mela et a., 1997).

(4)

4

Promotions that are effective in one category are not necessarily effective in another. Research described price sensitivity as an actor in why consumers respond to price promotions (Krishnamurti & Raj, 1991). Gordon, Goldfarb & Li (2013) found consumers become more price sensitive for categories which have high share of wallet, low perishability, high substitution and/or a high market concentration.

A multiplicative sales model is developed and tested for three price sensitive (paper towels, frozen pizza, tortilla chips), and three price insensitive categories (peanut butter, mayonnaise, carbonated soft drinks).

The results show that cross brand affects the own brand sales levels. The cross brand price elasticities supported by feature and display promotions are even greater than the price only parameters for cross brand effects in most models. Hence, hypothesis 1 is confirmed. The magnitude of the switching effect is greater when separation promotions are applied. However the results are not as hypothesized. The price sensitive categories benefit greater from separating promotions than price insensitive categories. Furthermore the results show that the magnitude of cross brand elasticities are greater in the sales model for low tier brands, compared to those of high tier brands. Hence, these results confirm the expectations described in hypothesis 3. The magnitude of the cross brand elasticities strongly deviate across price sensitive categories. Where paper towels are strongly affected by cross brand elasticities, tortilla chips showed the opposite effect. Therefore a comparison of cross brand effects across the level of price sensitivity resulted in inconclusive results. This implies that hypothesis 4 is rejected. The high tier brands do not suffer strongly from cross brand effects. However, when the high tier brand models are affected by high tier cross brand effects, the high tier rival brand has greater impact on high tier brands in price insensitive categories, than on high tier brands in price sensitive categories. Hence, hypothesis 5 is rejected.

(5)

5

PREFACE

Wednesday September 2nd_{2015 was the day it all started. The date that the first meeting found place,}

which marked the start of the thesis writing process. That eventually led to this report. Somewhere at that time the topic choice was made. I chose for brand switching because I am interested in how the concepts of brand loyalty and promotion effectiveness interrelate. Concepts of which I learned a lot during several courses in the Master Marketing, and of which I hope to learn even more when applying it in practice, after my graduation.

Although things did not always work out as expected, I enjoyed the process of writing this master thesis. I learned a lot during the process, for which I am very thankful. One example is working with large data sets. The effort it takes to understand how large data sets work is something I clearly underestimated.

Finally, I would like to use this opportunity to give special thanks to prof. dr. Wieringa for his advice and support. He helped me finding the right course of action, and spend a lot of time correcting my writings. Not to forget the trouble associated with the arrangement of the data.

(6)

6

1. INTRODUCTION

Marketing accountability has become more and more important for marketers. In order to get strategic influence in the organization, it has become a necessity for marketing to prove the effectiveness of its investments (Verhoef & Leeflang, 2009). In many firms marketing has lost its strategic decision influence (Homburg, Workman & Krohmer, 1999). Key marketing mix decisions as pricing and distribution are now covered by sales and finance. Verhoef & Leeflang (2009) found that marketing accountability and innovation are key antecedents for the marketing department’s influence in the firm. A better cooperation of the marketing department with departments as finance, sales, and research & development has a positive effect on market learning and the market orientation of the firm. This can increase firms’ financial and customer relationship performance (Luo, Slotegraaf & Pan, 2006). Hence, being able to justify its (marketing) expenditures, and get better understanding of the market, is good for the firm performance. Managers must know how promotions work for their specific brands, and which type of promotions will be most effective in reaching the marketing objectives. One way to achieve this accountability is by getting insights in (the magnitude of) sales promotion effects. Research distinguishes two different kinds of sales promotion effects: primary and secondary demand effects. Primary demand effects represent the effects that increase the category demand, and can lead to cannibalization of future demand as a result of purchase timing acceleration, stockpiling and/or increased consumption. An example of a secondary demand effect is brand switching. Unlike the primary promotion effects, brand switching does not affect the category demand during the promotion. Brand switching only increases the demand for the promoted brand by attracting customers who are loyal to other brands. Brand switching does not increase the category demand, but takes away demand from non-promoted brands. So, as long as retailers do not receive higher margins for the promoted items, it is not a beneficial promotion effect for retailers (van Heerde, Gupta & Wittink, 2003).

(8)

8

Even though brand switching is not a beneficial side effect of sales promotions, it does not make this field less interesting to study. Differences in perceived quality tiers for example can be one reason to switch brands during sales promotions. Most switchers are considered shoppers of low-quality, low price-tier brands (Sethraman, 1996; Blattberg & Wisniewski, 1989). Nevertheless, this does not preclude consumers who normally buy high quality, A-labels to switch to other brands during promotions. Especially since, over years consumers have become more price sensitive (Bijmolt, van Heerde & Pieters, 2005). This raises the question: how does this price sensitivity relate to perceived brand quality? And did this effect reach a point that even loyal customers are considering switching to a competing brand? Hence, what is the proportion of A-label loyal customers that switch to another A-label during promotions? This article evaluates the switching effect on different product categories. To compare the proportion of the effect over price sensitive versus price insensitive categories. There have been multiple studies that investigate the (magnitude of the) promotion effect of brand switching, whereas the focus was merely on how large the effect was for different categories (Gupta, 1988; van Heerde et al., 2003). Blattberg and Wisniewski (1989), studied how brand prices influence the market shares during promotions, and which types of (price tiers) brands are affected by these promotions. Less research is done to the origin of the brand of which consumers switch from, and whether price sensitivity of the category has impact on this switching behavior. This study is aimed to provide insights in the promotion effect of brand switching. Specifically in whether the brand switchers are high brand tier shoppers or merely low brand tier shoppers. The results will be compared over different categories. This is useful because it can help retailer and brand manufacturers in their selection process of promotion instruments for product categories. These insights will help selecting the best instrument for the most beneficial category, which accounts for the least levels of waste of investment by promotion effects caused by brand switching. My objective is to study this gap in research, and come up with a model to calculate the brand choice elasticity given a certain level of perceived brand quality tier.

This all is summarized into the following research question:

Is the magnitude of the brand switching effect on the sales bump affected by the level of price sensitivity of the category? And does this differ for perceived brand tiers?

(9)

9

(10)

10

2. LITERATURE REVIEW

2.1 Promotion Effects

Sales promotions are part of the marketing mix and are used to trigger consumers to purchase the promoted product or service at that moment (Kotler & Armstrong, 2014). This generates a bump of extra sales compared to the sales level in a non-promotion period in time. The sales bump can be distinguished in extra sales caused by purchase of increased quantities, purchase timing acceleration, and brand switching (Gupta, 1988). These first two effects mentioned are briefly explained in the covariate section of this report. In this article I primarily focus on how promotion and synergies of the most often used elements of the promotion mix have impact on brand switching, and how this is affected by the type of brand consumers switch from and the type of category. Research found that promotion mix (interactions) could have different effects for different brand tiers (Lemon & Nowlis, 2002). For that reason, I will discuss the theoretical implications of promotion effects on brand switching for price promotion and non-price promotion separately.

Price promotion

After distribution, price promotions are the most effective element of the promotion mix in short term (Ataman, van Heerde & Mela, 2010). Discounting only has very small long term effects (1%), and does not do much of brand building (Ataman, et al., 2010), price cuts thus only trigger sales bumps in the short term. As mentioned briefly in the introduction, consumers have become more price sensitive over time (Bijmolt et al., 2005). Long term exposure to price promotions has increased consumer sensitivity to prices and promotions. Price promotions let consumers focus on price cues, which cause prices to become more salient (Mela, Gupta & Lehman, 1997).

(11)

11

Next to the consequences to price sensitivity, price promotions lead to a decrease in brand equity over time (Blattberg, Briesch & Fox, 1995). The use of price promotions cause consumers to reduce their reference price, which is the price they consider as appropriate to pay for a specific product. A lower reference price reduces the premium the brands can charge for the brand, which engenders a decreased brand equity (Blattberg et al., 1995). This has negative effects for brand and sales value in the long run (Ailawadi, 2001). Consequently, consumer price promotions simply encourage deal-to-deal purchasing behavior and brand switching. A decreasing brand equity diminishes the gap of the specific brand to other brands. Brands become more similar to each other. The price cut on a different brand can be the final step that bridges the perceived difference to their usual brand. In this line of reasoning, price promotions makes other brands attractive too. Hence, during price promotions the likelihood to switch to a better offer, even if it is not consumers’ favorite brand, will increase.

Non-price promotion

Non-price promotions, like displays and feature promotions, are effective instruments because they generate extra attention to the promoted brand (Dhar, Hoch & Kumar, 2001). Research on consumer decision making and choice has shown that choices are affected by the number and characteristics of the alternatives under consideration, the characteristics of the decision environment, the context effects, and choice task (Bettman, Luce & Payne, 1998). This could explain why consumers react differently to offers when they are separated from each other compared to when they are offered side-by-side and direct comparison is possible. Separating promotion instruments are promotions instruments that separate one brand from the other brands from the same category in a way that the brand can be judged individually without the influences of direct comparison of competing alternatives (Lemon & Nowlis, 2002). These instruments are amongst others (end-of the aisle) display promotions and feature advertising. In case of end-of-the-aisle display promotions, consumers cannot easily compare the promoted brand to other alternatives, because there are no other alternatives placed in the direct (shelf) space around the promoted product.

(12)

12

Research found that separating promotion instruments are especially beneficial for high tier brands, compared to low tier brands (Lemon & Nowlis, 2002). In section 2.2 these brand tiers are explained in more detail. Strong brands carry a strong set of brand associations which can be evaluated separately, whereas prices are hard to compare without any reference of other alternatives (Lemon & Nowlis, 2002). How does this work? Low tier brands generally benefit from low prices. So, when a low tier brand is displayed at the end of the aisle, the consumers are not able to easily compare the offer to the price of their regular brand choice. Since the low tier brand lacks strong associations, the consumer has few associations to evaluate the choice. This reduces the probability to switch to the low tier brand. The brand associations cause that separating promotions are more likely to induce brand switching for high tier brands. That does not mean that separating promotion instruments are ineffective for low tier brands. Separating promotions instruments in this case will generate primary demand effects, such as increased quantities by consumer who are loyal or more familiar to the brand (van Heerde et al., 2003). Thus, high tier brands will benefit from separating promotion instruments like display and features, because of their higher brand equity.

Display promotion is an effective promotion instrument, nevertheless it could be more effective for specific categories (Dhar et al., 2001). It can be more effective for brand switching behavior in price

insensitive categories. End-of-the-aisle display promotion leads to dramatically increased shelf-space,

(13)

13

This evaluation increases the importance of brand equity in the process of choosing brands. Hence, it could stimulate the importance of brand equity for low tier shoppers in the decision process as well. In this line of argumentation, low brand tier customers are more likely to switch to higher tier brands in feature advertising without extra support.

All of the above leads to the following hypotheses:

H1: A non-price promotion enhances a brand’s value, hence it increases the switching effect for consumers of low brand tiers to switch to higher brand tiers.

H2: In price insensitive categories the magnitude of the switching effect is greater when separating promotion instruments are applied than in price sensitive categories. Interaction effects

Next to the individual main effects of the promotions instruments on brand switching there could be interaction effects of these instruments as well. Prior research found a strong interaction effect of feature and display promotion on item sales (Blattberg et al., 1995). Additionally, Lemon & Nowlis (2002) found an interaction effect of price promotion with feature or display support, which will benefit low tier brands more than high tier brands. However, there is no strong evidence to explain what promotion effect has caused this increased effect. It could be caused by brand switching, but stock piling is a likely possibility too. For this reason I did not include a hypothesis for interaction effects, but it still should be included in the measurement to find insights in these effects.

Dynamic effects

(14)

14

Similarly, lagged effects represent the sales dip after the promotion. Customer could have stockpiled the product during the promotion, and fulfilled their needs for the product. In this way, the promotion effects are dynamic and influence the demand in the weeks after the promotion (Leeflang et al., 2015). This dip can be spread out over multiple weeks (van Heerde et al., 2000). Although dynamic effects sometimes tend to be difficult to detect, because of cross brand effects, research has proved the presence of these effects (van Heerde et al, 2000). Moreover, research found that (sales) models that allow parameters to vary for the magnitude and timing of past promotions will lead to more valid results (Foekens et al, 1999).

When evaluating the effects for different promotion instruments it becomes clear that brand equity has a dominant position in whether consumers will switch brands or not. Since this study aims to find out what proportion of consumers will switch from different quality tiers, it is necessary to take a look into the influences of brand equity.

2.2 Brand Equity and Brand Tiers

As explained, brand equity is an important factor when it comes to brand switching. This section will answer the question: what are brand tiers and hoe does this influence the magnitude of brand switching?

Brand tiers in general

(15)

15

High tier brands include brands that score high on brand equity and have relatively high prices, compared to the other brands in the category. These brands are mostly national brands, which are heavily promoted by use of advertisement in order to link positive associations to the brand and create value (Kumar & Steenkamp, 2007). Shoppers of these brands often perceive a higher quality difference between high tier brand and low tier brands (Sethuraman, 2001). Low tier brands are characterized by lower prices and less brand value compared to the high tier brands in the same category. These brands have less advertising support, and have less (specific) brand associations (Kumar & Steenkamp, 2007). Shoppers of these low tier brands are often familiar with the quality of the brand, which they perceive as acceptable and satisfactory to fulfill their needs. This makes them less willing to pay a premium for high tier brands (Sethuraman, 2001; Steenkamp, van Heerde & Geyskens, 2010).

Choice studies show that consumers choose the alternative with the highest perceived utility. During promotions the promoted brand gets more attention, so the perceived utility of the promoted brand increases. When the perceived utility of brand A outperforms brand B, the consumers will probably select brand A. The promotion will increase the perceived utility of the promoted brand compared to a customers’ regular brand, which could lead to switching behavior. However, a consumer who is loyal to the high tier brand will not necessarily switch to a lower tier brand during promotions. The brand equity of a high tier brand is higher than the low tier brand. Consumers of high tier brands are brand sensitive, and weigh the brand equity in their perceived utility (Blattberg & Wisniewski, 1989; Lemon & Nowlis, 2002). A low tier brand lacks this brand value, which leads to a lower perceived utility for these consumers compared to the high tier brand. So by its nature, brand switching is affecting low tier brand demand. Consumers of these low tier brands often switch to the higher tiered promoted brands (Blattberg & Wisniewski, 1989). Although there are differences in the height of brand equity, high tier brands have higher brand equity than low tier brands have. Therefore it is expected that the low brand tier shoppers have a greater win when they switch, than high brand tier shoppers have. Thus,

(16)

16 Private label and brand tiers

As mentioned briefly, over the years retailers have invested heavily in (lower price tier) private labels. They have become an important instrument for (grocery) retailers (Nielsen, 2014). Retailers invest heavily in the quality and the perceptions consumers hold against private labels, to decrease the perceived gap with national brands (Kumar & Steenkamp, 2007).

As a result of this investments in the recent years many retailers introduced multiple-tiered private label programs (Geyskens, Gielens & Gijsbrechts, 2010). These programs include additional private label lines to the regular private label line, which has been present for a long time. These additional lines often have an economic and a premium pricing strategy. This multiple tier private label program provides the retailer an assortment for value shoppers, mid-tier level for the mainstream shoppers and a premium level, high quality product line (Geyskens et al., 2010). The last one should be able to compete against the high tier, which are the premium quality national brands. Hence, this leads to major differences in the quality perception consumer hold towards private labels.

These different private label tiers ensure that one cannot assume private label brands to be low tier brands based on the sole fact that it is a private label. Moreover, a private label brand can become a high tier brand when consumers perceive the price-quality to be high. Kumar & Steenkamp (2007) distinguish two types of high tier private label propositions: 1) premium lite and 2) premium price. Premium Lite is a private label proposition that is characterized by better or equal quality than the leading manufacturer, while selling at a lower price (Kumar & Steenkamp, 2007). Retailers who apply this proposition benefit from their settled brand name, while competing on price. However, this lower price could influence the perceptions that the private label still is inferior to the national brand. It takes retailer effort to prove the customer that the product is better or equal, unless the private label is delinked from the store (name). The second type of high tier private label is Premium Price, where the quality is superior and prices higher compared to the leading manufacturer (Kumar & Steenkamp, 2007). Retailers who apply this proposition actively compete with the premium national brands on quality and brands values, by labeling the premium level private labels under a different name than the one that has no mention of the store brand. Delinking allows the private label more pricing and positioning flexibility.

(17)

17

Furthermore, Geyskens et al (2010) found when quality variation of private label lines increases through up- or downscale extensions, consumers become less confident in the private label brand name as a signal of a quality level. This becomes visible by high cannibalization effects of the standard level private label after the introduction of a premium level private label line (Geyskens et al., 2010). Although these additional tiers influence a larger variation in the quality perception consumers hold, they perceive (mid-tier) private label brands still as lower in quality compared to national brands (Kumar & Steenkamp, 2007). Hence, mid-tier private labels still can be considered low tier brands.

2.3 Price Sensitivity across Categories

Consumers can be high brand tier shoppers and low brand tier shoppers simultaneously, when they value high tier brands in one category, and are satisfied with a low tier brand in another category. This makes it interesting to look for category differences. In some product categories consumers are more likely to respond to promotions than in other categories, which has effect on the marketing effectiveness of promotions. Promotions that are effective in one category are not necessarily effective in another category. Like the specific promotion effect, brand switching is subject to these categorical differences (Van Heerde et al., 2003). Moreover category specific factors have greater impact on brand switching than brand-specific of consumer specific factors (Bell, Chiang & Padmanabhan, 1999). One way to compare promotion effects of categories is on their level of price elasticity. Research described price sensitivity as an actor in why consumers respond to price promotions (Krishnamurti & Raj, 1991). Based on this theory it seems logical to assume that magnitude of brand switching and the level of price sensitivity are strongly related. Hence, it is useful to check whether price sensitivity causes the differences in the magnitude brand switching over categories.

(18)

18

Narasimhan, Neslin & Sen (1996) found that categories characterized by high prices often indicate high savings during price promotions and an increased promotion effect. Whereas at the same time categories characterized by high prices mean a greater brand loyalty and less brand switching. Bell, Chiang & Padmanabhan (1999) found similar results for categories characterized by high share of wallet. This study found for high share of wallet and storable products that higher loyalty reduces the sales promotion effect of brand switching, whereas it increases the impact of primary demand effects (Bell et al., 1999). A reason for this result could be that consumers are more likely to have strong preferences for items that have a large share of wallet (Bell et al., 1999).

Brand switching always involves the risk that the promoted brand is not able to meet the expectations. This risk prevents loyal customer to switch brands. Hence, high share of wallet will stimulate the promotion effect of stockpiling and deal-to-deal purchasing, and has a negative impact on brand switching (Bell et al., 1999).

At the same time, prices are more salient in categories with high shares of wallet and categories where consumers are price sensitive. Steenkamp & Dekimpe (1997) found that the perceived quality is the primary factor that ensures that private labels fail in some categories. Extensive advertising and brand building generates high levels of quality perceptions and brand equity for high tier brands. This can generate trust and reduces the risk of underperformance for consumers. Although high share of wallet reduces sales bumps caused by brand switching, the level of brand equity of the high tier brand can provide enough cues for trust. Especially for those high brand tier shoppers who not necessarily are loyal to one specific brand, but for example switch between a subset of high tiered brands. For these consumers the brand tier provides trust, which reduces the risks involved with brand switching. Hence, when the prices are salient in price sensitive categories, high brand tier shoppers are more likely to switch to other high tier brands, compared to price insensitive categories.

Research has found that price promotions are more effective in price sensitive categories

(Narasimhan et al., 1996; Nijs, Dekimpe, Steenkamp & Hanssens, 2001). Since brand switching is a dominant effect from price promotions (Bell et al., 1999), it is hypothesized that consumers are more likely to switch brands in price sensitive categories. Thus:

H4: The magnitude of brand switching is larger in categories characterized by high price sensitivity than in categories characterized by price insensitivity.

(19)

19

2.4 Covariates

Besides promotions there could be other variables influencing consumers to switch brands. Store specific factors for example, can influence promotion effectiveness (Lam, Vandenbosch, Hulland & Pearce, 2001). Loyal customers to store A are more likely to be affected by promotions in store A than in store B. Furthermore, when a customer visits store B it could be the case that its favorite brand is not available and therefore switches brands. Hence, the model for brand switching should account for store (chain) loyalty.

Consumers can be more loyal to one brand than to other brands. Likewise it is possible that they switch between a subset of brands. Research found that brands with more repeated purchases and more loyal consumers are less affected by brand switching as a result of other brands’ promotions (Krishnamurthi & Raj, 1991). Moreover, these brands generate greater purchasing quantity effects as a result of sales promotions (Bell et al, 1999). If consumers score high on brand loyalty it decreases their motivation to switch brands. Similarly, loyalty to a specific package type of package size can withhold consumers from brand switching. Such preferences can be consumers’ attributes that act similar as brand loyalty (Guadagni &Little, 1983).

Another factor that could influence switching behavior is consumer demographics. Bell, Chiang & Padmanabhan (1999) found that age has a negative effect on brand switching, and concluded that older people are more loyal to brands. Other characteristics like family size, occupation and disposable income may also influence brand switching. An increase in family size will increase the motivation to look for deals, and increases switching behavior (Gupta, 1988). A higher income or better profession can reduce the motivation to look for discounts, and therefore are more loyal. However, Bijmolt, van Heerde & Pieters (2005) could not find any support for income. Therefore this will not be included in the measurement.

2.5 Magnitude of Brand Switching

(20)

20

Therefore van Heerde, Gupta & Wittink (2003) contributed with a complementary approach in the combination of elasticity and unit sales decomposition in order to allow for more complete assessment of sales promotion effectiveness. They found, measured across several categories, an elasticity decomposition of approximately 75% and unit sales decomposition of 33% on average. This complementary approach shows a different assessment of sales promotion effectiveness.

Hence, the magnitude of the promotion effect of brand switching is on average 33% of the gain in unit sales (van Heerde et al., 2003). However, this is the aggregate effect for all consumers over different categories. Assuming that low brand tier shoppers are more likely to switch brands than high tier shoppers, this average percentage of switchers are probably higher for low tier shoppers, and lower for high brand tier shoppers, than the 33% found by van Heerde et al (2003).

In short all hypotheses together leave the following conceptual model, shown in figure 1.

FIGURE 1 CONCEPTUAL MODEL

Price Sensitivity of the

Promotion

Type of Brand tier

(21)

21

3. METHODOLOGY

3.1 Measurement

Treatment

To test the hypotheses that were developed in chapter 2, a brand sales model is developed and tested. First, to account for the different brand tiers, the sales of different kinds of brands were estimated. For the low brand tiers, private label products were selected. As discussed in chapter 2, private labels can come in different brand tiers. Nevertheless, consumers still perceive mid-tier private labels as lower in quality compared to national brands (Kumar & Steenkamp, 2007). To control that the brand tier actually is low, the most regular variant of the private label product was selected and the prices were controlled to be lower than those of the leading national brands. Private labels are available in almost every product category in grocery stores (Nielsen, 2014). This accounts for a sufficient number of data points in all selected categories. For these reasons it is appropriate to use private labels as low tier brands, and national brands as high tier brands. The bestselling private label item that fulfilled these demands was selected.

To account for the high brand tier effects, the two bestselling A-label brands that shared the product characteristics (product type, package size and/or sugar level) of the private label were selected. Pearson correlation was used to detect correlations of brand sales and preclude a selection of brands which have no cross-brands effects at all. Appendix A (Table A2) includes a more detailed description of the selected items for measurement. Based on these brands, the model is able to account for cross-brand effects from both high tiered and low tiered cross-brands.

(22)

22

Because of the lack of cross-brand correlations it was decided to select tortilla chips instead of potato chips. Tortilla chips has less differences between items, which simplified the process of selecting appropriate candidate brands for the measurement. Moreover the category still scores severe on price sensitivity, which allows the category to be measured as a price sensitive category.

Consumers can be loyal to a specific package size, however the promotion can stimulate consumers to increase the purchase quantity or to buy a different size (Gupta, 1988). These consumers are not necessarily switching brands, but only switching package sizes. This should be accounted for during the measurement. Consumers who switched package size during the promotion were not considered as brand switchers. Finally, store chains have different retail strategies. These strategies determine among others the pricing strategy and the type of promotions the chain uses. Therefore the chain is fixed in the measurement.

Data description

The analysis used store level scanner data for three brands of frequently purchased consumer (food) product over the years 2001 until 2005. To avoid biases, only one retail chain is selected in the measurement. This chain is chosen based on the availability of a private label in the selected categories. A chow test is performed on the models for the categories mayonnaise and peanut butter. Both categories showed with highly significant results that pooling is not allowed. Therefore the store of which the private label sales significantly correlated with the price and display variables of the a-label brands over all categories is chosen for further analysis. In this way there will be no bias by store differences. The selected store is located in San Diego and is performing under the sample mean. In the first year of the study the estimated annual sales in dollars was for the selected store $13.2 million, whereas the estimated annual sales of the total sample was $17.5 million on average (SD = $13.2 million). During the measurement the store increased in performance similarly to the sample mean. In year 5 the mean of the estimated sales in dollars for all stores is $18.4 million (SD = $14.4 million), whereas the estimated annual sales for the selected store is about $14 million.

(23)

23 TABLE 1 DATA DESCRIPTIVES Paper Towels Frozen Pizza Tortilla Carbonated

Soft drinks

Mayonnaise Peanut Butter

high low high low high low high low high low high low

N 217 248 204 172 130 120 104 104 260 260 260 260 Sales min 3 3 1 1 1 2 37 2 7 1 1 1 Sales max 105 243 49 46 57 27 308 28 271 24 138 107 Sales mean 35.27 28.60 6.47 7.01 12.92 8.03 118.09 8.25 43.61 7 15.93 10.55 Sales SD 1.324 2.31 6.571 6.944 8.783 6.19 49.473 4.285 33.437 5.21 14.88 10.626 Price median 1.99 1.799 3.5 3.34 2.947 1.59 1.336 .79 3.49 2.59 2.99 2.39 Feature 50 9 60 37 38 3 101 24 37 18 29 28 Display - 5 - - 29 - 41 4 28 - - 5

(24)

24

3.2 Model and Estimation

In order to measure the proportion of the brand switching, a multiplicative sales model is appropriate. This model is attractive to use, since it allows for several explanatory variables. Besides, it allows for estimation of the own and cross brand elasticities (Bijmolt, 2005). These elasticities were used to compare the own brand effects to the cross brand effects across categories. Where the own brand elasticities accounts for the primary demand effects and the cross brand elasticities represent the brand switching effects. The dependent variable in the model is unit sales of brand j, for which is taken the natural logarithm. Where j is the number of brands, i represents the store, and t is the number of weekly observations per store.

𝑆𝑖𝑗𝑡 = Unit sales of brand j, in week t in store i. Data transformation

The model includes different levels of feature promotion, because the dataset made a clear distinction between these levels. Although there was no clear explanation of what the different levels mean. Given that a large promotion could have a broader scope and therefore a greater effect than medium sized advertising has on sales, the levels were kept as given in the data. However, this distinction is not clear and this is just an assumption. Therefore it does not hypothesize any significant difference in effects. The variable for display promotion is split into minor display promotion and end-of-the-aisle display promotion to include the effects of separating promotion instruments.

Unfortunately a multiplicative model only allows dummy variables in the model in the form of multipliers. Multipliers will not generate elasticities, which can be compared across categories. By recoding those variables into price indices when promotion is present, the variance is still included in the model. Besides it allows to use the variable as an elasticity. Since the values are transformed into price variables no conclusions can be made about non-price effect without the support of price changes. To avoid multicollinearity issues, the price index without support of other promotion instruments is recoded likewise, i.e. in all cases feature advertising is present, the price index for feature advertising take the value of the price index, while the variable price index without support takes the value 1. Vice versa in all cases feature advertising is not present the variable for price index supporting feature advertising take the value 1. For all values of these variables is taken the natural logarithm, except for the outlier variable, which is a dummy variable.

(25)

25 Explanatory variables

The study includes the following independent variables. For all variables is taken the Natural logarithm, except for the values of the outlier variable.

𝑂𝑊𝑁_𝑃𝑗𝑡 = The price index of brand j in week t. The index is calculated as the price divided by the

median of the price. This variable includes the price variance without the support of other promotion instruments. When other promotion instruments are applied for brand j, the variable will take the value 0 (after taking the LN).

𝐶𝑅𝑂𝑆𝑆_𝑃𝑗𝑡 = The price index of the competing brands of brand j, without the support additional

promotion instruments, in week t. The index is calculated similar as the own price index. When other promotion instruments are applied for the cross brand, the variable will take the value 0 (after taking the LN).

𝑂𝑊𝑁_𝑃𝐼𝐹𝑚𝑗𝑡 = The price index for brand j in week t, if medium sized feature advertising is applied. If

not applied the variable takes the value 0 (after taking the LN).

𝐶𝑅𝑂𝑆𝑆_𝑃𝐼𝐹𝑚𝑗𝑡= Like OWN_Fm, the price index for the competing brands of brand j in week t, if

medium sized feature advertising is present. If not applied the variable takes the value 0 (after taking the LN).

𝑂𝑊𝑁_𝑃𝐼𝐹𝐿𝑗𝑡 = The price index for brand j in week t, if large sized feature advertising is applied. If not

applied the variable takes the value 0 (after taking the LN).

𝐶𝑅𝑂𝑆𝑆_𝑃𝐼𝐹𝑙𝑗𝑡= Like OWN_Fl, the price index for the competing brands of brand j in week t, if large

sized feature advertising is present. If not applied the variable takes the value 0 (after taking the LN).

𝑂𝑊𝑁_𝑃𝐼𝐷𝑗𝑡 = The price index for brand j in week t, if display promotion is present. If not applied the

variable takes the value 0 (after taking the LN).

𝐶𝑅𝑂𝑆𝑆_𝑃𝐼𝐷𝑗𝑡= Similar to OWN_Display, the price index for the competing brands of brand j in week

t, if display promotion is present. If not applied the variable takes the value 0 (after taking the LN).

𝑂𝑊𝑁_𝑃𝐼𝐴𝑖𝑠𝑙𝑒𝑗𝑡=The price index for brand j in week t, if end-of-the-aisle display promotion is present.

(26)

26

𝐶𝑅𝑂𝑆𝑆_𝑃𝐼𝐴𝑖𝑠𝑙𝑒𝑗𝑡= Similar to OWN_Aisle, the price index for the competing brands of brand j in week

t, if end-of-the-aisle display promotion is present. If not applied the variable takes the value 0 (after taking the LN).

𝑃𝐼_𝐿𝑒𝑎𝑑𝑗𝑡+1= Brand price index of brand j, in week t+1.

𝑃𝐼_𝐿𝑎𝑔𝑗𝑡−1= Brand price index of brand j, in week t-1.

𝑂𝑢𝑡𝑙𝑖𝑒𝑟𝑠𝑗𝑡 = Multiplier for outliers in the dataset, caused by extremes or omitted variables. The 1

when the case is indicated as an outlier, 0 for all other cases. The model to estimate the cross-brand elasticities in week t, is given in equation 1. 1) 𝑆𝑖𝑗𝑡 = 𝛼𝐵𝑅𝐴𝑁𝐷,𝑗 (𝑂𝑊𝑁_𝑃𝑗𝑡𝛽𝑂𝑊𝑁_𝑃,𝑗 𝐶𝑅𝑂𝑆𝑆_𝑃𝑗𝑡𝛽𝐶𝑅𝑂𝑆𝑆_𝑃,𝑗 𝐶𝑅𝑂𝑆𝑆_𝑃𝐼𝐹𝐿𝑗𝑡𝛽𝐶𝑅𝑂𝑆𝑆_𝑃𝐼𝐹𝐿𝑗 𝑂𝑊𝑁_𝑃𝐼𝐷𝑗𝑡𝛽𝑂𝑊𝑁_𝑃𝐼𝐷𝑗 𝐶𝑅𝑂𝑆𝑆_{𝑃𝐼𝐷𝑗𝑡}𝛽𝐶𝑅𝑂𝑆𝑆𝑃𝐼𝐷𝑗_𝑂𝑊𝑁 𝑃𝐼𝐹𝑚𝑗𝑡𝛽𝑂𝑊𝑁𝑃𝐼𝐹𝑚,𝑗 𝐶𝑅𝑂𝑆𝑆𝑃𝐼𝐹𝑚𝑗𝑡𝛽𝐶𝑅𝑂𝑆𝑆𝑃𝐼𝐹𝑚,𝑗 𝑂𝑊𝑁𝑃𝐼𝐹𝐿𝑗𝑡𝛽𝑂𝑊𝑁𝑃𝐼𝐹𝐿,𝑗 𝑂𝑊𝑁_{𝑃𝐼𝐴𝐼𝑆𝐿𝐸 𝑗𝑡}𝛽𝑂𝑊𝑁𝑃𝐼𝐴𝐼𝑆𝐿𝐸,𝑗_{𝐶𝑅𝑂𝑆𝑆} 𝐴𝐼𝑆𝐿𝐸 𝑗𝑡 𝛽_{𝐶𝑅𝑂𝑆𝑆𝐴𝐼𝑆𝐿𝐸𝑗} 𝑃𝐼_{𝐿𝐸𝐴𝐷𝑗𝑡+1}𝛽𝑃𝐼𝐿𝐸𝐴𝐷,𝑗_𝑃𝐼 𝐿𝐴𝐺 𝑗𝑡−1𝛽𝑃𝐼𝐿𝐴𝐺,𝑗 𝛾𝑂𝑢𝑡𝑙𝑖𝑒𝑟𝑠𝑗𝑂𝑢𝑡𝑙𝑖𝑒𝑟𝑠𝑗𝑡)𝜀𝑖𝑡

Estimation and model selection

The estimation of the model is based on the first 90% of the data, the remaining 10% of the data is kept for (prediction) validation. Before applying the model, possible confounds were searched. The Pearson correlation statistic was used to detect correlations between the independent variables. When the correlation was significant, multicollinearity issues were overcome by re-coding of the independent variables. This was present for display and feature promotions, for which an interaction variable was added to the model. In the case of frozen pizza the lag and lead variables were strongly correlating to the price variable. Because the prices in these categories were merely constant and price changes maintained for a couple weeks, it was decided to eliminate the (lag or lead) variable that has the smallest parameter estimate.

(27)

27

(28)

28

4. RESULTS

4.1 Face Validity

The model fit is highly significant for all models (F > 5.614, p < .000). As presented in appendix B (table B1) the adjusted R-squares are all between 0.215 and 0.716, which is sufficient.

Price parameters without support

Table 2 presents the parameter estimates of the model for each category. In general, the own price parameters of the brand sales resulted as expected. The elasticities for own price show significant and negative in all categories. This confirms the basic theory of economics; the sales levels decrease when prices rise and vice versa. One exception is low brand tier model of carbonated soft drinks, for which none of the price only variables had a significant effect on the sales. Several reasons could have caused this effect. First, the category is categorized to be price elastic (Gordon, et al. 2013). This could imply that an increase in price does not have strong effects on the sales. This argument would hold for a small effect of price, however no effect at all, as it is in this case seems extreme and unlikely. Another explanation arises from the data description in table 1. This category is subject to intense promotion activity. Whereas the variable price is indicating the price without support of promotion instruments, which leaves only a few data points for the high tier brand. Although the low tier brand does have many weeks without feature or display support. The lack of significant effect of price parameters without support could have been influenced this effect. Since the price (promotion) is often supported by feature, display or end-of-the-aisle display promotion, the variance of price is most likely explained by these price with support variables.

(29)

29

Lastly, the selection of brands could explain these unexpected results. The high tier brands in the tortilla chips and frozen pizza categories were not the bestselling brands, whereas the high tier brands often were the bestselling brands in the other categories. The selected brands were chosen based on their shared product characteristics to the bestselling private label brand in the category. This could have influenced the unexpected results of tortilla chips and frozen pizza. Possibly this could have biased the results if the selected high tier brands are actually low tier brands. An increase in price than could have influenced the quality perception consumers hold, and switched to the brand with the highest quality.

Quality perceptions are not formed based on price only. Brand name, brand associations, and product composition characteristics are determinants of perceived quality too (Jacoby, Olson & Haddock, 1971). Hence, there is no satisfying argumentation why the price parameters for cross brand effect in the categories tortilla chips and frozen pizza are negative.

Price parameters with feature and/or display support

(30)

30 Dynamic effects

(31)

TABLE 2 ELASTICITIES

Price Sensitive Categories

Paper towels Frozen pizza Tortilla chips

highᵇ lowᵇ highᵇ lowᵃ highᵃ lowᵇ

Elasticity Standard Error Elasticity Standard Error Elasticity Standar d Error Elasticity Standard Error Elasticity Standard Error Elasticity Standard Error OWN_P -2.347*** .303 -2.755*** .441 -2.714*** .269 -3.344*** .397 -1.727*** .562 -1.341*ⁱ .694 PI_Lead -.323** .161 NS. - NS. - NS. - NS. - NS. - PI_Lag -.360** .155 -.848** .385 NS. - NS. - NS. - NS. - OWN_PIFm -.679*** .191 -1.647*** .501 NS. - -4.133*** .465 NS. - NS. - OWN_PIFL NS. - NS. - -3.472*** .684 -4.704*** .663 -1.094*ⁱ .646 NS. -

OWN_PID NA. - NS. - NA. - NA. - -4.429*** 1.352 NA. -

OWN_PIAisle NA. - NS. - NA. - NA. - -4.006*** 1.887 NA. -

OWN_Fm*Aisle NA. - NA. - NA. - NA. - -1.812** .711 NA. -

OWN_FL*Aisle NA. - NA. - NA. - NA. - -1.541*** .441 NA. -

CROSS¹_P NS. - 1.726** .730 NS. - NS. - -1.603*** .514 -2.100*** .798

CROSS²_P NS. - NS. - -,943*ⁱ .493 .589** .234 -.604ⁱ .414 NS. -

CROSS¹_PIFm NS. - NS. - NS. - NS. - -1.413** .671 NS. -

CROSS²_PIFL NS. - .995** .500 NS. - NS. - NS. - .757**ⁱ .649

CROSS¹_PIAisle NS. - 3.756** .730 NA. - NA. - NS. - NS. -

Outliers -1.352*** .157 NS. - -1.901*** .259 NS. - -.743*** .187 -1.750*** .129

(32)

32

TABLE 2 CONTINUED Price Insensitive

4.2 Results

The elasticities presented in table 2 are compared amongst each other in order to verify the hypotheses developed in chapter 2.

Promotion effects Hypothesis 1

(34)

34

Subsequently the magnitude of the cross price without support elasticity is compared to those of the cross price supported by feature and display promotion. The cross brand price elasticities supported by feature and display promotions are even greater than the price only parameters for cross brand effects in the categories paper towels (cross PI aisle¹: B = 3.756 SD = .730 > cross price¹: B = 1.726 SD = .730), and peanut butter (cross¹ PI aisle: B = 2.284 SD = .800 & cross¹ PI feature: B = 1.274 SD = .586 > cross price¹: B = 1.119, SD = .289 & cross price²: B = .707 SD = .191). In the categories mayonnaise and carbonated soft drinks cross price without support elasticities did not have a direct effect, but the price indices with support of other promotion instruments does. Hence, hypothesis 1 is confirmed.

Hypothesis 2

The magnitude of separating promotions is measured by use of the price index supported by end-of-the-aisle promotions. In line with Lemon and Nowlis (2002), the results show strong effects of separating promotion effects. When comparing the price sensitive categories amongst each other, the results show that paper towels is the only category which sales is strongly affected by cross brand separating instruments. Still, the end-of-the-aisle-display promotion by high tier brand 1 has a large effect on the own brand sales of the low tier brand (B = 3.756, SD = .730, p < .000). The separating promotion effects in the price insensitive categories showed more often significant results.

(35)

35 Brand tiers

As discussed in the paragraph about promotion effects, the magnitude of switchers is greater from low tier brand to high brand tier than vice versa. In this part the effects of rival high tier brands on the sales of another high tier brand is measured and discussed. Whether the effect of high brand tier shoppers switch to another high tier brand is significant, is captured in the Cross¹ brand effect on the high brand tier model. The magnitude of the Cross¹ brand effects on the own sales represents the magnitude of switchers to that particular rival brand. When comparing the cross brand¹ effect in the low tier model to those in the high tier model hypothesis 3 can be tested. The other cross brand effects (= Cross² brand effects) are not used for comparison, in order to keep the brand stable for cross brand effects in both brand tiers.

On one hand the category carbonated soft drinks is subject to cross brand effects. The high tier brand price supported by medium sized features influences consumers to switch brands, whereas this is not significant for the low tier brand sales. On the other hand low brand tier shoppers are likely to switch brand to a high tier brand when the rival brand is promoted by large feature advertising. Whereas this does not affect the high tier brand sales. However, since these variables are both feature advertising. These results suggest that for carbonated soft drinks the magnitude of switchers from low tier brands to high tier brands is greater, than from high tier brands to another high tier brand (cross¹ PI FL: B = 1.216, SD = .686 > cross¹ PI FM: B = .731, SD = .208). Similarly in the other categories the cross brand elasticities are larger in the sales model for low tier brand compared to those of high tier brand. Hence, these results confirm the expectations described in hypothesis 3.

(36)

36 Category differences

Hypothesis 4

Hypothesis 4 can be tested by comparing the magnitude of cross brand elasticities over types of categories. There are multiple ways to check this hypothesis. Table 2 shows that the price insensitive categories have a higher number of significant variables (19 out of 47) that impact the own sales, than the price sensitive categories have (9 out of 28). This extra number of variables is caused by the number of interaction variables, which were necessary in some categories to avoid multicollinearity. The number of significant interactions suggest that these variable do include extra variance, and therefore were necessary to include. Based on the number of significant variables, the price insensitive variables are more affected by brand switching than the categories characterized by high price sensitivity. When comparing the individual magnitude of each elasticity, the cross brand variables of paper towels have the greatest impact on own sales. This result supports the hypothesis. Conversely, the cross brand effect of price insensitive categories mainly depends on the many large and significant cross brand variables in the low tier models of peanut butter and mayonnaise. At the same time the predictive validity of the low tier model of mayonnaise is very bad. This makes the results of the mayonnaise low tier model less valid to use as an argument against the confirmation of the hypothesis. Nevertheless, paper towels is the only price sensitive category that accounts for strong cross brand effect, which makes the results less reliable. Hence, multiple reasons inhibit to confirm or reject the hypothesis easily. Since the difference in number of variables is present and inevitable, the average effect of the cross brands is calculated and presented in table 4, in order to find more arguments for consideration.

TABLE 4

SIZE OF THE CROSS BRAND EFFECTS ON THE LOW BRAND TIER MODELs

Price sensitive Price insensitive

Number of significant variables

Average effect Number of

significant variables

Average effect

Paper towels 3 2.159 Peanut butter 5 1.166

Frozen pizza 1 .589 Mayonnaise 6 1.3

Tortilla chips 2 -.672 Carbonated soft

drinks

2 1.122

(37)

37

As mentioned before the categories of tortilla chips and frozen pizza had unexpected results and suffered by missing data, therefore these categories generated less reliable results. Since the predictive validity and model significance were sufficient for frozen pizza, it is not possible to completely ignore the results. Therefore, it can be conclude that the results indicate that price sensitive categories are more affected by cross brand effects than price insensitive categories are. However, the hypothesis cannot be confirmed.

Hypothesis 5

The last hypothesis focuses on the effect of category on the magnitude of switchers from high tier to high tier brand. Therefore a similar measurement is used as for hypothesis 3. As mentioned earlier, the high tier brands do not suffer strongly from cross brand effects. However, when the high tier brand models are affected by high tier cross brand effects, the high tier cross brand has greater impact on high tier brands in price insensitive categories than on high tier brand in price sensitive categories. This suggests that high tier brands in price insensitive categories suffer more from cross brand promotion activity, which is the opposite of the hypothesized effect.

(38)

38

5. DISCUSSION

5.1 Discussion of the Results

Promotion instruments

The results showthat non-price promotion instruments increase the effect price promotions have on brand switching. Since the parameter estimates needed to be all elasticities in order to compare them across categories, it was not possible to measure the single effect of non-price promotions. Cross brand promotion influences the magnitude of brand switching. Still, the results prove that feature and display promotions are highly effective and influence the magnitude of brand switching. Also the results confirm the study by Lemon & Nowlis (2002), and conclude that separation promotions are highly effective. Extra brand attention and interaction effects of promotion instruments increased the brand switching. Hence, the use of non-promotions generates extra attention towards the brand and increases the switching effect.

More specifically, the results found that separation promotions are highly supportive to switch brands as well. However, this effect is not strong enough to decrease the effect of loyalty. Therefore, separation promotion did not have stronger cross brands in price insensitive categories than price sensitive ones, as is described in hypothesis 2. This result might be explained by the type of measurement. The hypothesis assumes that when separating promotions are applied there are no price cuts, whereas the variable tested did include prices when the promotion was present. Consumers of price sensitive categories can be stimulated to switch because of the price cut or because of the perception that end-of-the-aisle display promotion is always combined with a price cut.

(39)

39 High versus low brand tier shoppers

Although the study did not find enough evidence to conclude that price sensitivity influences the magnitude of brand switching, the study proved that the difference in brand tiers does affect the magnitude of brand switching.

The price sensitivity level does not influence the likelihood that high tier consumers will switch to another high tier brand. This confirms studies that find that shoppers of high tier brands are brand sensitive, and weigh the brand equity strongly in their utility perceptions (Blattberg & Wisnieuwski 1989; Lemon & Nowlis, 2002). The perceived utility for high tier brands is at a certain level that is not easily outweighed by a rival high tier brands’ promotion activities. Whereas this effect is strongly visible for low tier brands. So, brand loyalty remains to have a strong influence on the perceived utility, which inhibits high brand tier shoppers to switch.

Meanwhile shoppers of low tier brands are unlike the high brand tier shoppers not loyal to the brand. Low brand tier shoppers are strongly affected by promotions of high tier brands. The results confirm the theories that found the majority of the brand switchers are shoppers of low brand tiers (Sethuraman, 1996; Blattberg & Wisnieuwski, 1989). Thus, promotions keep attracting low tier shoppers to make a single purchase. This makes that a part of the promotion budget is spilled to these customers. However, the loss in this situation is only relative. Brand switching is just one part of the sales bump. When primary demand effects outweigh the brand switching effect, the promotion could still be a success.

(40)

40

5.2 Recommendations

In general as Bijmolt et al (2005) find that the overall brand sensitivity has increased. This study contributes to this theory finding that the categories have become more equal to each other in terms of being characterized as price sensitive. The fact that the study did not find enough evidence to indicate that the level of price sensitivity of the category has impact on the magnitude of brand switching. This implies that consumers are sensitive to price changes and promotions in all categories. Not just the price sensitive categories, but the considered price insensitive categories are subject to these price sensitive (low tier) consumers too. The increased overall price sensitivity strongly influenced low tier shoppers to respond to promotions.

Although the brand loyalty preventhigh tier shoppers to change brands, this brand loyalty is an actor that stimulates these consumers to perform primary promotion behavior, as a result of this increased overall price sensitivity (Bijmolt et al. 2005). As discussed in the literature review section, increased price sensitivity and learning effects stimulated consumers to perform deal-to-deal purchasing behavior and other primary promotion effects (Bijmolt et al., 2005; Nijs et al., 2001). This suggests that the highest sales bumps come with price cuts and decreased margins. Price cuts have negative effects on the reference prices. As reduced reference prices could lead to decreased brand equity, price promotions are not beneficial on long term (Blattberg et al., 1995). This means that brands will increasingly suffer from post-promotion dips and decreased margins. Consumers are sensitive to price promotions. So unless consumers really feel necessary to buy the product in unpromoted periods, they will postpone the purchase to times that promotions are present.

(41)

41

Loyalty programs will enable retailers and brand manufacturers to offer individualized promotions. These individualized promotions allows managers to target specific consumers for each promotion based on their individual consumption behavior. Therefore, the low brand tier shopper, who switch more often, will not targeted for promotions of brands that they do not regular buy, which reduces the sales bump caused by brand switching.

This study found that sales bumps caused by brand switching are mainly caused by low tier shoppers that switch to higher brand tiers, than for switchers between high tier brands. This does not imply that cross brand effects across high tier brands are not present at all. Therefore brand manufacturers should monitor the reduced sales of their own brands, when rival brands are promoted. Especially when brand manufacturers produce the private label products besides their own brands.

5.3 Limitations and Ideas for Future Research

This study suffered from limitations. In order to keep the models strong and significant only one store is used for the measurement, which implies that only within-store variation of price and promotions is used. Other stores and retail chains could have different effects and specifically encounter effects of different magnitudes. Especially since the selected store did not often use end-of-the-aisle-display promotion, which had a direct effect on the results of hypothesis 2. Moreover, the selected store suffered from gaps in the data in several categories. For example the models for tortilla chips, frozen pizza and carbonated soft drinks suffered from a small data set. This could have biased the results, and the reliability of the model. Hence, the selection of the store used for measurement has impact on the generalizability of the results.

(42)

42

Furthermore the study is based on grocery sales data, which are characterized fast moving consumer goods. The results are therefore not necessarily generalizable for other markets. The fact that customers need to renew the purchase in a relative short term can influence the impact of brand loyalty. In markets were the time between purchase and a renewal of the purchase is longer, this impact of brand loyalty for high tier brands can be lower. The involvement to make the best purchase could influence in those markets that more (high tier) brands are compared among each other. This can imply that the magnitude from brand switchers between high tier brands are greater. Likewise the gap between the magnitudes of switchers from low tier to high tier versus switchers from high tier to another high tier brand can be lower in those situations. On the other hand, markets similar to supermarkets, which are characterized by frequent price promotions (i.e. consumer electronics) could benefit from these results. However, in markets where consumers are less educated about the specific differences between the specific high tier brands, the magnitude of switchers among high tier brands will probably be greater.

(43)

43

6. Conclusion

Price sensitivity and brand switching seem to be closely related. Price has a large impact on the attractiveness of an offer (Ataman et al., 2010). Furthermore, price promotions have the negative effect of decreasing brand equity in the long run (Blattberg et al., 1995). Over years the overall price sensitivity has increased (Bijmolt et al., 2005). This makes it relevant to investigate how brand switching is related to price sensitivity, which led to the research question:

“Is the magnitude of the brand switching effect on the sales bump affected by the level of price

sensitivity of the category? And does this differ for perceived brand tiers?”

A multiplicative sales model is developed, tested and used to generate cross brand elasticities for different categories and brand tiers. The elasticities have shown that there are indications that there are differences in the magnitude of brand switching between price sensitive and price insensitive categories. However the results were inconclusive, which inhibit conclude that high price sensitivity lead to increased magnitudes of brand switching.

Moreover, the study found that the type of brand tier influences the magnitude of brand switching. Shoppers of low tier brands are regardless the price sensitivity of the category strongly affected by promotions of high tier brands. Vice versa, shoppers of high tier brands are not affected by price changes or promotion of low tier brands. However these shoppers of high tier brands are to a small extent affected by promotions of rival high tier brands.

PRICE SENSITIVITY AND BRAND SWITCHING

PRICE SENSITIVITY AND BRAND

SWITCHING

A COMPARISON OF BRAND TIERS ACROSS PRODUCT

CATEGORIES

PRICE SENSITIVITY AND BRAND

SWITCHING

A COMPARISON OF BRAND TIERS ACROSS PRODUCT

CATEGORIES

MANAGEMENT SUMMARY

PREFACE

TABLE OF CONTENTS

1. INTRODUCTION

2. LITERATURE REVIEW

2.1 Promotion Effects

2.2 Brand Equity and Brand Tiers

2.3 Price Sensitivity across Categories

2.4 Covariates

2.5 Magnitude of Brand Switching

Price Sensitivity of the

Category

Promotion

Type of Brand tier

3. METHODOLOGY

3.1 Measurement

3.2 Model and Estimation

4. RESULTS

4.1 Face Validity

4.2 Results

5. DISCUSSION

5.1 Discussion of the Results

5.2 Recommendations

5.3 Limitations and Ideas for Future Research

6. Conclusion