Master Thesis Competitive price reactions in the Dutch food retail market: An empirical study

(1)

Master Thesis

Competitive price reactions in the Dutch

food retail market: An empirical study

University of Groningen Faculty of Economics and Business

Department of Marketing

MSc Marketing Management & Marketing Intelligence

Thijs Kegelaar Bedumerweg 43 9716AD Groningen +31 620259304 T.J.H.Kegelaar@student.rug.nl Student number: S2599643

(2)

(3)

1

Acknowledgement

Before diving in the complex world of competitive reactions, I would like to take the opportunity to express my gratitude to the persons who helped me with the realization of this thesis. First of all, I would like to thank my supervisor Prof. Dr. Laurens Sloot. His critical and useful feedback always steered me in the right direction when needed. Second, I would like to thank Dr. J.E.M. van Nierop for sharing his interesting and useful thoughts about the data. This really helped creating the foundation of this thesis. Third, I would like to thank Andries Mooij and Alwin van der Velden from Superscanner.nl for providing the data and sharing relevant insights. Last, I would like to thank my fellow thesis students for sharing their thoughts and pleasant meetings.

(4)

(5)

3

Competitive price reactions in the Dutch food retail

market: An empirical study

Thijs Kegelaar

University of Groningen, The Netherlands

ABSTRACT

This study focuses on the underlying variables that are influencing the decision to react to price changes of competitors, the reaction speed of this competitive reaction and the strength of this reaction. The aim of this study is to develop a better understanding of how food retailers in the Dutch market respond to their competitors. First, it is

investigated when retailers respond. Second, variables that influence the speed and strength of these reactions are revealed. The findings indicate that case specific variables are the largest determinants of competitive reactions. Category and product related variables have less significant impact. Moreover, findings show that retailers have different reaction strategies. Larger retailers make use of category and product variables and show passive behavior more often. Smaller retailers are more responsive and solely use of case specific variables.

Keywords: Retail; Competitive reactions; Dutch food retail; Reaction speed; Strength

competitive reaction

1. INTRODUCTION

The Dutch supermarket price war, initiated by Albert Heijn in 2003, shows the importance of competitive reactions. Understanding competitive reactions and

(7)

5

reactions will be identified and explained. The questions when and why retailers respond and, maybe just as important, why and when retailers do not respond will be tried to answer.

The findings of multiple studies shows it is common for retailers to use the same instrument to react on competitive actions (Steenkamp, Nijs, Hanssens and Dekimpe, 2005; Leeflang and Wittink, 1991). Therefore, in general, advertisement of a competitor is countered with advertisement, price promotions of competitors are countered by price promotions etc. Besides the type of competitive response, Steenkamp et al (2005) find that retailers often show passive behaviour and that there are barely long-term consequences of these competitive responses. This is consistent with earlier research from Steenkamp, Nijs, Hanssens and Dekimpe (2001). Besides passive behaviour, accommodating behaviour and strong retaliation are options for firms to react on competitors (Steenkamp et al, 2005). “The consequences of a competitive action for an attacker depend at least in part on the number of responses that action provokes (Chen and Miller, 2008)”. When many firms in the market respond to an attack, the outcome for the attacker might be negative. On the other hand, if many competitive firms choose to show passive behaviour, the chance on positive effects will increase. Therefore, Chen and Miller (2008) argue that in order to be successful, firms should avoid retaliation of competitors.

(8)

6

their impact is relatively easily determined, and they tend to be more provocative” (Chen and MacMillan, 1992). Also, Gijsbrechts (1993) emphasizes the importance of pricing and an appropriate pricing strategy. Therefore, it necessary that factors influencing competitive reactions to price changes are known and understood. However, the importance of competitive reactions and pricing can be derived from previous research, not much empirical research is present about competitive reactions to changes of prices in the food retail market. Therefore, this research will try to fill this gap in literature. The leading research questions of this study will be: To which extend do retailers follow each other’s price changes, do reaction patterns differ between price increases and decreases and which other variables might explain reaction patterns.

This study will use the database of Supperscanner.nl. Superscanner.nl is a company that monitors prices of Dutch retailers that sell their products online. This means that Lidl, the largest player after Albert Heijn and Jumbo, is lacking in this database. Since only Dutch retailers are included in the database, this study will be limited to the Dutch food retail market. In the data section of this study, more information about the dataset will be provided.

(9)

7

2. LITERATURE REVIEW AND HYPOTHESES DEVELOPMENT

This section will provide a brief overview of the current existing retailing literature on competitive reactions of retailers. The goal of this study is to identify the variables explaining why retailers respond to price changes of competitors, and equally important, why customers do not respond to price changes. Based on this literature, hypotheses are developed. These hypotheses will be the basis of the conceptual map presented in the next section, which is the backbone of this thesis.

2.1.1 Competition –Size of retailers

One of the most important factors influencing pricing is competition (Abratt and Pitt, 1985). Large and powerful retailers have larger marketing budgets, wider

distribution (Reibstein and Farris 1995), and therefore price changes are noticed sooner compared to smaller retailers. Since awareness of a price change is a prerequisite for a competitive reaction, it is expected that competitive reactions to price changes of larger retailers will be occur more frequently. Besides, when a competitive price change is issued by a large retailer, this will be perceived as more alarming to other retailers. Therefore, consistent with the research of Steenkamp et al (2005), it is expected that the competitive reaction of retailers will be stronger when the price changes is issued by a large retailer. For the Dutch food retail market, this might imply that, for example, a price change initiated by Albert Heijn triggers more competitive responses compared to a price change of COOP.

However, the above-mentioned expectation will most likely hold in many cases, the size of the retaliating firm might influences the competitive response as well. Firms with different sizes will vary in their competitive reactions when under attack (Chen and Hambrick, 1995). First of all, a large firm has in general more resources which gives larger firms the ability to respond (Smith et al., 1991). Since smaller firms have less resources, they often cannot respond to competitive actions (Chen and Hambrick, 1995). Second, smaller firms often do not have the motivation to react since they want to avoid a new overwhelming retaliation (Steenkamp et al., 2005). Therefore, in general, larger or retailers of the same size are more likely to respond to price changes compared to

smaller retailers.

(10)

8

decreases. When Albert Heijn reduced their prices for many of their products in a wide variety of categories, all major competitors were able to respond within two days. Almost all priced that where lowered where matched or even further reduced (van Heerde, Gijsbrechts and Pauwels, 2008). Second, it shows that many retailers are willing to lower their prices, even when this means negative retail margins (Van Aalst, Sloot, van der Blom and Kivits, 2005; Holla and Koreman 2006).

2.1.2 Competition – Positioning of retailers

Firms operating in the Dutch food retail market are relatively similar to each other. Roughly 80% of the supermarkets are full service supermarkets in downtown and residential areas with a floor space between 500 and 1500 square meters (Pinckaers,

2016). The main differences between supermarkets can be found in differences between service and price. Data from GFK (2014) shows the perceived price and service level of the Dutch food retailers.Five types of supermarkets can be derived from this data and are shown in the graphic below. Different types of supermarkets have different

strategies and therefore it is expected that different types of supermarkets respond differently to price changes of competitors. Since the full-service supermarkets cover roughly 80% of the market, these will be the main focus of this thesis. The database used in this thesis only provides sufficient complete data for the three main full-service

supermarkets in the Netherlands. These chains are Albert Heijn (35,3%), Jumbo (18,4%) and Plus (6,2%). Therefore, only these three supermarkets are covered in this thesis. Overall, literature argues that competitive reactions differ per firm, reasoned by multiple arguments. Therefore, the following hypotheses are formulated:

H1A: The strength of the competitive reaction is positively related to the market share of

the initiating retailer.

H1B: The reaction time of the competitive reaction decrease when the market share of the

(11)

9 2.1.3 Competition - Follow down, not up(FDNU)

In an oligopolistic setting, as the Dutch food retail market, the performance of a retailer depends strongly on reactions of competitors. Although markets are

concentrated, competition is fierce. In the Dutch food retail market, it is common for retailers to use the same instrument to react on competitive actions (Steenkamp et al., 2005; Leeflang and Wittink, 1991). Therefore, a price change of a retailer is, in general, countered by a price change of their competitors. However, retailers tend to follow price decreases of competitors, literature tells us that reactions to price increases are rare. One of the first theories explaining this phenomenon is the kinked demand curve theory (Hall and Hitch, 1938). “This conjecture produces an “imagined” demand curve with a kink at the current price, bounded by a less elastic portion for lower prices and a more elastic portion for higher prices” (Dickson and Urbany, 1994). This implies that

(12)

10

competitors, competitive information and stage of the product life cycle. Based on the FDNU theory, the following hypotheses are developed:

H2A: A retailer will respond stronger to price decreases compared to price increases. H2B: A retailer will respond quicker to price decreases compared to price increases.

2.1.4 Competition – Reaction of other competitors

When a retailer shows passive behaviour towards a change in price of a retailer, this response might be changed due to other competitors following the price change (Dickson and Urbany, 1994). When a price cut is initiated by a firm and followed by competitors, the pressure to cut prices for remaining competitor increases (Dickson and Urbany 1991; Desarbo et al, 1987). The opposite scenario might also occur. When a firm increases their price and is followed by their competitors, opposite to the FDNU theory, a firm might find it is justified to increase their prices as well (Kahneman, Knetsch, and Thaler 1986). Since only the three largest full-service supermarkets are covered in this thesis, the reaction of other competitors cannot be included in the study in a proper way. However, it is important to be aware of this variable.

2.2.1 Product – Role and category

Willart (2015) comes up with several factors that influence the results of retail pricing research. One of these factors is the category of the product. Two different meanings for the category of a product can be understood, which can both influence the competitive reaction of retailers. The category of a product may refer to the role a product has. From now on, this will be called role of a product to prevent confusion. The other meaning for category of a product refers to literally which category the product falls into, e.g. fruits, vegetables, meat, etc. The following hypotheses are conducted to explore the effect of the category on competitive reactions.

(13)

11

In the research of Dhar et al (2001), four different roles for products are classified. The roles are based on high/low penetration and high/low frequency. The four different roles classified are therefore (1) staples (high penetration/high

frequency), (2) niches (low penetration/ high frequency), (3) variety enhancers(high penetration/low frequency) and (4) fill-ins(low penetration/low frequency). The different roles of these products require different strategies. For example, products in the fill-ins role are less attractive for promotions. With frequently discounting these products, retailers will give away margins without much returns. On the other hand, it might be attractive to offer low prices for products in the staples role. These products have a high price sensitivity and therefore retailers want to offer low prices for these products in order to have a positive price image (Dhar et al, 2001). However, the downside of offering low prices for these products is that much revenues might be missed. Even the smallest decrease in price might cause an enormous decrease in revenues. This makes pricing one of the most important and complex issues retailers have to deal with. Since different strategies are used by retailers for different roles, it is essential to investigate products with different roles when conducting research in the area of retail pricing. Since frequently sold products are important for the price image, the following hypotheses are formed:

H4A: A shorter interpurchase time of the category increases the strength of a competitive

reaction.

H4B: A shorter interpurchase time of the category decreases the reaction time of

competitive reactions.

2.2.2 Product – Hedonic vs. utilitarian

(14)

12

utilitarian products are certainly not always mutually exclusive. As illustration,

toothpaste has hedonic and utilitarian characteristics. It preserves the teeth of the user as well as it provides pleasure from the taste.

Luxury products have, in general, a higher price elasticity of demand compared to necessities (Deaton and Muellbauer, 1980). Besides, customers find it harder to justify hedonic purchases compared to utilitarian purchases (Kivetz and Zheng, 2017).

Moreover, “research shows that people feel more guilt when they contemplate on engaging in hedonic consumption than engaging in utilitarian consumption (Lu, Liu and Fang, 2016)”. Therefore, for example, price promotions are more effective for hedonic products since it gives customers a justification for buying the product. Since customers are more price sensitive for hedonic products, sales will have a larger drop compared to utilitarian products when the price increases. Besides, utilitarian products have a larger influence on the price image of a retailer. Therefore, it is expected that competitive reactions are stronger and quicker for hedonic products compared to utilitarian products. This results in the following hypothesis.

H5A: A price change will trigger a stronger reaction of retailers when the product is of

hedonic nature.

H5B: A price change will trigger a quicker reaction of retailers when the product is of

hedonic nature.

H6A: A price change will trigger a weaker reaction of retailers when the product is of

utilitarian nature.

H6B: A price change will trigger a less quick reaction of retailers when the product is of

utilitarian nature.

2.2.3 Product – Private label vs. National Brand

(15)

13

brands need to cooperate with retailers and are therefore restricted and less motivated to compete with private label products (Steenkamp and Dekrimpe, 1997). Therefore, it is expected that competitive reactions will be stronger for national brands compared to private label products.

H7A: A retailer will respond stronger to price changes of national brands compared to

private label products.

H7B: A retailer will respond quicker to price changes of national brands compared to

private label products.

2.3 Local differences

Another factor found by Willart (2015) that is influencing retail pricing research are the characteristics of the retail trade area. As mentioned before, competition in the area can influence the pricing strategy of retailers. Recently, the Consumentenbond (2018) revealed that Dutch supermarket chains, Jumbo and Hoogvliet, are using

different prices for the same products in different stores. They adapt their prices to their competitors active in the area. When a low-priced competitor is active in the area, they decrease their prices. When only higher priced competitors are active, they will handle higher prices. Moreover, retailers might adapt their prices in order to fit local demand for different price levels (Montgomery, 1997). However, similar to Willart (2015), in this thesis it will be assumed that demographics are already embedded in the competitor variables since the demographics food retailers are facing are similar.

2.4 Size of price change

(16)

14

H8A: The size of the price change (%) will have an effect on the strength of the competitive

reaction.

H8B: The size of the price change (%) will have an effect on the reaction time of a retailer.

2.5 Price premium

Not only the size and sign of a price change are variables related to price that might influence a competitive reaction. When a product is expensive, in this thesis measured relative to the product category, competitive reactions of retailers may be different. Not much research has been done on this specific variable, however, it is known that customers respond different to high and low-priced products (Grewal, Krishnan, Baker and Borin, 1998). Since customers show different buying behaviour, it can be argued that retailers have different strategies for these products. Since low priced products have more influence on the price image (Dhar et al, 2001), it is expected that competitive reactions to price changes in low priced product will be stronger.

Therefore, the following hypothesis are formed.

H9A: The strength of a competitive reaction will be more powerful for low-priced products compared to high-priced products.

H9B: The reaction time to price changes will be shorter for low-priced products compared to high-priced products.

2.6 Strength and speed of the competitive reaction

Until now, many variables are mentioned and described that might influence the competitive reaction, measured in terms of speed and strength. However, the speed and strength of a competitive reaction can also be an explanatory variable for the other. When a retailer responds quickly, it is important to offer a competitive price and

therefore the strength of the competitive reaction will increase. On the opposite, when a competitive reaction is strong, it is likely that the reaction time will be shorter.

Therefore, the following hypotheses are presented.

H10A: The speed of the competitive reaction is positively related to the strength of the

competitive reaction.

H10B: The strength of the competitive reaction will decrease the reaction time of a

(17)

15

3. STUDY - MEASURING COMPETITIVE REACTIONS

3.1 Conceptual Model

Based on the hypothesis developed in the literature section of this thesis, the following conceptual model is proposed in figure 1. This model is a variant on the model used by Steenkamp et al (2005). For clarity of the conceptual model, the variables are clustered in category variables, product variables and case specific variables. Besides all the variables developed in the previous section, the variable total price changes is added to the conceptual model. This variable is added in order to investigate the relationship between the total number of price changes on a certain day on the dependent variables.

(18)

16 3.2 Method

3.2.1 Data

The daily price data used in this study is from the database of Superscanner.nl. This database contains prices of all products offered online by the largest retailers in the Netherlands. The total amount of products in this database is roughly 130.000. Since Superscanner.nl is a relatively young company, the data covers a short period of time. Besides, not all retailers were included from the beginning. Therefore, some important decisions were made when determining which data to use. After the preliminary

analysis of the data, the decision was made to only include the three largest full-service supermarkets. These supermarkets are Albert Heijn, Jumbo and Plus. These retailers are the main players in this market and data of a relatively long period is available. Besides, adding many chains would complicate the analysis extremely without an enormous increase in accuracy. The selected data covers the period from the 1th of June 2016 until the 1th of May 2018 and is aggregated on a daily level. In total, 99 products were

selected from 10 varying categories. In the 10 main categories, 8 nationals brand products and 2 private label products are selected. Due to unavailability of sufficient matching private label products in the laundry detergent category, only 1 private label product is selected in this category. In section 3.2.3, more detailed information about the variables can be found.

(19)

17 3.2.2 Measurement of variables and equations

In table 1, an overview is presented of all dependent and independent variables and how they are measured.

Table 1: Measurement of variables

Variable Measurement

Strength of competitive reaction(STCR) Measured in % of the initiating price change

Speed of competitive reaction (SPCR) Average response time to a competitive action for a product (measured in days)

Interpurchase time (IT) Gathered from own study. Likert scale (1-7) Hedonic characteristics (HC) Gathered from own study. Likert scale (1-7) Utilitarian characteristics (UC) Gathered from own study. Likert scale (1-7) Brand type (BT) National brand vs. private label. Gathered from the

Superscanner.nl database.

Price premium (PP) The difference in % from the average price of the category. Category (CAT) Categorical variable, 10 categories: Alcoholic beverages,

Preserves, Frozen food, Soft drinks, Biscuits, Sweets, Coffee, Pasta, Sauce and Detergent.

Initiating retailer (IR) Three retailers: Albert Heijn, Jumbo and Plus

Sign initial price change (Sign) Dummy variable, 1 in case of a price increase, 0 in case of a price decrease.

Size of initial price change (Size) The % change of the initial price.

Total price changes (TC) Total price changes of all retailers measured on daily level.

Since it is expected that the three retailers respond different to price changes, the models used are unit-by-unit models (Leeflang, Bijmolt, Pauwels and Wieringa, 2015). Benefits of a unit-by-unit model are that the data and so the estimates are per chain. Downside of this type of model is that there are fewer observations per parameter. The conceptual model presented in section 3.1 results in the following equations, where the variables are as defined in table 1 and table 2.

(20)

18 3.2.3 Descriptive analysis

Before the estimation of the models, presented in the section above, most important descriptive statistics about the data are examined. This will increase the understanding of the behaviour of the retailers and explore in which categories

extensive or little competition takes place. First, an overview of all price changes of the three retailers is presented in table 2. The most remarkable observation is that Albert Heijn has the least price changes, however, it is often the initiating retailer. Besides, Albert Heijn does not react very often to Jumbo and Plus compared to the behaviour of Jumbo and Plus. In total, there are more price decreases compared to price increases.

Table 2: Price changes and competitive reactions Retailer Total price changes Initiating

retailer Competitive reaction % reaction to competitors Increases Decreases Yes No

AH 191 (48,4%) 204 (51,6%) 320 (49,2%) 69 262 20,85% Jumbo 207 (37,6%) 344 (62,4%) 279 (42,9%) 214 160 57,22% Plus 183 (45,2%) 222 (54,8%) 51 (7,8%) 344 256 57,33% AH/Jumbo 1 0 1 - - - AH/Plus 1 2 3 - - - Jumbo/Plus 0 1 1 - - - Total 583 770 655 627 678 48,05%

Table 3: Reaction speed and strength

Retailer Average strength reaction Average reaction time in days

National

brand Private label National brand Private label

AH 0,9 0,6 14,2 12,8

Jumbo 0,9 1,2 3,1 2,2

(21)

19

Table 3 contains information about the competitive reactions to changes in national brands and private label products. It is remarkable that Jumbo is the quickest responder, followed by Plus. Albert Heijn responds on average after two weeks. In the histogram “Speed of competitive reaction AH”,

the reaction time of Albert Heijn is presented in days. It can be observed that the response time of Albert Heijn varies. In some cases, Albert Heijn responds quickly, however, a

reaction time of more than 18-20 days occurs the most.

The histogram “Strength of

competitive reaction AH”, shows how the variable strength of competitive reaction is distributed. The most common reaction strength is 1, or a slightly below 1. It is also remarkable that there are some cases that Albert Heijn decides to respond with an opposite reaction (e.g. increase price as reaction to price decrease).

The histogram “Speed of competitive

reaction Jumbo” shows the reaction time of

(22)

20

The strength of the competitive reactions of Jumbo can be observed in the histogram “Strength of competitive reaction Jumbo”. Similar to Albert Heijn, the strength of competitive reactions of Jumbo are in most cases 1 or slightly below one. A remarkable observation is that there are several extreme cases in which Jumbo responds with 4 times the price change and -2.

The speed of competitive reactions of Plus are visible in the histogram “Speed of competitive

reaction Plus”. The distribution is similar to the

speed of competitive reactions of Jumbo, however, there are more cases in which it responds later.

The strength of the competitive reactions of Plus can be observed in the histogram “Strength of competitive reaction

Plus”. This shows that in almost all cases, Plus

reacts with the same strength, which can also be observed in the fact that the mean, 1st quartile and 3th quartile is 1.

Now the average strength and speed of competitive reactions of the different retailers are described, the focus will shift to how the strength and speed differ per initiating retailer. Table 4 shows how quick and strong Albert Heijn responds to Jumbo and plus. Table 5 and 6 are identical, however they represent Jumbo and Plus

(23)

21

react very often to Plus (23,5%). Therefore it is possible that Jumbo only reacts to Plus when it feels that is has to react and therefore responds more powerful. Moreover, Albert Heijn does barely react to Plus (9,8%). Besides the strength of the reaction to Plus is low (73,3%)

Table 4: Reactions AH to Plus and Jumbo

Initiating retailer Number of reactions (%) Speed of reactions in Days Strength of reaction Jumbo 63 (22,5%) 13,5 89,4% Plus 5 (9,8%) 8,6 73,3%

Table 5: Reaction Jumbo to AH and Plus

Initiating retailer Number of reactions (%) Speed of reactions in Days Strength of reaction AH 199 (62,0%) 2,38 87,9% Plus 12 (23,5%) 9,1 110%

Table 6: Reactions Plus to AH and Jumbo

Initiating retailer Number of reactions (%) Speed of reactions in Days Strength of reaction AH 273 (85,1%) 3,1 101,2% Jumbo 70 (25,0%) 14,9 93,9%

In total, 19 of the 99 products (19,2%) in this dataset are private label products. When observing all price changes in the dataset, 177 of the 1258 price changes are price changes in prices of private label products (14,1%). However, there are insufficient products to prove a statistical difference between private label products and price changes, it can be observed that price changes in private label products are less common. This would also be in line with the research of Steenkamp et al. (2005). Moreover, table 3 shows that response time to price changes is lower for private label products compared to national brands for all retailers. The average strength of

competitive reactions is close to one for national brands. For private label products, the strength of the response differs per retailer. Observing price changes in different

(24)

22 3.3 Logit/probit regressions

The preliminary analysis of the data showed a chance of approximately 48% that a retailer responds to a price change of a competitor. However, there is a huge

difference between Albert Heijn and Plus and Jumbo. This section will shed some light on the underlying variables influencing the decision to respond to a competitive reaction, or to show passive behaviour. This analysis is performed with a logistic regression. This regression will use of the same model as used for measuring the

strength of the competitive reaction and speed of the competitive reaction. The equation for this model is presented below, where CR means competitive reaction (1=Yes, 0=No). The explanation of other variables can be found in table 1.

𝐶𝑅𝑖 = 𝛽0𝑖+ 𝛽1𝑖𝐼𝑇 + 𝛽2𝑖𝐻𝐶 + 𝛽3𝑖𝑈𝐶 + 𝛽4𝑖𝐵𝑇 + 𝛽5𝑖𝑃𝑃 + 𝛽6𝑖𝐶𝐴𝑇 + 𝛽7𝑖𝐼𝑅 + 𝛽8𝑖𝑆𝐼𝐺𝑁 + 𝛽9𝑖𝑆𝐼𝑍𝐸 + 𝛽10𝑖𝑆𝑃𝐶𝑅𝑖+ 𝛽11𝑖𝑇𝐶𝑖+ 𝜀𝑖

First, two types of logistic regressions are estimated, a logit and a probit logistic regression. Logit and probit models are very similar and therefore output of these models are not so different. After estimating the estimations, the models are checked on multicollinearity with the VIF-test. All regressions show an extremely high VIF-score for the variable category. Moreover, the variables hedonic characteristics and utilitarian characteristics show a VIF-score above 10. This problem is solved with removing the variable category from the model. Calculating the new VIF-scores only show scores below 4, which means that the issue of multicollinearity is solved.

In order to determine which models fit best, multiple measures of model fit will be used to compare the models. These measures are the pseudo R2_{, AIC, BIC, hitrate and}

the Likelihood ratio test. An overview of these measures can be found in table 7. The measures show, as expected, that the performance of the models does not differ significantly for a specific retailer, however the logit models perform slightly better. Since there are no significant differences between the logit and probit models, output of logit models will be used. Coefficients of probit models relate to changes in the Z-scores, which are less useful. Next, the logit models are compared with the null model. This comparison shows that the logit models of the three retailers all outperform the null models and proves that the models have explanatory power. The logit model has the most explanatory power for Plus, since it has the highest R2 _{and hitrate. The output of}

(25)

23 Table 7: Model fit comparison

Model Pseudo R2 AIC BIC LR Hitrate

AH - Logit 0.06 337.3 379.2 -157.7 0.58 AH - Probit 0.06 337.6 379.5 -157.8 0.58 Jumbo - Logit 0.09 488.2 531.4 -233.1 0.66 Jumbo - Probit 0.09 488.3 531.5 -233.1 0.66 Plus- Logit 0.33 573.8 622.2 -275.8 0.81 Plus- Probit 0.33 573.1 621.5 -275.6 0.80 3.3.1 Competitive reaction AH

The logit regression with the binary variable competitive reaction AH as

dependent variable has three significant variables. The first significant variable is brand type. This implies that a competitive reaction of Albert Heijn is more likely when the products is a national brand (β = 0.987; p <0,05). The second variable is initiating retailer Plus. When Plus is initiating the first price change, it is less likely that Albert Heijn responds (β = -0.901; p <0,05). The last significant variable is total price changes. When there are more price changes on a specific day, it is more likely that Albert Heijn responds to a price change of a competitor (β =0.079 ; p <0,1). To determine the

(26)

24 Table 8: Output logit regressions

Dependent variable: Independent variable:

Competitve reaction AH

Par. Est. (St. Err.)

Competitve reaction Jumbo

Competitve reaction Plus

Interpurchase time -0.298 (0.224) -0.322* _(0.181) _{-0.272 (0.172)} Hedonic characteristics 0.271 (0.235) 0.352* _(0.203) _{0.213 (0.183)} Utilitarian characteristics 0.136 (0.209) 0.041 (0.178) 0.071 (0.165) Brand type 1 (0-1, 1 = national brand) 0.987** _(0.478) _{0.064 (0.343)} _0.538* _(0.321) Price premium 0.421 (0.333) 0.128 (0.283) 0.425 (0.260) Initiating retailerJumbo/Plus -12.988 (882.744) Initiating retailerAH/Plus 14.907 (827.941) Initiating retailerPlus -0.901* _(0.519) _-1.577*** _(0.371) Initiating retailerAH/Jumbo 12.430 (535.411) Initiating retailerJumbo -2.939*** _(0.242)

Sign initial price change 1 (0-1, 1 =

price increase) -0.133 (0.427) -0.052 (0.361) -0.979*** (0.334) Strength first price change (%) 0.857 (2.359) 4.043 (2.730) 1.473 (2.035) Total price changes today 0.079* _(0.042) _{0.016 (0.032)} _0.092*** _(0.033) Constant -3.682** _(1.597) _{-0.617 (1.220)} _{0.405 (1.210)}

Observations 332 375 602

Log Likelihood -157.658 -233.079 -275.896

Akaike Inf. Crit. 337.316 488.159 573.792

Note: *_p<0.1;**_p<0.05;***_p<0.01

3.3.2 Competitive reaction Jumbo

(27)

25 3.3.3 Competitive reaction Plus

Performing the same regression for Plus leads to interesting results. First of all, the variable initiating retailer Jumbo has a negative estimate and is significant (β -2.939; p <0,01) The odds ratio (0,05), shows that Plus is 20 times more likely to respond to Albert Heijn compared to Jumbo. Furthermore, in comprehension with the FDNU theory (Dickson and Urbany, 1994), Plus responds significantly less to price increases

compared to price decreases (β -0.979; p <0,01) More precisely, the odds ratio (0,4) tells us that the chance of responding to a price increase is 2,5 times smaller compared to responding to price decreases. Similar to the regression of Albert Heijn, price changes of national brands are more likely to be followed (β 0.538; p <0,1). The related odds ratio (1,7) estimate that changes in national brands are more likely to be followed.

Furthermore, in accordance with Albert Heijn, total price changes is significant (β = 0.092; p <0,01). The related odds ratio (1,1) implies that the chance Plus reacts increases with 1,1 times when there is 1 more increase in price changes on the day of the initial price change.

3.4 Empirical results

Now it is known when and to who the three retailers respond and what influences this decision, empirical light can be shed on the main subject of this thesis. For each retailer, the model in section 3.2.2 will be estimated. The models are estimated with a multiple linear regression. The initial plan of analysis included an ordered logit regression, however, most variables in the model violated the parallel lines assumption. Since remedies of the parallel lines assumption are not desired, the decision was made to solely use a multiple linear regression. In order to find reliable results, the

assumptions of a linear model should be satisfied. Therefore, some issues are solved before the estimation of the models. First, to solve the issue of multicollinearity, showed by extremely high VIF-scores, the variable category was left out of the regression. This results in all VIF-scores being below 5.

(28)

26

order to test if heteroskedasticity is present, the Breush-Pagan Test was performed for the different models. This test showed that heteroscedasticity is present for the model with speed of competitive reaction of Jumbo as dependent variable and the models of Plus. A Box-Cox transformation was applied to obtain homoscedasticity for the model of Jumbo. The Box-Cox transformation was not sufficient to some the issue of

heteroskedasticity for the models of Plus. Since the dataset contains of specific cases instead of time series data, autocorrelation of residuals is not a problem. Furthermore, the number of observations is for all models larger than the number of independent variables, which is required for a linear regression. Lastly, the mean of residuals is for all models close to zero.

3.4.1 Albert Heijn

The output of the model with speed of the competitive reaction of Albert Heijn as dependent variable is presented in table 9. Four models are estimated, one with only category variables, one with product variables, one with case specific variables, and an estimation with all variables. The non-full models are estimated to measure the effect of the type of variables on the dependent variables. Since the full model is more complete, these parameters will be interpreted. The full model is based on 67 observations and has a R2_{of 0,365 and is significant (P < 0,01). Since the dependent variable is never 0, the}

intercept has no intrinsic meaning. Interaction effects where investigated, however, this did not result in any significant effects, or, extremely high VIF-scores.

(29)

27

opposite to H10B, it takes longer for Albert Heijn to respond. Other variables in the model are not significant and can therefore not be interpreted. Considering the R2 _and

significant variables of the models, it can be concluded that case specific variables have the largest impact on speed of the competitive reaction.

Table 9: Output Albert Heijn

Dependent variable:

Speed of competitive reaction AH

Category Product Case Full

Independent variable Par. Est. (St. Err.) Par. Est. (St. Err.) Par. Est. (St. Err.) Par. Est. (St. Err.)

Interpurchase time -0.331 (1.488) -0.770 (1.377) Hedonic characteristics -1.208 (1.611) -2.791* _(1.504) Utilitarian characteristics -0.440 (1.455) -0.078 (1.360) Brand type 1 (0-1, 1 = national brand) 1.118 (3.892) -4.976 (3.694) Price premium -1.204 (1.885) -0.654 (1.886)

Sign initial price change 1

(0-1, 1 = price increase) -4.625 (3.555) -7.828** (3.711) Strength first price change

(%) 10.404 (21.813) 25.283 (22.578)

Strength of competitive

reaction AH (%) 4.508** (1.965) 6.409*** (2.043)

Initiating retailer Jumbo 3.784 (3.691) 10.430** _(4.084)

Total price changes today 0.371 (0.244) 0.274 (0.252)

Constant 21.291** _{(10.551) 13.538}*** _(3.770) _{5.540 (3.961)} _20.087* _(11.507) Observations 67 67 67 67 R2 _0.020 _0.007 _0.217 _0.365 Adjusted R2 _-0.026 _-0.025 _0.153 _0.251 Residual Std. Error 8.031 8.024 7.297 6.859 F Statistic 0.437 0.210 3.381*** _3.216*** Note: *_p<0.1;**_p<0.05;***_p<0.01

The model with the strength of the competitive reaction of Albert Heijn as dependent variable is presented in table 10. Again, the model is based on 67

observations, has an R2_{of 0,379, and is significant (P < 0,01). Interaction effect have}

(30)

28

The full-model reveals two significant variables. In accordance the first model of Albert Heijn, speed of the competitive reaction of AH is significant (β = 0.023; p <0,01). When the competitive reaction takes longer, the competitive reaction will on average be more powerful. This is contradicting H10A, however it might be explained by the

reasoning that retailers might need more time to make a decisions about larger price changes. Furthermore, contradicting H9A, responses to high priced products are stronger compared to low priced products (β = 0.209; p <0,1).

Table 10: Output Albert Heijn

Strength of competitive reaction AH

Interpurchase time 0.126 (0.086) 0.054 (0.083) Hedonic characteristics -0.019 (0.093) 0.015 (0.093) Utilitarian characteristics -0.136 (0.083) -0.133 (0.080) Brand type 1 (0-1, 1 = national brand) 0.140 (0.206) 0.171 (0.225) Price premium 0.255** _(0.108) _0.209* _(0.110)

(0-1, 1 = price increase) -0.385* (0.220) -0.197 (0.231) Strength first price change

(%) 2.354* (1.332) 1.512 (1.362) Speed of competitive

reaction AH 0.018** (0.008) 0.023*** (0.007)

Initiating retailer Jumbo 0.184 (0.232) -0.214 (0.259)

(31)

29 3.4.2 Jumbo

For Jumbo, the same models are estimated as for Albert Heijn. The parameter estimates, standard error and the significance level of the first model are presented in table 11. The estimations are based on 214 observations and the full model has a R2 _of

0,232. Compared to Albert Heijn, the R2_{is lower, however it is acceptable. Moreover, the}

full model is significant with a P-value of 0,01. Observing the R2 _{of the non-full models,}

case specific variables have the largest influence.

Table 11: Output Jumbo

Speed of competitive reaction Jumbo

Interpurchase time -0.031 (0.039) -0.025 (0.036) Hedonic characteristics -0.009 (0.043) 0.018 (0.041) Utilitarian characteristics 0.034 (0.038) 0.068* _(0.035) Brand type 1 (0-1, 1 = national brand) 0.087 (0.076) 0.177** (0.073) Price premium 0.067 (0.058) 0.041 (0.053)

(0-1, 1 = price increase) -0.329*** (0.076) -0.340*** (0.074) Strength first price change

(%) 1.265* (0.075) 1.203** (0.606) Strength of competitive

reaction Jumbo (%) -0.033 (0.031) -0.024 (0.032)

Initiating retailer AH -0.412*** _{(0.097) -0.433}*** _(0.098)

(32)

30

The model for speed of the competitive reaction of Jumbo provides interesting insights. First of all, in line with H6B, utilitarian characteristics of a product delay the competitive reaction of Jumbo (β =0.068; p < ,05). Moreover, H1B is supported since the reaction of Jumbo is quicker to Albert Heijn compared to the baseline Plus (β =-0.433; p < ,01). Another significant variable is the sign of the initial price change (β = -0.340; p < ,01). This means that, opposite to the FDNU theory and H2B, but similar to Albert Heijn, Jumbo responds quicker to a price increase, compared with a price decrease. The next significant variable for Jumbo is the strength first price change (β = 1.203; p < ,05). This implies that the reaction time of Jumbo increases when the initial price change is larger. The last significant variable is brand type (β = 0.177; p < ,05). This means that opposite to H7B, Jumbo responds quicker to changes in private label products compared to national brands.

The output of the second model of Jumbo can be observed in table 12. The R2_of

this model is low compared to the other models. Besides, the model has a low

significance level (P < 0,1). Considering the R2_{and the significance levels of the non-full}

models, case specific and product variables have the largest influence.

(33)

31 Table 12: Output Jumbo

Strength of competitive reaction Jumbo

Interpurchase time 0.061 (0.079) 0.053 (0.080) Hedonic characteristics -0.135 (0.087) -0.131 (0.093) Utilitarian characteristics -0.035 (0.077) -0.057 (0.079) Brand type 1 (0-1, 1 = national brand) -0.265* (0.152) -0.250 (0.161) Price premium -0.101 (0.116) -0.121 (0.118)

(0-1, 1 = price increase) -0.018 (0.166) 0.075 (0.170)

Strength first price change

(%) 1.679 (1.331) 4.145** (1.742)

Speed of competitive

reaction Jumbo -0.003 (0.016) 0.004 (0.016)

Initiating retailer AH -0.220 (0.236) -0.213 (0.248)

Sign initial price change1*Strength first price change -6.505 ** _(2.855) Constant 1.486*** _(0.531) _1.235*** _(0.164) _0.995*** _(0.274) _2.139*** _(0.605) Observations 214 214 214 214 R2 _0.019 _0.023 _0.032 _0.084 Adjusted R2 _0.005 _0.013 _0.009 _0.035 Residual Std. Error 0.706 0.703 0.704 0.695 F Statistic 1.333 2.457* _1.371 _1.692* Note: *_p<0.1;**_p<0.05;***_p<0.01 3.4.3 Plus

The output of the first model of Plus is presented in Table 13. The estimation is based on 345 observations and has a R2 _{of 0.412. Compared to the models of Albert}

Heijn and Jumbo, the model has the highest R2_{. Besides, the model is significant (P<0.01)}

and in combination with the high R2_{, the results of this model are useful. Moreover, from}

the R2 _{of the non-full models, it can be derived that case variables have the largest}

(34)

32 Table 13: Output Plus

Speed of competitive reaction Plus

Interpurchase time 0.628 (0.637) 0.082 (0.506) Hedonic characteristics -0.914 (0.702) -0.191 (0.559) Utilitarian characteristics -1.005 (0.638) -0.226 (0.503) Brand type 1 (0-1, 1 = national brand) 1.586 (1.354) 1.490 (1.092) Price premium -0.265 (0.948) -0.993(0.745)

(0-1, 1 = price increase) -1.030 (1.049) -0.978(1.056)

(%) 1.344* (0.743) 14.473* (8.201) Strength of competitive

reaction plus (%) -0.055 (0.302) 1.314* (0.747)

Initiating retailer AH -12.104*** _{(0.812) -12.102}***_(0.820)

Constant 11.655*** _(4.410) _4.412*** _(1.375) _14.050*** _{(1.162) 15.332}*** _(3.623) Observations 345 345 345 345 R2 _0.007 _0.004 _0.406 _0.412 Adjusted R2 _-0.001 _-0.002 _0.398 _0.394 Residual Std. Error 7.568 7.569 5.869 5.886 F Statistic 0.834 0.690 46.437*** _23.386*** Note: *_p<0.1;**_p<0.05;***_p<0.01

(35)

33

Table 14 contains the output of the last model of Plus. The full models has the lowest R2 _{of all six models. Besides, the model is not significant and therefore the results}

should be interpreted with caution. Examining significant variables and the R2 _{of the}

non-full models, it can be concluded that the case specific variables have the largest impact on the strength of the competitive reaction of Plus.

Table 14: Output Plus

Strength of competitive reaction Plus

Interpurchase time 0.013 (0.036) 0.033 (0.037) Hedonic characteristics 0.011 (0.040) -0.011 (0.041) Utilitarian characteristics -0.013 (0.036) -0.022 (0.037) Brand type 1 (0-1, 1 = national brand) 0.076 (0.077) 0.035 (0.080) Price premium 0.003 (0.054) -0.008 (0.056)

(0-1, 1 = price increase) -0.005 (0.076) 0.057 (0.082)

(%) 0.480 (0.589) 1.214* (0.735)

Speed of competitive

reaction Plus 0.007* (0.004) 0.008* (0.004)

Initiating retailer AH 0.168** (0.075) 0.155** (0.076)

Total price changes today -0.003 (0.006) -0.002 (0.006) Sign initial price

(36)

34

(37)

35

4. DISCUSSION & CONCLUSION

This paper focused on the underlying variables that are influencing the decision to react to price changes of competitors, the reaction speed of this competitive reaction and the strength of this reaction. The aim of this study is to develop a better

understanding of how food retailers in the Dutch market respond to their competitors. First, a model was developed to understand which variables are influencing the decision of retailers to respond to a price change of a competitor. However, the results are not identical, the initiating retailer is a very important factor for the decision to respond to a price change of a competitor. More specifically, the results show that the retailers in the market follow price changes of large competitors more frequent compared to price changes of smaller retailers. This is showed by that Albert Heijn follows Jumbo most frequent, and Jumbo and Plus follow Albert Heijn most frequent. Moreover, Albert Heijn and Plus follow price changes of national brands more often compared to private label products. Also the total amount of price changes on the day the price change is issued increases the chance of a competitive reaction of Albert Heijn and Plus. For Jumbo, category related variables are influential. A high interpurchase time of the category has a negative influence while hedonic characteristics of a category have a positive influence on the decision to respond to a price change of a competitor.

The models with speed of the competitive reactions of the three retailers as dependent variable showed different significant variables. Overall, case specific variables have the largest influence on the reaction time of retailers. The initiating retailer is significant for all three retailers. Albert Heijn has a lower reaction time to Plus compared to Jumbo, Jumbo has a lower reaction time to Albert Heijn compared to Plus, and Plus has a lower reaction time to Albert Heijn compared to Jumbo. Furthermore, a price increase is followed quicker by Albert Heijn and Jumbo compared to price

(38)

36

utilitarian characteristics and national brands while these variables do not have impact on Albert Heijn and Plus.

The models revealing variables that are influencing the strength of competitive reactions also show different results for the different retailers. In general, case specific variables have the largest influence, while category related variables have no significant effect at all on the strength of competitive reactions. Jumbo and Plus both respond more powerful to strong price changes, while this effect is not statistically proven for Albert Heijn. Moreover, Albert Heijn and Plus both react more powerful to price changes when their reaction time is larger. Furthermore, Plus responds stronger to Albert Heijn compared to Jumbo, while Albert Heijn responds more powerful to price changes in expensive products. Also interaction effects have been found for Jumbo and Plus. A strong initial price increase will decrease the strength of a competitive reaction. This can be explained by that retailers only want to follow price increases to a certain extend since they do not want to hurt their price image.

(39)

37

5. LIMITATIONS & FURTHER RESEARCH

This study has its limitations regarding the data used and the analysis. When looking at the problems with the analysis, some models suffer from non-normality and

heteroskedasticity. Besides the variable category had to be left out of the analysis due to the issue of multicollinearity, indicated by high VIF-scores. Therefore, the results should be interpreted with a certain degree of caution.

Besides issues with the analysis, there are some limitations to the dataset. The first limitation is that only information about three retailers are included in the dataset. Therefore, the possible responses to other competitors are not included in the analysis. Also, the time frame of the data is limited. Due to the unavailability of sufficient reliable data, only data from the period of the 1th of June 2016 – 1th of May 2018 is included. The results would definitely be more precise if the time range of the dataset would have been expanded. Besides, only retailers in the Netherlands are included in this thesis. Therefore, the results of this research are not immediately transferable to other geographical regions. However, it should be noted that competitive reactions depend heavily on the market. Therefore, it is extremely difficult to find results that would be applicable to many different regions and countries.

This thesis made use of 99 different products. However, 99 products with a total of 1353 price changes is a decent amount, compared with 130.000+ available products, it is a rather small selection. Besides, it was hard to find sufficient private label products that matched. This is caused by the fact that most private label products differ from each other and are not similar as national brands. Therefore only 19 private label products are included. Furthermore, it was not possible to add any basic products to the analysis, since Jumbo does not offer products in this product class.

(40)

38

REFERENCES

Abratt, R., & Pitt, L. F. (1985). Pricing practices in two industries. Industrial Marketing

Management, 14(4), 301–306.

Batra, R., & Ahtola, O. T. (1991). Measuring the Hedonic and Utilitarian Sources of Consumer Attitudes. Source: Marketing Letters Marketing Letters, 2(22), 159–170. Bijmolt, T. H., Heerde, H. J. van, & Pieters, R. G. (2005). New Empirical Generalizations on

the Determinants of Price Elasticity. Journal of Marketing Research, 42(2), 141–156. Chen, M.-J., & Hambrick, D. C. (1995). Speed, Stealth, and Selective Attack: How Small

Firms Differ from Large Firms in Competitive Behavior. Academy of Management

Journal, 38(2), 453–482.

Chen, M.-J., & MacMillan, I. C. (1992). Nonresponse and Delayed Response to Competitive Moves: The Roles of Competitor Dependence and Action Irreversibility. Academy of

Management Journal, 35(3), 539–570.

Chen, M. J. (1996). Competitor analysis and interfirm rivalry: Toward a theoretical integration. Academy of Management Review, 21(1), 100–134.

Chen, M., & Miller, D. (2008). Competitive attack, retaliation and performance: An expectancy-valence framework. Strategic Management Journal, 15(2), 85–102. Chen, M.-J., Smith, K. G., & Grimm, C. M. (1992). Action Characteristics as Predictors of

Competitive Responses. Management Science, 38(3), 439–455.

Cohen, D., & Crabtree, B. (2006). Semi-structured Interviews Recording Semi-Structured interviews. Qualitative Research Guidelines Project, 2.

Deaton, A., & Muellbauer, J. (1980). An almost ideal demand system. The American

Economic Review, 70(3), 312–326.

DeSarbo, W. S., Rao, V. R., Steckel, J. H., Wind, J., & Colombo, R. (1987). A Friction Model for Describing and Forecasting Price Changes. Marketing Science, 6(4), 299–319. Dhar, S. K., Hoch, S. J., & Kumar, N. (2001). Effective category management depends on the

role of the category. Journal of Retailing, 77(2), 165–184.

Dickson, P. R., & Urbany, J. E. (1994). Retailer reactions to competitive price changes.

Journal of Retailing, 70(1), 1–21.

Ellickson, P. B. (2013). Supermarkets as a natural oligopoly. Economic Inquiry, 51(2), 1142– 1154.

(41)

39

Gijsbrechts, E. (1993). Prices and pricing research in consumer marketing: Some recent developments. International Journal of Research in Marketing, 10(2), 115–151. Grewal, D., Krishnan, R., Baker, J., Borin, N., & Krishnan Is, R. (1998). The Effect of Store

Name, Brand Name and Price Discounts on Consumers’ Evaluations and Purchase Intentions. Journal of Retailing, 74(3), 331–352.

Hall, R. L., & Hitch, C. J. (1939). Price theory and business behaviour. Oxford Economic

Papers, (2), 12–45.

Holla, J, & Koreman. A. (2006), “Prijzenoorlog en de Positioneringslag van Albert Heijn”, GfK jaargids, GfK Panel Services Benelux, 126-131.

Kahneman, D., Knetsch, J., & Thaler, R. (1986). Fairness and the assumption of economics.

The Journal of Business, 59(4), 285–300.

Kivetz, R., & Zheng, Y. (2017). The effects of promotions on hedonic versus utilitarian purchases. Journal of Consumer Psychology, 27(1), 59–68.

Leeflang, P. S. H., & Wittink, D. R. (1992). Diagnosing competitive reactions using

(aggregated) scanner data. International Journal of Research in Marketing, 9(1), 39–57. Leeflang, P., Bijmolt, T., Pauwels, K., & Wieringa, J. (2015). Modeling Markets: Analyzing

Marketing Phenomena and Improving Marketing Decision Making. International Series

in Quantitative Marketing; Vol. 1. New York: Springer.

Nielsen (2018). Marktaandelen 2007-2017 Nielsen. Retrieved from http://www.distrifood.nl/service/marktaandelen

Montgomery, A. L. (1997). Creating Micro-Marketing Pricing Strategies Using Supermarket Scanner Data. Marketing Science., 16(4), 315.

Montgomery, D. B., Moore, M. C., & Urbany, J. E. (2005). Reasoning About Competitive Reactions: Evidence from Executives. Marketing Science, 24(1), 138–149.

Nijs, V. R., Dekimpe, M. G., Steenkamp, J.-B. E. M., & Hanssens, D. M. (2001). The Category-Demand Effects of Price Promotions. Marketing Science, 20(1), 1–22. Pinckaers, M. (2016). The Dutch Food Retail Market (Report No. NL6037). The Hague.

South Holland

Primeaux, W. J., Bomball, M. R., Stigler, G. J., Phillips, A., Simon, J. L., Larner, R. J. Rasche, R. H. (1974). A Reexamination of the Kinky Oligopoly Demand Curve. Source

Journal of Political Economy, 82(4), 851–862.

(42)

40

Raworth, K., Narayan, S., Sweetman, C., Rowlands, J., & Hopkins, A. (2012). Conducting semi-structured interviews: Why master this skill? How to conduct a semi-structured interview. Oxfam, 1–5.

Reibstein, D. J., & Farris, P. W. (1995). Market Share and Distribution: A Generalization, a Speculation, and Some Implications. Marketing Science, 14(3), 190–202.

Ryan, R. M., & Deci, E. L. (2001). On happiness and human potentials: A Review of Research on Hedonic and Eudemonic Well-Being. Annual Review of Psychology, 52, 141–166.

Smith, K. G., Grimm, C. M., Gannon, M. J., & Chen, M.-J. (1991). Organizational information processing. competitive responses and performance in the U.S.domestic airline industry. Academy of Management Journal, 34(1), 60–85.

Steenkamp, J. B. E. M., & Dekimpe, M. G. (1997). The Increasing Power of Store Brands: Building Loyalty and Market Share. Long Range Planning, 30(6), 917–930.

Steenkamp, J.-B. E. M., Nijs, V. R., Hanssens, D. M., & Dekimpe, M. G. (2005). Competitive Reactions to Advertising and Promotion Attacks. Marketing Science, 24(1), 35–54. Urbany, J. E., Dickson, P. R., & Key, R. (1991). Actual and perceived consumer vigilance in

the retail grocery industry. Marketing Letters, 2(1), 15–25.

van Heerde, H. J., Gijsbrechts, E., & Pauwels, K. (2008). Winners and Losers in a Major Price War. Journal of Marketing Research, 45(5), 499–518.

Van der Staak, B. (2018, February 27) Ene filiaal Jumbo duurder dan andere.

Consumentenbond. Retrieved from:

https://www.consumentenbond.nl/nieuws/2018/ene-jumbo-duurder-dan-de-andere

Van Aalst, M, Sloot, L. van der Blom, L, & Kivits, L. (2005), "Het ‘Grote Voordeel’ van één Jaar Prijsoorlog," Erasmus Food Management Instituut 2005-01, January, ISBN 90-77015-18-3.

(43)

41

APPENDIX 1

1. Hoe vaak koopt u producten uit de volgende productcategorieën bij een bezoek aan een supermarkt?

Nooit (1) Zelden (2)

Af en toe (3)

Soms (4) Vaak (5) Meestal (6)

Altijd (7)

Product

group x

o

2. Utilitaire producten kunnen worden omschreven als producten die nuttig, praktisch, noodzakelijk, handig en functioneel zijn.

In hoeverre bent u het ermee eens dat de producten in de volgende productgroepen kunnen worden omschreven als utilitaire producten?

Zeer mee oneens (1) Mee oneens (2) Af en toe (3) Neutraal (4) Enigszins mee eens (5) Mee eens (6) Zeer mee eens (7) Product group x

o

3. Hedonische producten kunnen worden omschreven als producten die leuk, lekker, aangenaam, spannend en plezierig zijn.

In hoeverre bent u het ermee eens dat de producten in de volgende productgroepen kunnen worden omschreven als hedonische producten?