Assessing the optimal price for national brands at discounters

(1)

Assessing the optimal price

for national brands at

discounters

Supervisor: dr. J.E.M. van Nierop

Co- assessor: dr. F. Eggers

15/1/2018

University of Groningen H.T. Nolden

Master Thesis Marketing Intelligence 2064510

Abstract

(2)

1

Preface

The completion of my thesis means the final step in achieving my Master’s degree in Marketing Intelligence. During the track I developed special interest in handling data with various statistical methods. After my graduation I aspire to find a job in this field to continue working with marketing and data. I would like to thank my supervisor Dr. Erjen van Nierop for the excellent guidance during the process. During the several stages of writing he always provided me with effective feedback and was always open to answering questions in times there were difficulties in making progress. Furthermore, I would like to thank my fellow group members and wish them much luck in the future.

(3)

2

1. Introduction

The grocery retailing format has evolved over the years. Colla (2004) conducted a study about retail trends in European countries and found that changes in the retail environment are results of several factors. First demographic variables such as income growth, but also the increase of the avarage age and decrease of the population growth. Second, legislation is important as products have to meet specific standards before they are allowed to be sold. Third, technological development has enhanced the possibility of information transmission which increases efficiency for example. This evolution in the grocery retail sector eventually emerged into the shaping of four types of grocery retail formats: Full-line supermarkets, hypermarkets, convenience stores and discounters (Child, Kilroy, & Naylor, 2015). Discounters are distinctive from other retailers as their focus lies on price competition. Traditionally, discounters` strategies focused on selling private labels (PL), or store brands, at low steady prices while offering minimal extra services. This strategy is often referred to as the no-frills strategy. Discounters typically offer few choices and limited national brands. In order to offer low prices, they are more or less forced to give up other aspects such as service levels and store atmosphere. Also, products are often displayed in rough boxes instead of being neatly placed on shelves (Saba & Sharma, 2012). Hard-discounters aim to minimize operating costs by having an efficient supply chain as a result of focus on private labels and few SKU’s (Steenkamp & Kumar, 2009). Examples of retailers applying this strategy are Aldi and Lidl (Schwarz Group) which are the leading discount chains. These formats have grown substantially in the recent years and future prospects are still promising. Based on historic growth rates, the top nine discount chains will see their sales grow by 41% annually until 2020 (European Grocery Discounters Report, 2015).

(5)

4 format is responsible for the decline of sales of manufacturer brands. They ascribe this partly due to underestimation by service retailers that have several misconceptions about discounters. For example, they assume that discounters only attract consumers with low incomes. Despite the fact that the lower segments are important to the discounter, this assumption is not true. Furthermore, retailers think that discounters offer lower quality. Several research institutions such as GfK investigated this issue and proved it wrong. However, consumers tend to expect that national brands do have better quality signalled by higher prices (Ailawadi, Neslin, & Gedenk, 2001). Manufacturers of national brands are also interested in offering their products at discounters as they are threatened by the growth of private brands at traditional retailers. By being listed at discounters they have a stronger bargaining position with other retailers. Encouraging discounters to offer more national brands may be the most effective way to keep store brand growth in check (Dhar & Hoch, 1997).

Whereas manufacturers have to compete with private labels or store brands, discounters have to compete with each other and with service retailers. In the past, there has not been much competition between discounters and service retailers since they targeted contrasting segments (Fox, Montgomery, & Lodish, 2004). However, with the introductions of private labels at service retailers and national brands at discounters, they have increasingly been targeting each other’s segments. Cleeren et al. (2010) found that there is competition within service retailers and discounters, but that competition between service retailers is more intense. Furthermore, they argue that profits of supermarkets are only threatened when three or more discounters enter their area.

(6)

5 discounter. However, this finding seems contradictory as a larger within store price gap requires a higher price for the national brand, but a larger between-store price gap requires a lower price for the national brand. Lourenço & Gijsbrechts (2013) conducted a study about the price perception of customers of discounters listing national brands. Whereas a high price may weaken the discounters’ favourable value positioning, a low price may degrade quality perceptions compared to the private label. This study will focus on this price setting issue and examine what the optimal price for national brands at discounters is. On the one hand, the price should be low enough to induce consumers into switching stores from mainstream retailers to discounters. On the other hand, the price should be high compared to the private label in order to increase category demand. This optimal price will be assessed by investigating consumer preferences using logistic regression analysis. Furthermore, the relative importance of pricing compared to other factors such as price consciousness, distance to supermarket(s) and income will be regarded in order to find the magnitude of the pricing factor. This leads to the following research question:

What is the optimal pricing strategy regarding the price gaps for national brands at discounters?

This question will be answered using the following sub-questions:

- How large should the between-store price gap be in order to make national brand buyers switch from service retailers to discounters?

- How large should the within-store price gap be in order to prevent buyers of store brands at discounters to switch to national brands at the same store?

- What is the relative importance of price compared to other factors influencing store choice behaviour for national brands?

(7)

6

2. Theoretical Framework

National brands and store brands

Store brand choice versus national brand choice has been researched widely in the past. It is empirically generalized that consumers perceive national brands better than store brands in terms of general quality and other facets such as taste and reliability (e.g . Bellizzi et al., 1981 ; Cunningham et al., 1982). Nenycz-Thiel & Romaniuk (2014) argue that store brands differ from national brands because of inadequate advertising out of store and its small distribution across stores. Therefore, knowledge about the store brand is often missing. Store brands are often not used for certain product categories because they are not credible for them (Martenson, 2007). Examples are high involvement products and home appliances. National brands are generally higher priced than store brands. A reason for purchasing national brands is that they are more socially acceptable because of better packaging and image (Baltas, 1997). Sethuraman & Cole (1999) investigated factors that determine the premium consumers are willing to pay for national brands. First, one of the drivers is perceived quality. Second, a higher premium is paid for products providing high consumption pleasure. Third, higher premiums are possible for product categories that are bought infrequently. Fourth, younger people and females are willing to pay higher premiums for national brands.

Whereas perceived quality is in favor of the national brand, studies have demonstrated that there is no actual difference in quality between store brands and national brands. In an experiment, the quality of national brands was rated lower when participants were blinded compared to ratings of non-blinded participants (De Wulf et al., 2005). For store brands, there was no difference between settings. This finding implicates that national brands have higher brand equity. Additionally, they found that brand equity is positively moderated by brand loyalty and store loyalty for store brands. Therefore, Richardson, Dick, & Jain (1994) conclude that consumers’ primarily rely on extrinsic cues (price, brand, packaging) when comparing quality of products instead of intrinsic cues such as ingredients. Steenkamp, Van Heerde, & Geyskens (2010) confirmed this finding. They looked for drivers that make consumers pay a higher price for national brands and found that these are mediated by perceived quality. They considered marketing and manufacturing factors. They mention that innovation is a good way to enlarge the perceived quality gap between national brands and store brands. Furthermore, they found that distinctive packaging is the most important driver to increase willingness to pay. Manufacturing factors (PL maturity stage, required craftsmanship) also increase perceived quality and therefore willingness to pay.

(8)

7 Geyskens, 2013). European retailers even reported store brand market shares of 40%. According to Ailawadi & Harlam (2004), store brand generally have higher margins than national brands. Mainly because suppliers of store brands have weaker market power compared to national brand suppliers. Furthermore, national brands incur higher advertising costs which are reflected in retail prices. Store brands are not subjected to advertising costs which leads to higher margins. However, this finding does not apply to every category. Also, percentage margins may be higher for store brands whereas absolute margins may be higher for national brands. A side effect of store brands is that they typically are umbrella brands (Semeijn et al., 2004). This means that the brand covers various distinct categories such as food, but also laundry detergents. This could result into negative evaluations of the overall store brand because of an unpleasant encounter with one product category. Moreover, consumer confidence in the entire store can be damaged by negative experiences with the store brand (Thompson, 1999).

Corstjens & Lal (2000) studied the effect of store brands on store loyalty. They concluded that, if the store brands are of sufficient quality, these brands can create a differentiation effect, increase store loyalty and increase profitablility. Moreover, Sudhir & Talukdar (2004), found that store brands contribute more to store differentiation than to price sensitivities. Profits can increase even when store brands have lower margins than national brands. In that case, store brands and national brands act as complements as one creates loyalty and differentiation whereas the other enhances profitablity through high prices. On the other hand, lower quality store brands have opposing effects as consumers become more prone to prices. All-in all it is important for the retailer to maintain a well balanced assortment of store brands and national brands.

Other research studied personality traits in relation with store brand attitude. Store brand buyers behave more independent and are less reliant on behavioural norms (Becherer & Richard, 1978). Familiarity with store brands is also important in buying decisicions as a lack of familiarity leads to the belief that purchasing store brands is a risky decision because of quality concerns (Richardson, Jain, & Dick, 1996). A paper of Sinhaa & Batrab (1999) investigated price consiousness in categories that are expected to be risky. They confirmed their expectation that consumers are less price consious in riskier categories. In their study, products like medicines were perceived to be most risky and frozen vegetables least risky.

Price gaps

(9)

8 prices than service retailers, price sensitive consumers may choose to buy the brand at a discounter. This is favourable for the discounter, as incremental earnings increase and store traffic increases as well. Second, the within store price gap is the price difference between store brands and national brands within the chain. Lower within-store price gaps may encourage consumers switching from store brands to national brands.

Several studies address price setting behaviour of national brands and store brands. Bontemps et al., (2008) Investigated how national brand pricing is affected by the success of store brands. They found that national brand prices rise as store brands are developing, even when quality changes are limited. Therefore, national brand characteristics are not the only determinant of national brand pricing. Cotterill & Putsis Jr (2000) conducted a study about price setting behaviour of national brands and store brands. They had several findings. First, they concluded that national brand and store brand prices both increase as national brands sales increase. Second, leading supermarket chains are able to narrow the within store price gap when retail concentration increases. Third, national brand prices affect store brand sales whereas there is no evidence of the reverse relation.

(10)

9

Group Description Manufacturer Discounter

I

In-store brand switchers: purchased store brands at

discounters

+

-

II

New brand and store buyer: purchased store brands at

service retailers

+

III

Store switchers: Purchased national brands at service

retailers

-

+

Notes:” + “means that the new buyers are beneficial and “ - “ means that they are harmful to profitability.

Figure 1: Potential national brand buyers at discounters (Deleersnyder & Koll, 2012)

According to Lourenço & Gijsbrechts (2013), within-store comparisons can have opposing effects. On the one hand, larger gaps between store brand and national brands may signal superior quality of the national brand. Kahn & Lehmann (1991) mention that a national brand may stand out among an assortment of lower priced brands which increases consumer perception of the assortment. Higher perceived and more varying assortments can lure new customers. Furthermore, store brand sales may increase because the price of the private label may be perceived more favourable due to the greater contrast with national brand prices (Meredith & Maki, 2001). On the other hand, the high pricings may influence the overall price perception of the discounter to be more expensive which may deter the lower income segments. On average, price gaps between store brand and national brands are 40%, ranging between 20% and 40% across categories (Dhar & Hoch, 1997). According to Ailawadi et al. (2001), store brand buyers are typically price conscious, less quality conscious and loyal to the store. Whenever a national brand is on promotion, they are likely to switch to that brand. However, discounters usually do not take part in (deep) promotions as part of their everyday low pricing strategy. Overall, Higher within store price gaps have proven to enhance category demand and are therefore beneficial to discounters (Deleersnyder et al, 2007). Undesired within-store switchers (Group I) are deterred through high national brand pricing.

H1: The higher the within-store price gap, the less unprofitable consumers will switch from store brands to national brands within discounters

(11)

10 they have the opportunity to save costs (Broniarczyk et al., 1998). It can be expected that the more the price of the national brand at the discounter will be decreased, and thus an increasing between-store price gap, the more consumers will be buying the brand at the discounter.

H2a: The higher the between-store price gap of national brands at service retailers versus national brands at discounters, the more beneficial consumers will switch from service retailers to discounters.

For completeness, store switchers that are currently buying store brands will also be included in the study. Since private labels of service retailers are likely to be priced lower than national brands at discounters if they would be listed there, the between-store price gap is expected to have a negative effect here. A smaller price gap is expected to increase the probability of switching from a store brand at a service retailer to a national brand at a discounter.

H2b: The lower the between-store price gap of store brands at service retailers versus national brands at discounters, the more beneficial consumers will switch from service retailers to discounters

Store choice attributes

(12)

11

Table 1 (Saba & Sharma, 2012)

Lamey (2014) studied the performance of discounters during times of different economic prosperity. The findings confirm that discounters perform better during economic downturns. Furthermore, the research indicated that the increased market share is maintained after the economic contraction for both, hard and soft discounters. This means that recession has a short term, but also a permanent positive effect on discounter performance. By introducing more varying assortments by listing national brands, discounters attempt to intensify competition with service retailers.

(13)

12 may only visit the discounter to buy a few national brands for low prices. However, if this occurs more frequently, familiarity with the discount store may increase and therefore loyalty and profitability may increase as well. Also, every additional customer attracted by listing national brands is beneficial.

Demographics also are important attributes in store choice behaviour (Cummins et al., 2008). First, as age increases, shopping frequency, and shopping time both increase as well. Also, Lumpkin et al. (1985) found that older people are more price sensitive than younger people. Second, married people value grocery shopping more than single people. Third, shoppers with higher incomes shop less frequent, but spend more time on their shopping trips. Baltas & Argouslidis (2007) looked for characteristics of store brand buyers. Surprisingly, they found that people with greater income and higher education levels are more prone to store brands. They ascribe this to the promise of making smart buying decisions by choosing good quality for low prices. Another explanation is that more intelligent consumers are able to process intrinsic attributes such as ingredients better. However, Ailawadi et al. (2001) argue that higher income reduces price sensitivity and therefore preference for store brands decreases. Since this thesis isinvestigating switching behaviour, it is expected that people with higher incomes who usually buy a brand at a service retailers are less inclined to switch to a discounter due to lower price sensitiveness.

H3: Income level decreases the effect of the between-store price gap between national brands at discounters versus retailers on national brand buying at discounters.

Not all consumers are perfectly loyal to a specific store. Often, consumers visit multiple stores to purchase different sets of products. Customers only buying one or two products are not very profitable. Multiple store patronage is defined as the number of stores patronized by consumers. Factors determining the number of stores visited are for example availability of time and store brand proneness (Baltas, Argouslidis, & Skarmeas, 2010). Store brand proneness is indicative of aiming for best buying decisions from an economic point of view (Batra & Sinha, 2000). Therefore, multiple store patronage can be expected as these consumers are looking for the best deals. This study specifically focuses on finding prices for national brands that persuade buyers of that brand to switch from service retailers to discounters. Consumers that are price conscious are interesting in that regard as they can probably get a better deal at discounters. Therefore, it is expected that consumers who have higher price consciousness, to be more likely to purchase a national brand at a discounter.

(14)

13 Geographic location also plays a role in store choice (Huff, 1964). Logically, consumers choose between stores that are located close to them. Costs of shopping increase as the distance to the supermarket increases because of transportation costs. Furthermore, a shorter distance between grocery stores increases the likelihood that multiple supermarkets will be visited as incremental transportation costs are lower (Fox & Hoch, 2005). Therefore, It is expected that a lower travel distance increases switching from service retailers to discounters. In this study, a distance measure will be used that is relative to the distance to the service retailer that a consumer usually visits.

H5: Lower distance to the discounter relative to the service retailer increases the effect of the between-store price gap on the probability to switching to a national brand at a discounter

Conceptual model

In summary, this study aims to find out what the best strategy is for national brands to get more profitable customers for discounters by optimizing the between-store and within-store price gaps. The first objective is to find price gaps where buyers of store brands or national brands at service retailers switch to national brand buying at discounters. Simultaneously, the group that switches from store brands to national brands within discounters should be kept as small as possible as this group has proven to be unprofitable. The second objective is to find out how relevant the price level actually is compared to other attributes that determine store choice. Figure 2 shows an overview of the relationships in a conceptual model.

National brand buying at discounter Between-store price

gap between service retailer and discounter(H2)

- Income level - (H3)

- Price consciousness + (H4) - Distance – (H5)

Within-store price gap PL – NB at discounter

(15)

14 Figure 2: Conceptual Model

3. Research Design

Data collection

Consumer preferences will be investigated by means of binary logit analysis. In a binary logit model, the dependent variable either has a 0 or 1 value. For example, 0 can mean “no” and 1 can mean “yes”. In logistic regression, the coefficients indicate how the variables influence the probability that the outcome of the dependent variable will be 1. As this study concerns within-store switching as well as between-store switching, separate predictions will be performed. The dependent variable either

indicates whether customers switch from service retailers to the national brand at a discounter (0 = Stay, 1 = Switch) or switch from a store brand to a national brand within a discounter. By means of a survey, participants are asked to specify in which store and of which brand they currently buy a specific product. They can choose between the two largest discounters (Aldi, Lidl) and the two largest service retailers (Albert Heijn, Jumbo) in terms of market share (Figure 3). A national brand that is not yet available at discounters is preferred because people may already have switched if the brand is available there. Thus, these switchers cannot be asked whether they would switch or stay as they have already switched from the service retailer to the discounter.

In order to reach as many complete responses as possible, a product is desired that is bought by a high percentage of households. Product that meet these requirements is milk from the national brand Campina as their product totals add up to a market penetration of 85% (Gfk, 2015). After the respondents have specified what their current preferred brand is and from which chain, they get to choose whether they would consider buying Campina at a discounter or stick to their initial choice. Respondents stating that they already buy at a discounter will be provided with choices regarding in-store switching whereas respondents stating they usually purchase at service retailers will be provided with choices regarding between-store switching. Both groups will be shown 8 different choice-sets

(16)

15 with varying price gaps. A situation will be sketched where consumers are asked to express where they would go shopping in case they need only one product, as it is not very plausible that consumers will alter their major shopping trips format to a discounter by adding only one national brand to the assortment. Besides these choice options, survey questions will be asked in order to collect data about travel time, income levels, price consciousness and demographics. For distance, a cognitive value is produced by asking the respondent to estimate the travel time in minutes to the nearest of each chain which is comparable to the method used by Kraus et al. (2015). The reason for measuring perceived distance instead of exact geographical distance is that travel time differs for people who travel by car, foot or other methods. The survey questions can be found in Appendix A. An overview of the measured variables with possible values is visible in Table 2

In order to validate whether the constructs used for measuring price consciousness are valid, factor analysis will be used. KMO and Bartlett’s tests and Cronbach’s’ alpha will be used as measures. Factor analysis is appropriate when the KMO value exceeds 0,5 and Bartlett’s test turns out significant (Malhotra & Birks, 2009). The Cronbach’s alpha is used to verify whether the constructs together are a reliable factor. The value should be at least 0,6. The number of factors depends on several criteria. First, eigenvalues should be higher than 1. Second, the cumulative variance explained should be at least 60%. Third, the individual variance explained of each variable should at least be 5%>

Variable

Values

Gender Male, Female

Age <25, 25-45, 45-65, >65

Income <1000, 1000-2000, 2000-3000, 3000-4000, > 5000

Education University, higher, middle, lower, other

Household size 1, 2, 3, 4, 5 or more

Number of chains regularly

visisted 1 - 5

Distance to chain (minutes) 0 – 30 Inter-format distance (minutes) 0 - 30

Current choice Discounter PL, Retailer PL, Retailer NB

Price consciousness 1 (Low) – 7 (High)

Between-store price gap Retailer NB – Discounter NB

(17)

16 Between-store price gap Retailer

PL – discounter NB

€0,32 - €0,53 (Settings shown to the respondent with intervals of 3 cents, randomly displayed) Within-store price gap

Discounter

€0,32 - €0,53 (Settings shown to the respondent with intervals of 3 cents, randomly displayed)

Switch Dependent variable for every price gap(Yes/No)

Table 2: Measured variables

The price gaps are manipulated by displaying two products with differing prices. The true prices for the relevant chains are displayed in Table 3. Respondents are asked to choose between their initial choice and the national brand at one of the discounters. Prices of the initial choice are set around the true prices at the store. Prices of the national brand at the discounter are fluctuating to manipulate the gaps and find the optimal price at which consumers are likely to switch.

Chain Price store brand Price national brand

Albert Heijn € 0,75 € 1,23

Jumbo € 0,73 € 1,22

Lidl € 0,75 -

Aldi € 0,85 -

Table 3: True milk prices per chain (Groningen, The Netherlands)

People are asked to fill in the survey through several channels on the internet. First, social media has been used to reach the personal network (friends and family). Second, the survey link has been posted in relevant internet fora to collect data from people outside the personal network. Furthermore, QR-codes with a link to the survey were distributed. In order to motivate respondents to fill in the entire survey, the choice sets were presented as simple and short as possible to prevent exhaustion. Also, a gift card has been offered to those respondents who completed the survey in order to create an incentive to finish the survey.

Plan of analysis

(18)

17 𝑦_𝑖∗_{= 𝛽}

0+ 𝛽1𝑥1+ 𝛽2𝑥2+ ⋯ + 𝛽𝑘𝑥𝑘+ 𝜀𝑖

In a binary logit model, 𝑦∗ is assumed to be a latent variable which follows a linear model. If y* takes on a value above 0, the respondent will be classified as switcher. If y* takes on a value below 0, the respondent is predicted to stay. 𝑦𝑖∗ Represents the predicted value which can either be stay (0) or switch (1) based on probabilities, and not on observed values, for respondent i.

The predictor variables are the within-store price gap and between-store price gap. For every price gap, respondents have indicated whether they would switch or not. These variables are continuous as the price gaps are in the form of an interval with steps of 3 cents. In the analysis, the coefficients indicate how a change in the price gap affects the probability that a respondent will switch based on odds (Leeflang et al., 2015). The odds define the ratio of the probability that a consumer would switch versus the likelihood that a consumer will stay. Probabilities are estimated by using the cumulative distribution function:

𝑝_𝑖 = 𝐹(𝑋_𝑖′𝛽) = exp (𝑋𝑖 ′_𝛽) 1 + exp (𝑋_𝑖′_𝛽)

When the predicted probability that a consumer will switch to the national brand at a discounter is higher than 0,5, that consumer will be classified as switcher. The logit model used in the analysis is defined as:

𝑦_𝑖∗_{= 𝛽}

0+ 𝛽1𝑋1+ 𝛽2𝑋2+ 𝛽3𝑋3+ 𝛽4𝑋4+ 𝛽5𝑋5 Explanations of the parameters are given in Table 4.

Explanation

y

_i∗ _{Latent variable}

𝛽

₀ Intercept

𝑋

_𝑗 Value for variable j (j = 1, 2, 3, 4, 5)

𝛽

₁ Coefficient Between-store price gap

𝛽

₂ Coefficient within-store price gap

𝛽

3 Coefficient Income level

𝛽

₄ Coefficient price consciousness

(19)

18

Validation

Model fit will be assessed by means of Maximum likelihood, Cox & Snell R square and Nagelkerke R square. The closer the value of the loglikelihood to 0, the better the model fits the data. The R square values are based on comparisons between the model with predictors and the model with only a constant. The Cox & Snell R square is calculated as: 1 − (−2𝐿𝐿_−2𝐿𝐿0

𝑘)

2/𝑁_{. The Nagelkerke R square is} calculated by dividing the Cox & Snell R Square by the term: 1 − (−2𝐿𝐿0)2/𝑁. Furthermore, the hit-rate will be assessed. The hit-hit-rate indicates the percentage of observations that is predicted correctly by the model. In order to give more insight in the optimal price gaps, a graph will be made that displays the cumulative percentage of within-store switchers or between-store switchers with increasing price gaps.

In contrast to linear regression, logistic regression does not have to meet assumptions like normality and homoscedasticity of the residuals. There are other assumptions that logistic regression does have to satisfy (Hosmer Jr., Lemeshow, & Sturdivant, 2013). First, independent variables should not be correlated to each other. Correlations of the independent variables regarding between-store are reported in Table 5: Correlations between independent variables. The table shows that the highest correlation is 0,22 which means that correlation between independent variables is no issue. Second, the independent variables should have a linear relationship with the log odds ( ln_1−𝑝𝑝 ). A linear regression is performed showing that all variables have a significant relationship with the log odds (Table 6). Third, the sample size should be sufficiently high. A common rule of thumb is 10 cases for each predictor variable. With a total of 305 respondents and case replication for every displayed price gap (times 8) the dataset is sufficiently large.

Price gap Income Price consciousness Relative distance Price gap 0,00 0,00 -0,03 Income 0,00 0,03 0,22 Price consciousness 0,00 0,03 -0,16 Relative distance -0,03 0,22 -0,16 Table 5: Correlations between independent variables

(20)

19

Constant

-3,56 0 -407766350,8 0

Relative distance

0,036 0 0,256 125817026,1 0

Price consciousness

0,387 0 0,544 274293362,2 0

Price gap

0,12 0 0,839 429829902,1 0

Income

-0,059 0 -0,066 -32971568,51 0 Table 6: Linear regression of independent variables on log odds

Data cleaning

Before performing the analysis, some issues with the dataset have to be resolved. First, double entries are removed by comparing IP-addresses. Of course, different people on the same internet connection could have filled in the survey. Therefore, those respondents will be checked more thoroughly to check if they are truly equal by comparing e-mail addresses or other variables. Also, respondents that chose the option ‘other’ when they were asked to indicate which brand they currently buy will not be included in the analysis as the questions regarding switching behaviour were not displayed to them. Furthermore, incomplete responses are removed from the data as well. The second issue is that values for the variable ‘Price gap’ are spread over column names. To

(21)

20

4. Results

Descriptives

After cleaning the dataset, a total of 259 respondents remained. Figure 4 shows how they are divided based on what kind of brand they currently buy at what format. Campina contains the lowest number of buyers which is surprising because Campina has a market penetration of 85% in the Netherlands. People assigned to ‘other’ are for example buyers of biological milk or do not buy dairy products at all. Appendix B provides some frequency plots for the demographic and socio-economic variables of the respondents that are categorical. Additionally, Figure 5 shows boxplots of how price consciousness levels over the groups. It shows that

store brand buyers on average are more price conscious than national brand buyers (Campina).

Figure 5: Boxplots of price consciousness level per group

(22)

21

Factor analysis

The value of 0,633 and statistical significance for the Bartlett’s test show that factor analysis is appropriate. Cumulative variation explained shows that two factors are the best solution (Appendix C). Factor loadings suggest that the number of chains visited is not explaining the same information as the three questions regarding price consciousness. Therefore, the number of chains will be left out. The Cronbach’s alpha for the other three variables has a value of 0,728 and therefore form a reliable factor. This factor will be used as the price consciousness variable in further analysis

Within-store switching at discounter (PL – NB)

Within-store switching concerns consumers that already purchase a store brand at a discounter and are asked whether to switch to the national brand if it is introduced, given certain price gaps. Figure 6 shows the percentage of switchers at every price gap. The amount of switchers seem relatively low. Table 7: Goodness of fit contains the R2

values of the model. The model

explains between 4,7% and 17% of the variation which is very low. The number of switchers is very low with the given price gaps. Hence, the model predicted not a single switcher and yet has a hit-rate of 95,6% Figure 8: Percentage of switchers for every price gap (NB service retailer – NB discounter. The hit-rate shows the percentage of cases that were

classified correctly. However, the hit-rate is not useful in this case since there are no cases where

Table 7: Goodness of fit

-2 Log likelihood

Cox & Snell R Square Nagelkerke R Square 187,472 0,047 0,170

Predicted

Percentage

Correct

Stay Switch Observed Stay 627 0 100 Switch 29 0 0 95,6 Table 8: Classification table

(23)

22 the model predicts switching successfully.

According to Wooldridge (2012, pp. 590), a cut-off value of 0,5 is inappropriate when it is not likely that an event will occur (Y = 1) because the predicted probability never exceeds 0,5. Since the data is skewed towards staying with the store brand, the model does not predict switching adequately. In that case, the cuf-off value should be set to the same percentage as that an event occurs. Switching only occurs in 4,3% of the observations for private label buyers within discounters. Therefore, a new model was estimated with a cut-off level of 0,043. Although the overall hit-rate is worse than in the initial model because of worse prediction of non-switchers, the enhanced model predicts switchers more effectively (Tables 9 and 10).

Beta Standard error Wald Odds ratio

Pricegap -0,210 0,046 20,747 0,810

Constant 4,900 1,659 8,719 134,310

The wald index shows statistical significance for the price-gap variable (Table 11). The probability that a store brand buyer at a discounter will switch to a national brand at a discounter increases with 19% with a unit decrease in the within-store price gap (0,810 – 1 = -0,19). Therefore, the hypothesis that an increase in the within-store price gap has a negative effect on switching from a store brand to a national brand within the discounter is confirmed (H1).

Between-store switching (PL service retailer – NB discounter)

Predicted Percentage Correct Stay Switch Observed Stay 415 231 64,2 Switch 5 21 80,8 64,9 -2 Log likelihood

Cox & Snell R Square

Nagelkerke R Square

187,472 0,047 0,170 Table 9: Goodness of fit with cut-off level of 0,043

Table 10: Classification table with cut-off level of 0,043

(24)

23 Between-store switching concerns two groups: Consumers who already buy the national brand at a service retailer and consumers who buy a store brand at a service retailer. Similar to the within-store switchers, the number of switchers from store brands at service retailers to national brands is very low. Figure 7 shows the number of switchers at each price gap. Out of 72 of the consumers, less than 15% indicated to switch to a national brand at a discounter at the smallest set price gap. Just as in the model concerning within-store switching, this model predicts zero switchers and therefore the hit-rate is not useful. As the event of switching for this group occurs for 4% of the observations, the cut-off level will be set to 0,04.

Figure 7: Percentage of switchers at every price gap

The results for the between-store switchers that currently purchase store brands at service retailers are displayed in tables 12 and 13. This model contains the same variables as for the national brand buyers as they concern store choice. Here, the price gap is expected to have a negative effect instead of a positive effect, because the store brands are priced lower than the national brands. Therefore, an increasing price gap should lower the motivation to switch to a national brand at a discounte.. The wald index shows statistical significance for the 1% level (Table 13). Therefore, hypothesis 2b, that a lower between-store price gap (PL SR – NB discounter) has a positive effect on national brand buying at a discounter is confirmed. Furthermore, a significant effect was found for relative distance and income (p < 0,05).

-2 Log likelihood

Cox & Snell R Square

Nagelkerke R Square

Hit-rate

156,802 0,061 0,215 73,1

(25)

24

Beta S.E. Wald df Odds ratio

Pricegap _-0,195 _0,047 _{17,370** 1} _0,823

Income _-0,450 _0,207 _4,730* ₁ _0,638

Relative distance _0,070 _0,027 _6,504* ₁ _1,072 Price consciousness -0,260 0,192 1,835 1 0,771

Constant _6,133 _2,094 _8,581** ₁ _460,879

Table 13: Estimation results between-store price gap (PL – NB) *Significant at P <0,05

**Significant at P <0,01

According to this model, the utility for a store brand buyer at a service retailer to switching to a national

brand at a discounter can be calculated

as: p(Y =1) = _{1+ exp (6,133−0,195∗𝑝𝑟𝑖𝑐𝑒𝑔𝑎𝑝−0,450∗𝑖𝑛𝑐𝑜𝑚𝑒𝑙𝑒𝑣𝑒𝑙+0,70∗𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒)}exp (6,133−0,195∗𝑝𝑟𝑖𝑐𝑒𝑔𝑎𝑝−0,450∗𝑖𝑛𝑐𝑜𝑚𝑒𝑙𝑒𝑣𝑒𝑙+0,70∗𝑟𝑒𝑙𝑎𝑡𝑖𝑣𝑒 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒) . For example a respondent with an income level of 2, who has 5 minutes shorter traveling time relative to the service retailer while the price gap is 41 cents has a probability of _{1+exp(−2,607)}exp (−2,607 * 100% = 6,87%. As this is above the threshold of 4%, this respondent is classified as switcher. As the data involves very few switchers, no moderation analysis or examination of relative importance of the predictors will be performed.

Between-store switching (NB service retailer – NB discounter)

The Campina buyers at service retailers consist

of 336 observations (42 * 8). Consumers were given the option to choose between their original choice and the national brand at one of the two discounters given several price gaps. In some cases respondents could choose for Aldi and in other cases for Lidl as a discounter. This may distort the effect of the between-store price gap as respondents logically are more likely to switch to one of the two

discounters when one is closer than the

other. Figure 8 reveals this effect. The graph shows the number of switchers moving up and down first, but the trend is upwards. Price gaps 3, 9, 15, and 21 were displayed in combination with switching to Aldi. Lidl was displayed combined with the other price gaps.

(26)

25

-2 Log likelihood Cox & Snell R Square Nagelkerke R Square

178.406 0.157 0.219

The model is statistical significant (Chi-square = 14,005(8), P < 0,10). The hit-rate of the model is 68,9%, meaning that 68,9% of the predicted classifications were correct (Table 15).

None of the coefficients showed statistical difference from 0 (Table 16). However, insignificant signs can be a result of multicollinearity. Especially with multiple interaction terms, multicollinearity can become an issue (Hayes & Matthes, 2009). Therefore, the hypotheses will be tested by first analysing the main effects and then add the interaction effects separately. By building the model step by step, it

Predicted Correct Stay Switch Observed Stay 96 15 86,5% Switch 36 17 32,1% 68,9%

Table 15: Classification table

Table 16: Moderation analysis

Variable B S.E. Wald df Odds ratio

(27)

26 was found that the variables become insignificant after adding the interactions of price consciousness or relative distance. The results for the main effects are displayed in tables 17 and 18.

-2 Log likelihood

Cox & Snell R Square

Nagelkerke R

Square Hit-rate

377,528 0,163 0,222 70,4%

Beta Standard Error Wald Odds ratio

Constant -3,560 0,646 30,418*** 0,028

Price gap 0,120 0,019 37,635*** 1,127

Income -0,059 0,117 0,255 0,942

Price consciousness 0,387 0,098 15,513*** 1,472

Relative distance 0,036 0,020 3,444* 1,037

Table 18 * Significant at 0,10 level **Significant at 0,05 level ***Significant at 0,01 level

The model shows higher R square values than the models for the store brands. Furthermore, the price gap variable shows a significant effect at the 1% level. Therefore, hypothesis 2a, that the between-store price gap has a positive effect on switching from a NB at a service retailer to a NB at a discounter is confirmed. The odds ratio takes on a value of 1,127 which implies that an increase in the price gap by 1 cent increases the probability that a consumer will switch by 12,7%. Furthermore, significant effects of price consciousness and relative distance were found. According to this model, the utility of a national brand buyer at a service retailer will switch to the same national brand at a discounter can be calculated as: -3,560 + 0,120 * Price Gap + 0,387 * Price consciousness + 0,036 * Relative distance. For example, a consumer that has a level of 6 for price consciousness and whose travel time to a discounter is 10 minutes shorter than to the service retailer while the between-store price gap is 15 cents has a utility of : -3,560 + 0,120 * 15 + 0,387 * 6 + 0,036 * 10 = 0,992. the

probability that this consumer will switch is _{1+exp (0,992)}exp (0,992) = 72,95%. As this probability exceeds the cut-off value of 0,5 the consumer will be classified as switchers.

The results of the moderation effect of income are provided in tables 19 and 20. All variables show statistical significance, except for the moderator variable income. Therefore, Hypothesis 3, that Income level decreases the effect of the between-store price gap on national brand buying at

(28)

27 Beta Standard error Wald Odds ratio

Pricegap 0,142 0,054 6,886*** 1,152 Income 0,055 0,283 0,038** 1,056 Price consciousness 0,39 0,099 15,555*** 1,478 Relative distance 0,036 0,02 3,469* 1,037 Income * Pricegap -0,008 0,018 0,196 0,992 Constant -3,898 1,007 14,991*** 0,02

Table 20: Moderation analysis of price gap * Significant at 0,10 level

**Significant at 0,05 level ***Significant at 0,01 level

Leeflang et al. (2015) posit various solutions for dealing with multicollinearity. Examples are specifying other models or recoding variables. Here, the moderation effect will alternately be tested by using the median split method. This method holds that the data will be split according to a cut-off value (the median) following with logistic regression on both sets. When the effect of the independent variable is stronger in one group than in the other, it can be concluded that there is a moderation effect. As income level is not significant and it is coded as an ordinal variable with only 4 levels, income will not be included in the moderation analysis. The results of the moderation analysis of price consciousness are displayed in Table 21, 22 and 23

Low price consciousness

High price consciousness

-2 Loglikelihood 170,756 186,018

Cox & Snell R Square 0,209 0,206

Nagelkerke R Square 0,290 0,277

Hit-Rate 71,3% 72%

-2 Log likelihood

Cox & Snell R Square

Nagelkerke R

Square Hit-rate

377,332 0,164 0,223 70,4%

(29)

28

Constant -3,268 0,995 10,779*** 0,038

Price gap 0,11 0,029 14,395*** 1,116

Income -0,967 0,31 9,738*** 0,38

Price consciousness 1,198 0,293 16,673*** 3,312

Distance relative to service

retailer 0,081 0,036 5,138** 1,048

Table 22: Estimates under low price consciousness * Significant at 0,10 level

Constant -7,908 2,052 14,851*** 0

Price gap 0,144 0,029 25,247*** 1,155

Income 0,196 0,155 1,595 1,217

Price consciousness 0,949 0,316 9,007*** 2,583

Distance relative to service

retailer 0,028 0,028 1,081 1,029

Table 23: Estimates under high price consciousness * Significant at 0,10 level

As the price gap variable has a greater impact in the respondents with high price consciousness, there is evidence that price consciousness enhances the effect of the price gap on the probability that a consumer will switch from a national brand at a service retailer to a national brand at a discounter. A t-test is used to test whether the parameters of the price gap variable in both models are statistical significant according to the method used in Paternoster et al. (1998).

𝑡 = 𝑏1− 𝑏2 stdev(𝑏1− 𝑏2)=

𝑏1− 𝑏2 √var(𝑏1− 𝑏2)

The calculations require the coefficients and standard errors of the price gap variable in both models. The parameters are significantly different if the absolute value of the t-value is larger than 1,96 (p < 0,05). As there is no covariance between the variables because they are estimated in separate models, var(𝑏1− 𝑏2) can be computed by adding var(𝑏1) to var(𝑏2). This results in var(𝑏1− 𝑏2) = 0,0292_{+ 0,029}2_{= 0.001682. The t-value that follows is calculated as : 𝑡 =}0,11−0,144

(30)

|-29 2,62| exceeds the critical value of 1,96 meaning that the parameters are significantly different at the 95% confidence level. Therefore, the hypothesis that higher price consciousness enhances the effect of the between-store price gap on the probability of switching to a national brand at a discounter is confirmed (H4). Figure 9 shows how the percentage of switchers is higher under high price

consciousness.

Figure 9: Percentage of switchers at every price gap under high and low price consciousness (NB service retailer – NB discounter)

Results for the median split of distance relative to the service retailer are provided in Table 24. High relative distance means that the distance to the discounter, compared to the service retailer, is larger than under low relative distance.

High relative distance

Low relative distance

-2 Loglikelihood 179,602 194,544

Cox & Snell R Square 0,179 0,157

Nagelkerke R Square 0,246 0,211

Hit-Rate 72,0% 68,3%

Wald

Odds ratio

Wald

Odds ratio

Constant 22,928*** 0,013 6,195** 0,072

Price gap 18,354*** 1,128 18,564*** 1,125

Income 0,029 1,028 1,583 0,787

(31)

30 Distance relative to service retailer 0,02 1,005 1,184 1,079 Table 24: Estimation with median split for price consciousness

* Significant at 0,10 level **Significant at 0,05 level ***Significant at 0,01 level

The t-test shows a value of 0,120−0,118

√0,001405 = 0,06526 which is below the threshold of 1,96. Therefore, the results show no substantial difference of the effect of the between-store price gap on switching to a national brand at a discounter over the group with high relative distance compared to the group with low relative distance. Therefore, hypothesis 5, that shorter distance relative to the service retailer enhances the effect of the between-store price gap (NB SR – NB discounter) on switching to a national brand at a discounter, is rejected.

Relative importance of store choice factors

Relative importance of the independent variables will be calculating using the weights of the standardized betas. The standardized betas are calculated following the formula proposed by Menard (2011) :

𝛽∗ = 𝛽 ∗ 𝑠𝑑𝑥∗ 𝑅 𝑠𝑑𝑙𝑜𝑔𝑖𝑡(𝑌)

Where 𝛽∗ is the standardized beta, 𝛽 the unstandardized beta, 𝑠𝑑𝑥 the standard deviation of variable X, R the root of the cox & snell R square and 𝑠𝑑𝑙𝑜𝑔𝑖𝑡(𝑌) the standard deviation of the log transformed probabilities of Y=1. The relative importance of the variables is then calculated by dividing the standardized beta of each variable by the sum of the standardized betas. The findings are reported in Table 25: Relative importance of independent variables The price gap variable shows the highest relative importance with 49,20% followed by price consciousness, relative distance and income level.

Variable Unstandardized Beta Standard deviation Standardized Beta Relative Importance Pricegap 0,120 6,884 0,404 0,982 0,339 49,20% Price consciousness 0,387 1,37969 0,404 0,982 0,220 31,93% Income -0,059 1,095 0,404 0,982 -0,027 3,92% Relative distance 0,036 6,911 0,404 0,982 0,103 14,95%

(32)

(33)

32

5. Conclusions & Recommendations

Conclusion

This study examined the effects of the between-store price gap and within-store price gap on switching to national brand buying at discounters. The research question that was aimed to be answered was formulated as:

What is the optimal pricing strategy regarding the price gaps for national brands at discounters?

The answer to this question was determined by answering three sub-questions:

- How large should the between-store price gap be in order to make national brand buyers switch from service retailers to discounters?

- How large should the within-store price gap be in order to prevent buyers of store brands at discounters to switch to national brands at the same store?

- What is the relative importance of price compared to other factors influencing store choice behaviour for national brands?

These sub-questions are answered consecutively.

Within-store price gap

The within-store switchers belonged to the first group in Figure 1. Deleersnyder (2012) argued that this group is unprofitable to a discounter because of lower margins of national brands opposed to store brands. There were various findings. First, a higher within-store price gap has proven to deter undesired within-store switchers. Second, discounters do practically not have to reckon with store brand buyers switching to buying the national brand at the discounter as this group is quite low. shows that less than 15% of store brand buyers at discounters switch to a national brand at the lowest price gap (32 cents). If the discounter still believes this percentage is too high, the price gap could be increased to a minimum of 38 cents as the graph flattens at that point.

(34)

33 store brands indicated via an optional comment that they simply buy the cheapest milk or go to the closest chain as the quality of all brands is the same. Furthermore, Ailawadi et al., (2001) argue that store brand buyers are typically price conscious and less quality conscious and therefore show less desire for national brands. Another explanation is that store brand buyers are more loyal to a store and are therefore less likely to switch to another chain (Corstjens & Lal, 2000).

Between-store price gap

Between-store switchers belong to either group 2 or group 3 of Figure 1. Group 2 involves private label buyers from service retailers and group 3 national brand buyers at service retailers. Both groups are beneficial to a discounter because they formerly visited service retailers meaning that profits increase because of switching. Similar to store brand buyers at discounters, only a small fraction of store brand buyers at service retailers switch to a national brand at a discounter. In fact, Figure 10: Percentage of switchers for every price gap (PL – NB from service retailers to discounters and within discounters)shows that both groups more or less follow the same pattern at the various price gaps. As group 1 is harmful and group 2 beneficial to performance, these groups neutralize each other. This is under the assumptions that both groups are approximately equal in numbers and that store brands of both formats have about the same prices. Taking this into account, only the third group is relevant. Therefore, the within-store price gap can be set at a price that makes the between-store price gap optimal.

Figure 10: Percentage of switchers for every price gap (PL – NB from service retailers to discounters and within discounters)

(35)

34 Hoyer, & McAlister, 1998). Higher price gaps increase the probability that a consumer will buy the national brand at a discounter if it would be introduced there. This implicates that there are opportunities for the discounter to attract new customers by new national brand listings. Based on Figure 8, the recommended price gap for national brands at discounters is either just below the service retailers’ price (3 or 6 cents), or at 15 cents as the graph shows a sudden increase in switchers. The reason for this boost is not clear. There is no certain psychological threshold like €0,99 as all prices for the national brand were set above €1,-. Gupta & Cooper (1992) cconducted a study about the minimum threshold of a price promotion in order for a consumer to change buying intentions. They concluded that a discount of 10% is required for national brands and a discount of 14%-15% for store brands. The price difference at the 15 cent price gap in this research was displayed with a price of €1,29 which leads to a percentage gap of 11,6% (1,29− 1,14_1,29 ∗ 100%). This is in line with the findings of Gupta & Cooper (1992) as this result concerns national brands. Van Heerde et al., (2001) investigated the shape of the deal effect curve of price discounts and concluded that the curve is in an s-shape. This shape is also visible in the results in Figure 8.

Moderation effects

Price consciousness positively moderates the effect that the between-store price gap has on national brand buying at a discounter. This follows from the expectation that people who value low prices highly are more likely to switch to a discounter in order to get a better deal (Fox & Hoch, 2005). No moderation effects of income and relative distance were found. As discussed in the literature, income can have ambiguous effects. On the one hand, people who have higher incomes are likely to have higher educations which relates to making smart buying decisions (Baltas & Argouslidis, 2007). On the other hand people with lower incomes are more price sensitive and may therefore be more likely to switch to lower priced alternatives as well (Ailawadi, Neslin, & Gedenk, 2001) . Distance relative to the service retailer also showed no evidence of moderating the relationship between the between-store price gap and switching to a national brand at a discounter. However, there is some evidence that relative distance has a direct effect.. Especially the consumers with high price consciousness take a leap as is displayed in Figure 9: Percentage of switchers at every price gap under high and low price consciousness (NB service retailer – NB discounter).

Relative Importance

(36)

35

Limitations and Future Research

(37)

36

6. Bibliography

Ailawadi, K. L., & Harlam, B. (2004). An Empirical Analysis of the Determinants of Retail Margins: The Role of Store-Brand Share. Journal of Marketing, 68(1), 147-165.

Ailawadi, K. L., Neslin, S. A., & Gedenk, K. (2001). Pursuing the value-conscious consumer: store brands versus national brand promotions. Journal of Marketing, 65(1), 71-89.

Baker, J., Parasuraman, A., Grewal, D., & Voss, G. B. (2002). The Influence of Multiple Store Environment Cues on Perceived Merchandise Value and Patronage Intentions. Journal of Marketing, 66(2), 120-141.

Baltas, G. (1997). Determinants of store brand choice: a behavioral analysis. Journal of Product & Brand Management, 6(5), 315-324.

Baltas, G., & Argouslidis, P. C. (2007). Consumer characteristics and demand for store brands. International Journal of Retail & Distribution Management, 35(5), 328-341.

Baltas, G., Argouslidis, P. C., & Skarmeas, D. (2010). The Role of Customer Factors in Multiple Store Patronage: A Cost–Benefit Approach. Journal of Retailing, 86(1), 37-50.

Batra, R., & Sinha, I. (2000). Consumer-level factors moderating the success of private label brands. Journal of Retailing, 76(2), 175-191.

Becherer, R. C., & Richard, L. M. (1978). Self-Monitoring as a Moderating Variable in Consumer Behavior. Journal of Consumer Research, 5(3), 159-162.

Bellizzi, J. A., Krueckeberg, H. F., Hamilton, J. R., & Martin, W. S. (1981). Consumer Perceptions of National, Private, and Generic Brands. Journal of retailing, 57(4), 56-70.

Bontemps, C., Orozco, V., & Réquillart, V. (2008). Private Labels, National Brands and Food Prices. Review of Industrial Organization, 33(1), 1-22.

Briesch, R. A., Chintagunta, P. K., & Fox, E. J. (2009). How Does Assortment Affect Grocery Store Choice? Journal of Marketing Research, 46(2), 176-189.

Broniarczyk, S. M., Hoyer, W. D., & McAlister, L. (1998). Consumers' Perceptions of the Assortment Offered in a Grocery Category: The Impact of Item Reduction. Journal of Marketing Research, 35(2), 166-176.

Carpenter, J. M., & Moore, M. (2006). Consumer demographics, store attributes, and retail format choice in the US grocery market. International Journal of Retail & Distribution Management, 34(6), 434-452.

(38)

37 Child, P., Kilroy, T., & Naylor, J. (2015). Modern grocery and the emerging-market consumer: A

complicated courtship. Perspectives on retail and consumer goods.

Chintagunta, P. K. (2002). Investigating Category Pricing Behavior at a Retail Chain. Journal of Marketing Research, 39(2), 141-154.

Cleeren, K., Verboven, F., Dekimpe, M. G., & Gielens, K. (2010). Intra- and Interformat Competition Among Discounters. Marketing Science, 29(3), 456-473.

Colla, E. (2004). The Outlook for European Grocery Retailing: Competition and Format Development. The International Review of Retail, Distribution and Consumer Research, 14(1), 47-69.

Corstjens, M., & Lal, R. (2000). Building Store Loyalty Through Store Brands. Journal of Marketing Research, 37(3), 281-291.

Cotterill, R. W., & Putsis Jr., W. P. (2000). Market Share and Price Setting Behavior for Private Labels and National Brands. Review of Industrial Organization, 17(1), 17-39.

Cummins, S., Findlay, A., Petticrew, M., & Sparks, L. (2008). Retail-led regeneration and store-switching behaviour. Journal of Retailing and Consumer Services, 15(4), 288-295.

Cunningham, I. C., Hardy, A. P., & Imperia, G. (1982). Generic Brands versus National Brands and Store Brands. Journal of Advertising Research, 22(5), 25-32.

De Wulf, K., Odekerken-Schröder, G., Goedertier, F., & Van Ossel, G. (2005). Consumer perceptions of store brands versus national brands. Journal of Consumer Marketing, 22(4), 223-232.

Deleersnyder, B., & Koll, O. (2012). Destination discount: A sensible road for national brands? European Journal of Marketing, 46(9), 1150 - 1170.

Deleersnyder, B., Dekimpe, M. G., Steenkamp, J.-B. E., & Koll, O. (2007). Win–win strategies at discount stores. Journal of Retailing and Consumer Services, 14, 309-318.

Dhar, S. K., & Hoch, S. J. (1997). Why Store Brand Penetration Varies by Retailer. Marketing Science, 16(3), 208-227.

Donovan, R., Rossiter, J. R., Marcoolyn, G., & Nesdale, A. (1994). Store atmosphere and purchasing behavior. Journal of Retailing, 70(3), 283-294.

Fox, E. J., & Hoch, S. J. (2005). Cherry-Picking. Journal of Marketing, 69(1), 46-62.

Fox, E. J., Montgomery, A. L., & Lodish, L. M. (2004). Consumer Shopping and Spending across Retail Formats. The Journal of Business, 77(S2).

Fung Business Intelligence Centre. (2015). European Grocery Discounters Report. Gfk. (2015). Merk van het Jaar Verkiezing. Leusden.

(39)

38 Gupta, S., & Cooper, L. G. (1992, December). The Discounting of Discounts and Promotion

Thresholds. Journal of Consumer Research, 19(3), 401-411. Retrieved 01 12, 2018 Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2013). Multivariate Data Analysis. Pearson

Education Limited .

Hayes, A. F., & Matthes, J. (2009). Computational procedures for probing interactions in OLS and logistic regression: SPSS and SAS implementations. Behavior Research Methods, 41(3), 924-936.

Hosmer Jr., D. W., Lemeshow, S., & Sturdivant, R. (2013). Applied Logistic Regression (3 ed.). Wiley-Interscience.

Huff, D. L. (1964). Defining and estimating a trade area. Journal of Marketing, 28(3), 29-37. Kahn, B. E., & Lehmann, D. R. (1991). Modeling Choice Among Assortments. Journal of Retailing,

67(3).

Kraus, S., Ambos, T. C., Eggers, F., & Cesinger, B. (2015). Distance and perceptions of risk in internationalization decisions. Journal of Business Research, 68(7), 1501-1505.

Lamey, L. (2014). Hard economic times: a dream for discounters. European Journal of Marketing, 48(3/4), 641-656.

Leeflang, P. S., Wieringa, J. E., Bijmolt, T. H., & Pauwels, K. H. (2015). Modeling Markets: Analyzing Marketing Phenomena and Improving Marketing Decision Making. New York: Springer.

Lourenço, C. J., & Gijsbrechts, E. (2013). The impact of national brand introductions on

hard-discounter image and share-of-wallet. International Journal of Research in Marketing, 30(4), 368-382.

Lumpkin, J. R., Greenberg, B. A., & Goldstucker, J. L. (1985). Marketplace needs of the elderly: Determinant attributes and store choice. Journal of Retailing, 61(2), 75-105.

Malhotra, N. K., & Birks, D. F. (2009). Marketing Research. Pearson Education (US).

Martenson, R. (2007). Corporate brand image, satisfaction and store loyalty: A study of the store as a brand, store brands and manufacturer brands. International Journal of Retail & Distribution Management, 35(7), 544-555.

Maruyama, M., & Wu, L. (2014). Multiple store patronage: the effects of store characteristics. Journal of Retailing and Consumer Services, 21(4), 601-609.

Menard, S. (2011). Standards for Standardized Logistic Regression Coefficients. Social Forces, 89(4), 1409-1428.

Meredith, L., & Maki, D. (2001). Product cannibalization and the role of prices. Applied Economics, 33(14), 1785-1793.

Assessing the optimal price for national brands at discounters