Master Thesis
MSc. Marketing Intelligence
How to Seduce Customers with Discounters on the Lurk?
“Moderating Effects of Distance to Discounters on Assortment, Pricing and Promotional Strategies of Traditional Supermarkets: An Empirical Study”
Jamie Geluk S3820599 Riemsdijkstraat 1a 6701 BC, Wageningen jamiegeluk@outlook.com
+316 57278228
University of Groningen Faculty of Economics and Business
MSc. Marketing Intelligence
Completion date: 17-6-2019
Supervisor: Prof. Dr. L.M. Sloot Second supervisor: Dr. J.P.R. Thomassen
Abstract
The rise of the hard discount format in the Dutch grocery retailing market during the last decades has negatively affected the market shares of traditional supermarkets (TS). Dutch TSs have reacted in various offensive and defensive ways in order to win back their customers.
Their marketing reactions involve adjustments in assortment composition, pricing strategy and promotional strategy. Nevertheless, the effect of a hard discounter (HD) on the performance of a TS is found to be dependent on the distance to a HD, which suggests a local marketing approach. This study examines the effect of distance to a HD on the effectiveness of the marketing instruments of TSs in order to optimize marketing instrument reactions. Monthly data of 4419 customers, who (partly) shopped their groceries at the three biggest Dutch TSs over the past 24 months, is collected and combined with assortment, pricing and promotional data of five product categories. The results show evidence for heterogeneity in the effectiveness of marketing instruments used by TSs, depending on the distance to a HD. In particular, full- service operating TSs appear to be more affected in both direct effects on share-of-wallet and in the effectiveness of their marketing instruments, depending on the distance to a HD, compared to TSs with an ‘every day low prices’ strategy. The study shows the need for local marketing strategies and ends with managerial implications and future research suggestions.
Keywords: Food Retail, Discounters, Traditional Supermarkets, Distance Effect, Share-of- Wallet, Inter-Format Competition, Assortment Composition, Pricing Strategy, Promotional Strategy, Private-Label, National Brand.
Acknowledgement
The process of writing my master thesis started with a personal interest in the field of food retailing from the time when I was a young boy. This personal interest, combined with a curiosity in the possibilities of data science regarding marketing, has led to a preference to write my marketing intelligence thesis concerning food retailing under the supervision of Prof. Dr.
Laurens Sloot. Deepen myself in the field of food retailing for my thesis in the last months convinced me in my ambition to develop myself further in this area in the coming years.
Further, I would like to use this part of my thesis to thank those who played an important role in creating this thesis and those who provided me with support. First, I would like to thank Prof.
Dr. Laurens Sloot for his encouraging and constructive guidance. His outstanding expertise in the field of food retailing, combined with his social and enthusiastic character, inspired me to push the boundaries of my knowledge and analytical skills. Second, I would like to thank Marijn Bruggeman of EFMI Business School for his helpfulness in exploring the possibilities of the provided datasets. Third, I would like to thank EFMI Business School and Superscanner.nl for sharing their datasets, which are the foundations of all conducted analyses. Lastly, I would like to express my appreciation for the most important persons in my life: Koen, Marie, Peter, Alexandra and Iris. Family, and in particular grandparents Koen and Marie, thank you for all the guidance, care and support in life that could not be provided by my parents. This helped me to develop myself to the person I am now. Peter and Alexandra thank you; you are the proudest parents a son could ever wish for. Iris, your positivity and support during my study has been exceptional and made me succeed to achieve all the goals I set.
Table of Contents
Abstract ... 2
Acknowledgement ... 3
1. Introduction ... 5
2. Theoretical Background ... 8
2.1 Rise of the Hard Discount Format ... 8
2.2 Effect of Hard Discounters on Traditional Supermarkets ... 9
3. Conceptual Development ... 11
3.1 Assortment Composition ... 11
3.2 Pricing Strategy ... 14
3.3 Promotional Strategy ... 16
3.4 Control Variables ... 17
4. Methodology ... 18
4.1 Data Collection ... 18
4.2 Measurement ... 20
5. Results ... 22
5.1 Descriptive Statistics ... 22
5.1.1 Share-of-wallet ... 24
5.1.2 Distance ... 25
5.1.3 Assortment composition ... 31
5.1.4 Pricing strategy ... 34
5.1.5 Promotional strategy ... 36
5.2 Empirical Results ... 39
5.2.1 Pooled regression model ... 39
5.2.2 Partially pooled regression model ... 42
5.2.3 Unit by unit regression models ... 44
5.3 Additional Analyses ... 51
5.3.1 Competitors at near presence ... 51
5.3.2 Intra- and interformat competition effects ... 54
6. Conclusion and Discussion ... 56
7. Limitations and Future Research ... 59
7.1 Limitations ... 59
7.2 Future Research ... 61
References ... 62
1. Introduction
The first decades of this century in the field of grocery retailing are characterized by the rise of the, so-called, hard discounters. Hard discount (HD) is a grocery format that differs from traditional supermarkets (TSs) in the following way. They offer significantly fewer products (stock keeping units: SKUs), which mainly consist of private label (PL) products, for unmatched low prices. This HD format captured market share from the TSs and now has a noticeable position in the Dutch grocery setting in the past decades. In particular, German origin HDs Lidl and Aldi, which together own 916 stores in the Netherlands, account for 17.7%
market share (Distrifood, 2018). Their success has become a major source of concern for Dutch TSs (Cleeren, Verboven, Dekimpe & Gielens, 2009).
The increase in the market shares of Lidl and Aldi has been at the expense of TSs, such as Albert Heijn and Jumbo. Many previous studies affirmed that when a HD enters a market, TSs are highly affected and experience big sales losses. Contrary to this finding, other studies found that near presence of a HD has an ‘attraction effect’ and may encourage customers to engage in multiple store shopping patronage, resulting in a ‘twin location’ or ‘ideal neighbour’, suggesting a differentiated strategy for a TS (Gijsbrechts, Campo & Nisol, 2008; Vroegrijk, Gijsbrechts & Campo, 2013). Customers benefit from this, as they will take advantage of both store formats and engage in multiple-store shopping, which leads in a shift to share-of wallet (SOW) competition (Gijsbrechts et al., 2008). Moreover, research found that the distance to a HD plays a fundamental role in the way the presence of a HD affects the TSs (Vroegrijk et al., 2013). On the one hand, near presence has less effect, or may even have an attraction effect on customers, while on the other hand, a further distance to a HD has a stronger negative effect on mutual competition (Zhu, Singh & Dukes, 2006). This inverted U-shaped distance effect influences the optimal way a TS should react to the presence of a HD.
Knowing that the presence of a HD may lead to tremendous losses in sales has led to strong reactions of Dutch TSs in the past. These reactions varied from an increased focus on differentiation in service and quality, to substantial price reductions (Van Heerde, Gijsbrechts
& Pauwels, 2008). In 2003, the Dutch market leader Albert Heijn suffered from an enduring decrease in market share. They decided to permanently reduce the prices of thousands of products, resulting in a national price war between grocery retailers. The price war in Dutch grocery retailing made Albert Heijn succeed in countering their decreasing market share and
improving their price image, without harming their quality and service images (Van Heerde et al., 2008). An additional effect of the price war has been the increased price sensitivity of Dutch customers in grocery shopping. Ironically, this has led to improved performances of the HDs in the Netherlands, who saw their market shares increasing during the price war (Van Heerde et al., 2008). This implies that simply reducing prices does not have the intended effect on countering the rise of HDs. Instead of getting involved in price competition, other studies suggest traditional retailers to differentiate themselves from HDs, by providing a unique assortment or by offering frequent and deep price promotions (Zhu et al., 2006; Nielsen, 2007;
Cleeren et al., 2009).
Nowadays, the Dutch TSs Albert Heijn, Jumbo and Plus all use different defensive and offensive strategies to locally react on the presence of competitors. First, market leader Albert Heijn, which has a market share of 34.7% and operates as a full-service supermarket with a HiLo (High-Low) pricing strategy, provides consumers high-quality service for prices above average (EFMI Business School, 2011; Distrifood, 2018). As far as known, Albert Heijn is not involved in using local pricing, since they aim to differentiate themselves by offering service, quality and surprising offers (Marketingtribune, 2007). However, by introducing an economy private label and offering deep and frequent promotions, Albert Heijn nationwide reacts to the presence of HDs. Second, Jumbo is the runner-up in Dutch grocery retailing and has a market share of 19.1% (Distrifood, 2018). They operate as an ‘every day low prices’ supermarket with an above average service level, providing customers locally with a ‘lowest price guarantee’ for identical products, which urges Jumbo to use local pricing. However, since the assortment of HDs is dominated by unique private labels, this local pricing will barely affect them. Third, Plus is a smaller formula, which has a market share of 6.4% (Distrifood, 2018). They also operate as a full-service supermarket and locally use their Achilles-module, to be able to adjust their prices based on the competitiveness of the location. These three different Dutch supermarkets represent different TSs, since they use different strategies to react to local competitors.
Summarizing, previous research proposed several actions in the field of pricing, assortment and promotions to react to the presence of HDs. Moreover, the distance between a HD and a TS is crucial for using a local strategy of supermarkets. Dutch TSs seem to react on local competition by local pricing, assortment differentiation and promotions, but no research yet studied the effect of relative distance in order to identify appropriate marketing mix reactions. Therefore,
this study aims at answering the research question: “What is the effect of the distance to a HD in optimizing local reactions of TSs with marketing mix instruments assortment, price and promotions in order to maximize their share-of-wallet?”
From an academic perspective, this study contributes to the literature by examining the distance effect of HDs in the marketing mix reactions of TSs. As such, it fits into the gap in the literature that identified different distance effects of HD presence to a TS and the effects of proposed marketing mix reactions. As such, it fits into Gijsbrechts, Campo and Nisol’s (2008) call for more research on complementarity and substitutability of grocery retail formats and Cleeren, Verboven, Dekimpe and Gielens’ (2009) call for insights in the competition between HDs and TSs in the assortment composition, price dimension and promotional strategy. Besides, the findings of this study will help managers of TSs in making better marketing budget allocations among the marketing mix instruments in their competition with HDs.
The paper is organized as follows. In the next section, relevant literature about the presence, and distance effect of HDs and their corresponding appropriate marketing mix reactions will be discussed in the theoretical background. Based on this, a conceptual model will be developed, and hypothesis will be formulated. In the methodology, the used data and measurements of the variables will be discussed. In the results section, both descriptive statistics and empirical results will be presented and discussed. The paper concludes by answering the research question and discussing other findings of the study. Finally, the limitations of the study will be discussed, and future research suggestions will be provided.
2. Theoretical Background
In this chapter, literature about the rise of the HD format and the effect on TSs will be discussed.
2.1 Rise of the Hard Discount Format
The rise of the HD format has its origin in the German grocery retailing, where after the second World War small stores appeared with a limited assortment of PL products for low prices (Wortmann, 2004; Steenkamp & Sloot, 2018). These stores were named Aldi, which was derived from ‘Albrecht Discount’. Inspired by the success of Aldi, other retailers imitated this concept, of which Lidl has shown to be most successful. In 2015, Aldi and Lidl together accounted for approximately 37% of the grocery market share in Germany (AHDB, 2015). The HD format expanded beyond the German borders and gained ground in, for instance, other European countries and the United States (Wortmann, 2004; Steenkamp & Sloot, 2018). In particular, in the Dutch grocery market, Aldi and Lidl together account for 17.7% of the market share and respectively own 520 and 418 stores spread over the country (Distrifood, 2018).
Nowadays, the format of HDs in grocery retailing still rely on their principles from the past.
Their lean assortment only holds around 1000-1500 SKUs, which has small numbers of SKUs per category and predominantly consists of PL products whose quality is often as good as that of leading national brands (NB) (Colla, 2003; Steenkamp & Kumar, 2009; Cleeren et al., 2010).
However, despite this, Lidl increasingly adds NB products to their assortment to differentiate themselves and strengthen their competitive position (Deleersnyder, Dekimpe, Steenkamp &
Koll, 2007). Besides, HDs are barely involved in promotional or merchandising activities to increase their performance (Deleersnyder et al., 2007; Cleeren et al., 2010). Above all, the core of HD’s selling proposition is its focus on low prices (Steenkamp & Kumar, 2009). According to Colla (1994), their average price level can be considered to be approximately 15% to 30%
lower compared to the average domestic price for products of comparable quality. To offer low prices, HDs characteristically use a minimalistic and functional store design where consumers have to select their products from cardboard boxes, which are often stacked on pallets (Cleeren et al., 2010). The fact that HDs offer lower prices is the main criterion for consumers to choose for a HD (Colla, 1994; Lamey, 2014).
Earlier research held that HDs would especially be successful during economically hard times, such as post-war periods or economic crisis, because of their low prices (Kaas, 1994).
Moreover, it was believed that their success would decline in times of economic recovery.
However, Steenkamp and Kumar (2009) found that the market shares of Aldi and Lidl in Germany increased from 22% to 26% in the period of economic recovery from 2004 to 2007.
In other words, Steenkamp and Kumar found that (even wealthy) customers indeed become more price sensitive during recessions and therefore increasingly shop at HDs, but many of those customers keep returning for at least some of their purchases after the economic recovery.
Colla (1994) and Lamey (2014) affirm this, as they mention that HDs meet specific purchasing requirements by most customers, and therefore HD did not become popular only due to an economic crisis. This eye-opener increases the concerns of TSs about the presence of HDs.
2.2 Effect of Hard Discounters on Traditional Supermarkets
The rise of the HD format has been at the expense of TSs, such as Albert Heijn, Jumbo and Plus in the Netherlands. Previous research proved that the presence of HDs highly affects TSs, resulting in sales losses (Vroegrijk et al., 2013). Despite this, Vroegrijk and colleagues (2013) found that a HD is seldom the single store of choice for customers, which implies that customers engage in multiple-store shopping. This is confirmed by the finding that 61.9% of the customers are multiple-store shoppers, and even more remarkable: the combination of HD and TS shopping is by far the most common (Gijsbrechts et al., 2008). By purchasing at a HD as well as a TS, consumers can combine cost savings in price-sensitive categories at a discounter and purchase high-quality purchases, whereby choice variety and NB equity are more important, at a TS. Altogether, this leads TSs to shift from a share-of-customers to a share-of-wallet (SOW) competition (Gijsbrechts et al., 2008). The SOW in this study is defined as follows: “The share- of-wallet is the percentage of a customer’s total category expenditure captured by the firm”
(Du, Kamakura & Mela, 2007). Furthermore, the influence of distance to a HD is a crucial factor in the effect on the SOW of the TSs, as will explained next.
In contrary to what is expected, Zhu and colleagues (2006) found that a TS, which is near-by a HD, is better off than TSs located further away in terms of sales, which endorses the importance of the effect of distance. They explain this result as an indication that the near presence of a discounter draws away price-sensitive customers, causing TSs to focus on more profitable price-insensitive customers. Nevertheless, Vroegrijk and colleagues (2013) explained the near
presence of a HD differently. They conclude that a HD in the vicinity of a TS may turn the location in an attractive option for combined visits, which implies that a HD will attract additional clientele to the location who might also start visiting the TS. Besides, this, so called,
‘twin location’ leads to an ‘ideal neighbour advantage’, which enables customers to benefit from the advantages of both formats (Gijsbrechts et al., 2008). In the case of a mid-range distance or further distance to a HD, the effects for a TS are completely different (Vroegrijk et al., 2013). To underline this statement, Zu and colleagues (2006) found that entry of a HD further away from a TS leads to a 16% loss in sales and a 10% decrease in store traffic. In comparison, entry of a HD in the same shopping mall as a TS leads to a 4% loss in sales and a slight increase in-store visits. Therefore, the effect of HD presence on a TS depends on the distance to HD and is best explained through an inverted U-shaped distance effect (Vroegrijk et al., 2013). These differences in effect, depending on the distance to a HD, require different strategies for TSs. Literature proposed different defensive and offensive marketing reactions in the field of assortment composition, pricing strategy and promotional strategy, which will be discussed in the next chapter.
3. Conceptual Development
In this chapter, a conceptual model is developed based on proposed marketing mix reactions by literature in the field of assortment composition, pricing strategy and promotional strategy.
Moreover, the effect of distance to a HD is taken into account. Every sub-section highlights one of the marketing mix instruments, which leads to the formulation of the hypotheses of this study.
Figure 1. Conceptual model
3.1 Assortment Composition
Traditionally, assortment composition is a main driver of shopping decisions and is vital in solidifying affiliation with customers (Rhee & Bell, 2002; Carpenter & Moore, 2006; Vroegrijk et al., 2013). The two biggest distinctions in the field of assortment composition that can be made between HDs and TSs are the range and depth in categories offered and the share of PL products.
Firstly, the assortment of a HD holds around 1000-1500 SKUs with a small number of SKUs per category, compared to 15.000+ SKUs a TS holds and a larger number of SKUs per category (Colla, 2003; Steenkamp & Kumar, 2009; Cleeren et al., 2010). Conventional wisdom holds that a bigger assortment is more attractive for customers, since it increases the number of
Assortment composition
• Number of SKUs
• PL share o PPL share o SPL share o EPL share
• NB share
• Organic share
Pricing strategy
• Average price level o Average price level PL o Average price level NB
• Price dispersion
Promotional strategy
• Promotion share PL
• Promotion share NB
• Promotion depth PL
• Promotion depth NB
Distance to a HD
• Near presence
• Mid-range distance
• Further distance
Share-of-wallet TS
Control variables
• Age
• Income
• Size of household
• Education
• Distance to TS
options. Therefore, it is assumed that the positive effect of the number of SKUs on the SOW of a TS is not affected by the distance to a HD.
Second, the assortment HDs predominately consists of PL products, which is approximately 90% of their assortment (Colla, 2003; Steenkamp & Kumar, 2009; Cleeren et al., 2010). In contrast, TSs often use diversity in their assortment as a main selling proposition, by for example adding NB products to their assortment (Cleeren et al., 2010; Vroegrijk et al., 2013).
As NB are perceived as more attractive products, TSs can differentiate themselves from HDs which mainly offer PL products (Deleersnyder et al., 2007). Differentiation with assortment is repeatedly proposed as a strategy to compete against a HD (Zhu et al., 2006; Vroegrijk et al., 2013). Suggestions to do so vary from providing a more unique assortment with ethical and organic food, to introducing value-added PL products with a focus on quality (Zhu et al., 2006;
Vroegrijk, Gijsbrechts & Campo, 2016). Differentiation is especially recommended when a HD is located near a TS, since it drives complementarity (Gijsbrechts et al., 2008; Cleeren et al., 2010; Vroegrijk et al., 2013). Therefore, it is assumed that differentiation strategies as increasing NB share and adding ethical, organic or premium PL (PPL) products have more effect when a HD is nearly present than when at a mid-range or further distance.
The opposite strategy of acting complimentary is acting as a substitute of a HD. This strategy can be implemented by extending the assortment with standard PL (SPL) products to build store loyalty, and especially by economy PL (EPL) products to compete on price (Colla, 1994;
Vroegrijk et al., 2013; Vroegrijk et al., 2016). EPL products are products that are at very low prices and provide acceptable quality. A successful example of introducing with EPL products has been found in the United Kingdom, where TSs are very successful in competing with HDs by extending with an EPL product line (Zielke, 2010). However, the consequences of an EPL line has to be nuanced. On average, introducing EPL products in the absence of a near HD, does increase sales (Vroegrijk et al., 2016). In a case of the near presence of a HD, effects are contrasting. EPL products seem to be a useful instrument in near competition with a HD since it may preserve the SOW of its customers (Vroegrijk et al., 2016). In fact, research found that using this instrument cannibalizes current SPL sales and, moreover, may increase consumers’
focus on price and therefore drives them toward the lower priced HD (Chintagunta, Bonfrer &
Song, 2002; Geyskens, Gielens & Gijsbrechts, 2010; Vroegrijk et al., 2016). Therefore, it is assumed that the introduction of EPL products has a negative effect when confronted with a near HD and a positive effect when a HD is located at mid-range or further distance.
H1: The positive effect of the number of SKUs of a TS on its SOW is not moderated by the distance to a HD.
H2: The share of PPL products within an assortment of a TS has a more positive effect on its SOW when a HD is in near presence than when located at a mid-range or further distance.
H3: The positive effect of the share of SPL products within an assortment of a TS on its SOW is not moderated by the distance to a HD.
H4: The share of EPL products within an assortment of a TS has a negative effect on its SOW when a HD is in near presence and a positive effect when located at a mid-range or further distance.
H5: The share of organic products within an assortment of a TS has a more positive effect on its SOW when a HD is in near presence than when located at a mid-range or further distance.
H6: The share of NB products within an assortment of a TS has a more positive effect on its SOW when a HD is in near presence than when located at a mid-range or further distance.
Figure 2. Visualization of assortment composition hypotheses
SOW
Number of SKUs / Share of SPL products
Near presence Mid-range distance Further distance Main effect
SOW
Share of PPL / Organic / NB products
Near presence Mid-range distance Further distance Main effect
SOW
Share of EPL products
Near presence Mid-range distance Further distance Main effect
3.2 Pricing Strategy
As price is considered to be a main driver for shopping decisions, an often-proposed and straightforward strategy to compete against HDs is to substantially reduce prices (Ailawadi, Pauwels & Steenkamp, 2008). However, these nationwide price reductions may trigger price wars, as Albert Heijn initiated in the Dutch grocery market in 2003 (Van Heerde et al., 2008).
Nevertheless, by reducing the average price level, Albert Heijn successfully improved its price image, which directly resulted in an increased market share (Van Heerde et al., 2008).
Moreover, Van Heerde and colleagues (2008) found that the market share of HDs increased during the Dutch price war, due to the increased price-sensitivity of customers.
Another way to compete in the price dimension is by using local pricing strategies, which allows TSs to take into account the competitive environment of specific locations. Empirical research of Gomez-Lobo, Jiménez and Perdiguero (2015) in Gran Canaria showed that TSs lowered their price before near entry of a HD, and after entry increased their prices for products that were not sold at the HD (in other words: NB products) with 9% on average. In comparison, TSs located at a further distance (in their study located at 1.500 meters or more) did even increase the prices of both products that were sold at the HD. In addition, a similar study of Zhu and colleagues (2006) concerning the entry of a HD correspondingly shows that prices of products not offered by the HD on average increased, compared to pre-entry of the HD. This highlights the importance of the distinction of products offered by a HD and products not offered by a HD. Moreover, the finding of Gomez-Lobo and colleagues (2015) emphasizes the importance of heterogeneity in distance to select an appropriate pricing strategy. On the one hand, this implies that a near presence of a HD suggests a complementary strategy. On the other hand, it implies that a TS at a further distance to a HD is less affected by the HD and may even see opportunities to increase prices. As TSs at a mid-range distance do not benefit from a complementary strategy, Vroegrijk and colleagues (2013) suggest them to adopt a more defensive strategy, which may entail reducing prices or extending the assortment with an EPL line (Vroegrijk et al., 2016; Steenkamp & Sloot, 2018).
However, this EPL has its implications in the pricing strategy of a TS, since a low priced EPL extension leads to a greater price dispersion in the PL of a TS and their assortment as a whole.
In turn, this greater price dispersion may lower the price image of a TS (Alba, Broniarczyk, Shimp & Urbany, 1994). However, as mentioned in the sub-section about assortment
composition, may introducing EPL products lower the price image of a TS, but increases price sensitivity and drives customers to a HD when this is in the near presence (Chintagunta et al., 2002; Vroegrijk et al., 2016). Therefore, it is assumed that a greater price dispersion has a negative effect when a HD is in near presence, but a positive effect when a HD is at a mid- range or further distance.
H7: The average price level of PL products of a TS has a more negative effect on its SOW when a HD is in near presence than when located at a mid-range or further distance.
H8: The average price level of NB products of a TS has a less negative effect on its SOW when a HD is in near presence than when located at a mid-range or further distance.
H9: The price dispersion of a TS has a negative effect on its SOW when a HD is in near presence and a positive effect when located at a mid-range or further distance.
Figure 3. Visualization of pricing strategy hypotheses
SOW
Average price level of PL products
Near presence Mid-range distance Further distance Main effect
SOW
Average price level of NB products
Near presence Mid-range distance Further distance Main effect
SOW
Price dispersion
Near presence Mid-range distance Further distance Main effect
3.3 Promotional Strategy
HDs are characterized by their low prices and are known to use fewer promotional activities (Deleersnyder et al., 2007; Cleeren et al., 2010). By contrast, in particular TSs with HiLo pricing, are known for their promotional activities (Ellickson & Misra, 2008). These price discounts have been found to have a positive effect on customers purchase behaviour and thus on the SOW (Gilbert & Jackaria, 2002). However, both Rhee and Bell (2002) and Gijsbrechts and colleagues (2008) stress that temporary price promotions will only affect the short-term purchase decisions of customers, since customers will engage in cherry picking.
To compete with HDs, several studies recommended using deep and frequent price promotions, since this may attract customers from lower price segments (Rajiv, Dutta & Dhar, 2002;
Nielsen, 2007). In particular, Nielsen (2007) reports that customers are more likely to purchase the item with the deepest price cut, instead of simply purchasing the cheapest item. Especially, the depth of price promotions was mentioned to be crucial. Furthermore, Nielsen (2007) recommends conducting a well-planned and optimised promotional strategy.
In this study, a distinction is made between price promotions of PL products and NB products, since price promotions of NB products might have a stronger effect because they are perceived as higher quality and not offered by HDs (Deleersnyder et al., 2007). In line with the theory to differentiate when in near presence of a HD, it is assumed that share and depth of promotions of NB have more effect when near presence of a HD than when located at mid-range or further distance.
H10: The positive effect of the share of price promoted PL products within an assortment of a TS on its SOW is not moderated by the distance to a HD.
H11: The share of price promoted NB products within an assortment of a TS has a more positive effect on its SOW when a HD is in near presence than when located at a mid- range or further distance.
H12: The positive effect of the depth of price promotions of PL products within an assortment of a TS on its SOW is not moderated by the distance to a HD.
H13: The depth of price promotions of NB products within an assortment of a TS has a more positive effect on its SOW when a HD is in near presence than when located at a mid- range or further distance.
Figure 4. Visualization of promotional strategy hypotheses
3.4 Control Variables
Besides the before-mentioned variables, control variables are added to the analyses. These control variables are included to account for effects that can be controlled for. Controlling for the variables income, age and household size, education and distance to the TS provides for stronger testing of the hypotheses. The first control variable is age, which is included since a study found that older people are more price sensitive and may therefore prefer to shop at a HD (Blaylock, 1989; Hoch, Kim, Montgomery & Rossi, 1995). Second, a higher income leads to a decrease in price sensitivity (Ailawadi, Neslin & Gedenk, 2001). This may result in a lower preference for shopping groceries at HDs and a higher preference for TSs. Third, large households are to be more price sensitive than smaller households (Blaylock, 1989; Hoch et al., 1995). This could result in a higher preference to shop at a HD. Fourth, education has been found to have a significant effect on being a routine or random shopper. Consequently, routine shoppers have a preference for lower prices, which is likely to result in a preference to shop at a HD (Kim & Park, 1997). Lastly, conventional wisdom holds that distance to a shop is an important determinant in shopping decisions and therefore included as a control variable.
SOW
Promotion share of PL products / Promotion depth of PL products
Near presence Mid-range distance Further distance Main effect
SOW
Promotion share of NB products / Promotion depth of NB products Near presence Mid-range distance Further distance Main effect
4. Methodology
In this chapter, the data collection and the measurement of the variables will be discussed. In the sub-section data collection, the used datasets will be highlighted, as well as the choices for the traditional supermarkets and the assortment categories. In the sub-section measurement, the variables of the study and their corresponding measures are provided.
4.1 Data Collection
This study is conducted with data of the Dutch grocery retailing market, by using 4 different datasets from a period of 24 months from June 2017 until May 2019. For the dependent variable
‘SOW’ and the control variables ‘income’, ‘age’ and ‘size of household’, data from the EFMI Shopper Monitor was used, which is an extensive dataset based on monthly survey data of the main shopper of a household. In general, the data collection of EFMI takes place in the second week of a month. All data of respondents, that in the given timespan purchased (a part of) their groceries at an Albert Heijn, Jumbo, Plus, Aldi or Lidl, is included. In the 24 months covered by this study, 4419 respondents did their grocery shopping (partly) at Albert Heijn, Jumbo, Plus, Aldi or Lidl. Of these respondents, 1336 are male (30.2%) and 3083 are female (69.8%) and their age varies between 18 and 79 years old. In the 24 months covered by this study, 2764 respondents did their grocery shopping at Albert Heijn, 1883 respondents did their grocery shopping at Jumbo, 732 respondents did their grocery shopping at Plus, 1453 respondents did their grocery shopping at Aldi and 2004 respondents did their grocery shopping at Lidl. In addition, 750 respondents did their grocery shopping simultaneously at Aldi and Lidl and 4240 respondents were engaged in shopping at a TS and a HD in a month. The EFMI Shopping Monitor is provided with an extra variable ‘weighting’, which is used in this study to transform the SOW of a respondent in a weighted SOW. This weighted SOW improves the external validity, in particular the generalizability, of the study by making the respondents relative to the full Dutch population. Hence, it corrects for overpopulation and underpopulation of the sample of respondents.
For the moderating variable ‘distance to a HD’, two datasets were used. First, the EFMI Supermarket database was used to obtain the ZIP codes of the respondents. These ZIP codes are used to compute the distance of a respondent of EFMI Shopper Monitor to the TSs and HDs. Second, to take into account the nationwide heterogeneity in the degree of urbanity in areas of respondents, data of Centraal Bureau voor de Statistiek (CBS), the Dutch public
organization for statistical information, was used. They provide five categories for the degree of urbanization per ZIP code in the Netherlands. These categories, together with the distances of respondents to the TSs and HDs, have been used for cluster analysis to classify levels of urbanization in terms of the presence of the TSs and HDs. Correspondingly, the limits for their three levels of distances to a HD are calculated.
For the independent variables about assortment composition and pricing, data of five product categories from Superscanner.nl was used. The dataset of Superscanner.nl consists of daily obtained crawling data about assortment and prices. This data is aggregated on a monthly level to match with the intervals of the EFMI Shopper Monitor. The products in the assortment categories that are used in this study are the following: milk (dairy), mozzarella (cheese), apple juice (drinks), spaghetti (pasta) and laundry detergent (non-food). These products have been chosen since they cover important assortment categories and allow for variance in the different independent variables regarding assortment composition and pricing strategy.
The variables regarding the promotion strategy are computed with weekly promotion data, provided by EFMI Business School, covering all products of the TSs. Moreover, this weekly promotion data is appropriately aggregated on a monthly level. This data is not just aggregated for the month of the data collection. Instead, the promotional data is aggregated on the four weeks prior to the week of the data collection, to take into account that data collection of respondents takes place at the second week of the month and possible lag effects exist.
Table 1. Used datasets with the corresponding obtained information
Dataset Obtained information
EFMI Shopper Monitor SOW of respondents for Albert Heijn, Jumbo, Plus, Aldi and Lidl.
ZIP codes of respondents.
Control variables of respondents: income, age, size of household and education.
Superscanner.nl All assortment and price data of the five product categories for Albert Heijn, Jumbo and Plus aggregated on a monthly level.
EFMI Supermarket Database ZIP codes of all stores of Albert Heijn, Jumbo, Plus, Aldi and Lidl in the Netherlands.
CBS Degree of urbanization per ZIP code in the Netherlands.
EFMI Business School All promotion shares and promotion depth information of Albert Heijn, Jumbo and Plus on a weekly level.
This study contains data of three Dutch TSs which include Albert Heijn, Jumbo and Plus. These TSs are chosen because they are the biggest TSs in the Netherlands in terms of market share.
Firstly, Albert Heijn operates as a full-service supermarket and is the market leader with a market share of 34.7% and has 995 stores. It provides its customers with a big assortment for, on average, a higher price. However, they offer a lot of deep weekly price promotions compared to their competitors. Second, Jumbo operates as an ‘every day low prices’ supermarket and is the runner-up of the Dutch grocery retail market with a market share of 19.1% and has 618 stores. It has a mid-sized assortment and provides its customers locally a ‘lowest price guarantee’ for identical products. They do not offer a lot of weekly price promotions but offer changing seasonal price promotions. Finally, Plus operates as a full-service supermarket and is a smaller formula with 6.4% market share and has 260 stores. It offers its customers a mid- sized assortment, for a high price and offers weekly price promotions.
4.2 Measurement
The previous sub-section discussed how the data for this study is collected. This obtained data is measured and computed in various ways to use for analyses. The exact ways of how this data is measured and computed are provided in Table 2.
Table 2. Variables with corresponding measures
Variable Measure
SOW "Can you estimate what percentage of your total grocery purchases last month you purchased at which supermarket?"
Number of SKUs The number of SKUs in the five product categories in a month.
PL share The number of PL products in the five product categories divided by the number of SKUs in the five product categories in a month (in %).
PPL share The number of PPL products in the five product categories divided by the number of SKUs in the five product categories in a month (in %).
SPL share The number of SPL products in the five product categories divided by the number of SKUs in the five product categories in a month (in %).
EPL share The number of EPL products in the five product categories divided by the number of SKUs in the five product categories in a month (in %).
NB share The number of NB products in the five product categories divided by the number of SKUs in the five product categories in a month (in %).
Organic share The number of organic products in the five product categories divided by the number of SKUs in the five product categories in a month (in %).
Average price level Average price for all SKUs in the five product categories in a month (in €).
Average price level PL
Average price for all PL products in the five product categories in a month (in €).
Average price level NB
Average price for all NB products in the five product categories in a month (in €).
Price dispersion Average absolute deviation in the price for all SKUs in the five product categories in a month (in €).
Promotion share
PL Percentage of all PL products of a TS in price promotion in a month.
Promotion share
NB Percentage of all NB products of a TS in price promotion in a month.
Promotion depth PL
Average price promotion of all price promoted PL products of a TS in a month (in % of original price).
Promotion depth NB
Average price promotion of all price promoted NB products of a TS in a month (in % of original price).
Near presence* Dummy (0/1) for a distance of less than X meters to a HD.
Mid-range
distance* Dummy (0/1) for a distance between X meters and X meters to a HD.
Further distance* Dummy (0/1) for a distance of more than X to a HD.
* Distance X depends on the cluster for the urbanization level of a respondent (see Table 4).
5. Results
This chapter includes the descriptive statistics of the data and the empirical results. In the first sub-section, the descriptive statistics of the data are presented and clearly explained in tables, boxplots and graphs. In the second sub-section, the empirical results of the study are presented.
Lastly, some additional analyses are conducted, based on various discovered findings. For information, all analyses are conducted using the statistical R software version 1.1.456.
5.1 Descriptive Statistics
In this sub-section descriptive statistics, along with explorative charts are presented and discussed. Table 3 shows the means of the variables, together with their standard deviation and minimum and maximum value during the 24 months of data collection. Overall, the assortment, pricing and promotional variables show logical and explainable differences between the three different TSs.
The order in size of the SOW is logical and roughly corresponds to the market shares of the TSs in 2018: Albert Heijn 34.7%, Jumbo 19.1% and Plus 6.4%. The differences in the number of SKUs for the five different products are as expected. Albert Heijn offers the biggest assortment, while Jumbo and Plus rely on a smaller assortment. The differences in PL share are not big, but a notable difference is the smaller share of Jumbo. The subdivided shares of PL products show that Albert Heijn leads in PPL and SPL share, while Plus has the highest EPL share. The share of organic products is highest for Plus, which also has the most variance within.
Further, Albert Heijn and Jumbo almost have the same share of organic products. Comparing the average prices shows a representative view of the pricing strategy. On the one hand, Albert Heijn and Plus offer a comparable high price with some variation. On the other hand, Jumbo offers lower prices with less variation. These differences continue in the subdivided prices for PL products and NB products. However, the variation in the average price for PL products at Plus is notable. For the variable price dispersion, the high number for Albert Heijn is notable and could result from their big assortment and aim to have an assortment at every price level.
The promotion variables show differences in the promotion strategy between the TSs. Albert Heijn offers many and deep promotions. Besides, it concerns twice as much of a NB product compared to a PL product. Jumbo does not offer many promotions, and when they do it is less deep and equally focused on PL and NB products. Plus does not offer many promotions, but when they do it is a deep promotion and equally focused on PL and NB products.
Table 3. Mean values of the variables with standard deviation and minimum and maximum value Albert
Heijn Jumbo Plus
Mean
(SD) Min Max Mean
(SD) Min Max Mean
(SD) Min Max
SOW 30.23
(44.02) 0 100 18.43
(37.19) 0 100 5.61
(19.72) 0 100
Number of SKUs 143.8
(7.5) 128 153 105
(9.9) 87 117 99.4
(8.3) 79 107
PL share 38.83
(0.76) 37.50 40.00 27.61
(3.34) 22.77 32.74 34.95
(3.46) 34.95 38.68
PPL share 5.14
(0.25) 4.62 5.44 1.92
(0.19) 1.71 2.30 1.79
(0.50) 0.95 2.52
SPL share 29.18
(0.78) 27.89 30.07 24.82
(2.71) 20.79 29.20 28.06
(4.12) 18.60 32.08
EPL share 3.95
(0.35) 3.52 4.67 1.12
(0.81) 0.95 2.01 5.11
(0.50) 4.67 6.33
NB share 61.42
(1.23) 59.33 63.24 72.39
(3.34) 67.26 77.23 65.05
(3.46) 61.32 73.26
Organic share 13.03
(0.50) 12.24 13.85 12.42
(0.92) 10.89 14.16 13.25
(2.06) 11.21 17.72
Average price 12.05
(0.24) 11.59 12.38 10.92
(0.15) 10.68 11.22 11.74
(0.30) 11.39 12.45
Average price PL 9.09
(0.20) 8.68 9.37 8.09
(0.20) 7.87 8.56 8.28
(1.06) 7.09 10.61
Average price NB 13.54
(0.28) 12.98 13.98 11.72
(0.20) 11.35 12.05 13.43
(0.39) 12.73 14.06
Price dispersion 19.67
(0.42) 19.04 20.80 13.43
(0.78) 12.29 14.49 13.86
(0.52) 13.16 14.77
Promo share PL 6.02
(1.35) 4.29 9.74 7.51
(1.99) 4.13 10.22 4.64
(1.51) 2.86 7.85
Promo share NB 11.79
(1.65) 8.22 14.70 6.73
(0.87) 4.67 8.16 3.95
(0.28) 3.42 4.48
Promo depth PL 28.74
(1.71) 25.69 32.47 22.44
(1.51) 20.38 27.24 29.96
(2.73) 26.40 38.44
Promo depth NB 34.28
(1.52) 31.55 37.39 25.99
(1.14) 24.26 28.63 34.90
(1.69) 31.30 38.32
5.1.1 Share-of-wallet
To get a feel for the data, plots for the SOW are conducted to see the development of the dependent variable over time. Furthermore, several other ways to visually display the data are presented and explained.
As observed in Table 3, the SOW for the different TSs presented in Boxplot 1 is as expected and roughly corresponds to their market shares.
Besides, the SOW for the two HDs are included and show their substantial presence in the Dutch food retail market.
Boxplot 1. Scatter of the SOW for the TSs and HDs
This is confirmed by their market shares, since Lidl has a market share of 10.9% and Aldi a market share of 6.8% (Distrifood, 2018). Remarkable is the variance in the SOW, in particular of Albert Heijn and Lidl. This is an indication that over time there will a lot of changes in their SOW.
The development of SOW over time can be explored using explorative plots. Figure 5 shows this development during the period of 24 months. The development for Albert Heijn shows high variance over time, as expected from Boxplot 1. Besides it also shows high variance within the time. Furthermore, it shows declines in the begin and end of 2018 and a peak between July 2018 and October 2018. Although the development over time shows fewer extremes for Jumbo, the variance within the time period can still be considered as high. Furthermore, it shows slow growth in SOW. For Plus it can be seen that there exists little variance over time and within the time period. However, over the observed 24 months, Pus shows a gradual growth in SOW. The development of Aldi shows high variance within the time period and only little variance over time with a decline in the last few months. Lidl shows high variance over time and within the time period. It has its peaks at the start of both years which could come as a result of intended price sensitive behaviour. Moreover, the development in SOW for Lidl shows an increase over the two years.
Figure 5. Development of SOW over time for the TSs and HDs
5.1.2 Distance
To get an overview of how the distances to the different supermarkets are for the respondents, a boxplot and other figures are presented. The distance to a supermarket is considered as an important variable for grocery shopping choice, and therefore for the formation of SOW.
The distances to the supermarkets included in this study vary widely.
On average, Albert Heijn is most close to respondents, with only little variance. The distance to Jumbo, Aldi and Lidl seem to be equal and have approximately the same
Boxplot 2. Scatter of the distances to the TSs and HDs
variance. Plus, however, has a greater distance to the respondents, which also reflects the smaller number of shoppers at Plus and is likely to be a result of their small number of stores, which are often not located in highly urban areas.
A cluster analysis was conducted to take into account the heterogeneity of urbanization in different areas of the Netherlands. This cluster analysis is performed with the use of the five levels of urbanization per postcode by CBS and the distances to the supermarkets for the respondents. The results are presented in Table 4 and show the different distance limits for categorizing the variables ‘near presence of a HD’, ‘mid-range distance to a HD’ and ‘further distance to a HD’. These limits are based on the quartile distances of the aggregated distances to all supermarkets per cluster. The results show clear distinguished differences between the clusters, and therefore for the level of urbanization of their place of residence. The first cluster is the largest and the respondents can be described as living in urban areas. The distances for the second cluster are more than doubled compared to the first cluster. This cluster can be described as living in semi-urban areas. The third cluster has high distances to the supermarkets and the respondents are described as living in rural areas.
Table 4. Distance limits for the three different urbanization clusters
Near presence Mid-range distance Further distance
Cluster 1 (n = 3224) <1302 1302-2356 >2356
Cluster 2 (n = 661) <2925 2925-5139 >5139
Cluster 3 (n = 534) <4822 4822-7022 >7022
Vroegrijk and colleagues (2013) found that the effect of HD presence on TSs depends on the distance to a HD and could best be explained through an inverted U-shaped distance effect. In contrary to what Vroegrijk and colleagues (2013) stated, Boxplot 3 shows different Boxplot 3. Effect of distance to HDs on the aggregated SOW of TSs
results. It appears that near presence of a HD is most harmful to the SOW of TSs. Furthermore, mid-range distance to a HD is less harmful, and a further distance to a HD even slightly less harmful than mid-range distance, which is not in line with the research of Vroegrijk and colleagues (2013).
However, Boxplot 3 shows the aggregated SOW for all TSs, but the effect of distance to a HD may be different for the three different TSs in this study, since they do not all operate with the same strategy. Boxplot 4 shows the differences between the three TSs individually and indeed shows differences between the effect between the TSs. For Albert Heijn, it appears that near presence of a HD is most harmful, while a HD mid-range distance of further distance is equally harmful. For Jumbo the effect on SOW is different. Near presence and mid-range distance appear to be equally harmful, while a HD at a further distance is less harmful. This could be a result of the less differentiated way Jumbo operates as a supermarket compared to HDs, with lower prices. The effect of distance for Plus is less clearly, since they have a lower SOW by default. However, a HD at a further distance of Plus has unmistakably less effect on its SOW.
Boxplot 4. Effect of distance to closest HD on the SOW of TSs
To clarify the results of Boxplot 4 and get a better impression on the effect on SOW, plots are conducted. Figure 6 shows the relation of the distances to the closest HD on the individual TSs.
For Albert Heijn, a near distance to a HD is most harmful and their SOW increases gradually until approximately 3.000 meters. Jumbo is harmed as well when having a near distance to a HD and seems to have a small peak in SOW with a HD at a distance of around 1.000 meters.
Surprisingly, Plus its SOW is higher at a very near distance and lowers a bit around 500 meters and from there on increases gradually. However, the spread of the standard error shows a lot of variance around the very near distance, which makes it hard to draw conclusions. Concluding, the effect of distance to a HD seems to be different for the three TSs, but in general, contrary to literature is a near distance most harmful, after which the SOW increases.
Figure 6. Effect of distance to closest HD on SOW of TSs
Nonetheless, there may exist differences in the effect between the presence of Aldi and Lidl.
As known, Lidl and Aldi increasingly differ from each other, whereby Lidl increasingly focusses on quality may therefore be more or less harmful for TSs than Aldi (Deleersnyder et al., 2007). To explore these potential differences, Boxplot 5 is conducted. The boxplot confirms the potential differences between Aldi and Lidl. Lidl seems to be more harmful when present near-by and less when present at a mid-range distance or further distance. Aldi also harms TSs SOW, but less than Lidl. However, the presence of Aldi at a mid-range distance still seems to affect the SOW of TSs. This apparent difference between the effects of Aldi and Lidl leads to further inspection of the exact differences in effect on SOW of TSs.
