Masters Thesis
Msc. Marketing Intelligence
How Smart is Local Pricing?
The Impact of Local Pricing and Competitive Environment on the Price Image and Share-of-Wallet of a Locally Priced Retailer
Daniël Logt S3854752 Slachthuisstraat 96 9713ME, Groningen Daniel_logt@live.nl +31 640174351 University of Groningen Faculty of Economics and Business
MSc. Marketing Intelligence
Date: 05-04-2020
1
Abstract
2
Acknowledgements
3
Table of Contents
Introduction ... 4
2. Conceptual Development ... 5
2.1 Price Image and Share-of-Wallet ... 5
2.2 Pricing ... 5
2.1.1 Average Price Level ... 6
4
Introduction
Have you ever been to a petrol station, gassed up your car, paid and made your way just to find a gas station ten minutes further to display prices nearly 25% cheaper than the establishment you just visited? It’s common practice in petrol retailing to adhere prices to the local environment, but since traveling by car allows one to hop from one local environment to the next, the price discrepancies become more and more vivid. It’s one of those examples that comes to mind most easily when talking about local pricing. The general opinion on the matter seems to be of dishonest or even immoral operational management, as management is employing some form of market or monopoly power (Davidson, 2010). However, price discrimination, and local pricing more specifically, is not illegal in most western countries and the concept has therefore woven itself into more businesses than the exemplary petrol case mentioned before.
Retail chains essentially practice one of two broad strategies in setting prices across their stores. The more straightforward is to set a chain- or country- wide price. Alternatively, managers of retail chains may customize prices to the store level according to local demand and competitive conditions (Dobson & Waterson, 2008). Remarkably, uniform pricing has become the dominant strategy and market leaders in the UK (Tesco) and the Netherlands (AH), for instance, have chosen to roll out prices nationally (CC, 2003; SuperScanner, 2020). In the current literature, however, there does not seem to be a consensus on the matter of the effectiveness of a local versus a national pricing strategy. On the one hand it’s suggested empirically that retailers leave a significant amount of profit margins untouched when maintaining a uniform pricing strategy. (Dellavigna & Gentzkow, 2019) On the other hand a local pricing strategy could detrimentally impact a chain’s image and uniform pricing softens nearby competition and raises overall profits. (Dobson & Waterson, 2008)
The current literature focuses on business output as the indicatory variables of success but leaves customer sentiment unaccounted for. It is, however, of great interest to investigate if the customers’ opinion about stores, and more specifically stores’ prices, differentiate in the same way that their prices differentiate, which has yet to be researched. This study revolves around locally priced retailers and the impact of their pricing levels and competitive
environment on their price image and Share-of-Wallet. From an academic point of view it could be interesting to see whether price is still the main driver of price image at these
locations, as traditionally has been the case (Feichtinger, 1988). Managerially, it’s interesting to see if an X% increase in prices lead to the same increase in price image or in what
5
2. Conceptual Development
In this first section the individual constructs are defined and provided with relevant information and bodies of literature so as to get a clear view of the current perspectives on price image and it’s main drivers. Additionally, the main hypotheses of the current research are proposed. Lastly, a conceptual model will try to visualize the individual constructs and in what way they’re hypothesized to be linked.
2.1 Price Image and Share-of-Wallet
A common misperception about price image is that it is simply a reflection of a store’s average price level and, thus, that managing price image merely involves managing prices (Hamilton & Chernev, 2013). Indeed, consumers become more and more informed about their shopping choices and can compare prices of retailers more easily nowadays, and thus, the discrepancy between a store’s prices and it’s price image becomes less wide. However, despite increased availability of price information, the price image of many retailers does not always accurately represent their actual prices. For example, even though Target’s prices on many items are lower than Wal-Mart’s (Kavilanz, 2011), many consumers hold a consistently lower price image of Wal-Mart than of Target. These findings instigate a notion that price image is not solely driven by prices and as such, other factors come into play when a
consumers evaluates a retailer on whether it is expensive or not. While the price of a certain product is an inherently objective construct, the view of whether this price is relatively expensive or not is purely subjective. This research tries to explain the price image of a locally priced supermarket by its prices as well as its competitive environment.
In an article by Van Heerde and colleagues, they empirically proved that price image affects store choice and spending patterns (Van Heerde, 2008). However, this study focused on supermarkets with a national pricing strategy and as such, differences in price image of separate stores and their relation to share of wallet were not investigated. The share-of-wallet (SOW) is the percentage of a customer’s total category expenditure captured by the firm. If significant differences exist in the price image of the differently priced supermarkets, it is of course of great interest what effect this has on the share of wallet of the supermarkets.
2.2 Pricing
6 policy of a store under consideration. As the scope of this research is limited to one focal retailer, Jumbo, the majority of the nonprice factors can be considered to be generally equal across stores and said factors are therefore not taken into account when explaining price image in this analysis. Nonprice factors such as the size of the store are controlled for, however.
The price-related factors Hamilton and Cherney found to have an effect on price image, were segmented into five variables: average price level, dispersion of prices, price dynamics, price-related policies and price-related communications. The average price level can traditionally be seen as the main driver of the price-related factors (Feichtinger, 1988). It reflects how the prices of one retailer compare to that of another. The dispersion of prices entail the manner in which a retailers’ prices are competitive across categories. Inclusion of this variable accounts for the notion that consumers may form category-specific price images as one supermarket may, for instance, have an inexpensive bakery but expensive fresh products and another supermarket vice versa.
Price dynamics includes the fact that one supermarket may have quite static pricing whilst another may change their prices drastically over time. However, since our retailer of interest, Jumbo, follows a companywide Every Day Low Pricing (EDLP) strategy and the main price promotions are rolled out nationally, this variable is not taken into account in the current research as it is assumed not to vary across stores. The same goes for price-related policies and price-related communications, which involve the company wide policies towards prices (EDLP in the current case) and its mass marketing efforts, respectively. It can be assumed that these variables don’t vary across stores and therefore they are omitted of the current research.
2.1.1 Average Price Level
As mentioned before, price is an inherently objective construct and therefore,
7 Moreover, the way an average price level is construed is different from one person to the next. The vast amount of individually priced SKU’s, daily price changes and special promotions make a comprehensive evaluation of the average price level of a store impossible for the individual consumer. This leads them to make use of reference points; consumers are likely to use heuristics based on only a subset of the total possible information. (Desai & Talukdar, 2003) It must be acknowledged that the perception of the average price level is therefore malleable and susceptible to manipulation of the reference point (Alba, Broniarczyk, Shimp, & Urbany, 1994) To be able to measure the average price level in a valid manner, it is therefore wise to make use of the same reference points customers use as these exemplifying SKU’s will mostly, if not solely, influence the overall price image of the stores.
Based on a large body of literature on consumer behaviour and psychology (Alba & Hutchinson, 1987; Alba, Hutchinson, & Lynch, 1991; Berelson & Steiner, 1964) products influence the construct of the overall store price image most when excelling on the axes of two variables; consumption time span and unit price. Products having a short consumption span and high unit price (e.g. beer, ground coffee, shampoo) were found to have the most influence on the price image of a store, followed by products having a short consumption span and low unit price (e.g. toilet paper, potato chips, soda), products having a long consumption span and high unit price (e.g laundry detergent, batteries, shaving blades), and lastly products having a long consumption span and low unit price (e.g. light bulbs, tootbrush, trash bags) (Desai & Talukdar, 2003)The frequential encounter with a product and high score on the price dimension make short spanned/high priced products more easily recalled on when such information is needed and it makes the products store into the long term memory. Both of these effects facilitate more salience in the evaluation of an overall store price image. This information will be used to carefully construct a product index to use in the basket analysis so as to most accurately predict the price image.
H1: Average Price Level is not correlated with Price Image. H2: Average Price Level is not correlated with Share-of-Wallet.
2.1.2 Price Dispersion
8 Conventional wisdom would say that an upscale price extension would lead to a
higher price image whilst a downscale price extension would lead to a lower price image (Hamilton & Chernev, 2010). However, this effect is plausibly mediated by the directionally consistent shift in the average price level of a store’s product line, disintegrating the main effect of the dispersion itself. Because more frequently bought stockkeeping units
disproportionally affect price image formation (Sprott, Manning, & Miyazaki, 2003),
increasing variance around the products of purchase will likely result in increased uncertainty of choice within the categories and retailer as a whole. This may leave shoppers doubtful with regard to whether or not their purchase was a good trade (Iyengar & Lepper, 2000). Increased price dispersion is therefore hypothesized to detrimentally affect price image.
The dispersion of prices across product categories and the resulting
category-dependent price image may be one cause of cherry-picking behavior on the part of consumers (Fox and Hoch 2005). As prices within a supermarket get more dispersed, nearby competition is likely to be more competitive on at least one product-category. This instigates a notion that price dispersion negatively affects Share-of-Wallet as higher deviation might lead to category-specific shopping.
H3: Price Dispersion is negatively correlated with Price Image. H4: Price Dispersion is negatively correlated with Share-of-Wallet.
2.3
Competitive Environment
As a retailer is at least spatially differentiated from all other retailers, it must be acknowledged that a supermarket has a monopolistic position in the market to some extent. The most intense price competition for a given grocery store comes from stores offering the same array of goods in the same trading area (Cassady Jr., 1962). Because the subject under investigation prices locally, accounting for the spatial competition is of crucial importance in this study on price image. The spatial dimension has an important impact on pricing strategies in at least two ways: demographics and competition (Willart, 2014). For big supermarkets such as Albert Heijn or Plus, the competition component is of less interest when delineating their price image, as they (primarily) price nationally and if price leadership in a certain area or product category falls short, they can rely on their high service levels or spatial prominence to attract customers. It must be noted that Plus does have a few (<50) stores with a lower average price level (the so-called Achilles stores) but not much data is available on these stores. Albert Heijn does price differently at each of its formats (AH to Go, AH XL), but there the differentiated prices are not caused by an underlying geographical or demographical mechanism. As the subject of interest, Jumbo, prices locally, however, the local competitive environment is of interest.
9 competitive environment and the price image of a locally priced retailer, which has yet to be researched. It’s highly plausible that a supermarket is viewed as relatively expensive when there is a multitude of Hard Discounters (HDs) nearby while a supermarket maintaining the same price levels has an inexpensive price image when it’s surrounded by Traditional
Supermarkets (TSs). The competitive environment is measured in the amount of competitors (supermarkets) and the type of competitors (TS/HD) nearby (<1000m) and the link between competitive environment and price image of a locally priced supermarket is assumed to be moderated by the distance between the supermarket of interest and it’s competitors.
2.2.1 Type of Competition
Over the first decade of the twenty-first century, one of the most striking
developments in grocery retailing is the rise of the so-called hard-discounter (HD) format (Vroegrijk, Gijsbrechts, & Campo., 2013; Van Heerde, 2008). In the Netherlands alone, two of the top five supermarket chains consist of hard discounters, namely Aldi & Lidl
(Distrifood, 2020). Typically, these chains offer much smaller assortments of products and are characterized by their low service levels; there’s less personnel on the shop floor and products are stacked in the card-boxed packages they came in. Most noticeably, HDs offer fewer, or no, national brands (NBs) than traditional supermarkets (TSs); all so as to be able to offer drastically lower prices whilst upholding a profit margin. Prices at HDs can be up to 60% lower than those of leading NBs and as close as 40% lower than the prices of private-label products owned by traditional supermarkets (Cleeren, Verboven, Dekimpe, & Gielens, 2010). Paradoxically, however, these chains attract customers in low income and high income classes alike.
On the other hand, traditional supermarkets may face competition through retailers offering the same service levels and same assortments for relatively the same prices. Given the substantial differences in positioning, assortment composition, pricing, and store environment, Cleeren et al. (2010) expected the extent of interformat competition to be smaller than the extent of intraformat competition. In other words, we expect the nearby presence of traditional supermarkets to have a greater negative effect on Share-of-Wallet than the presence of Hard Discounters. However, nearby interformat competition is expected to have a greater effect on price image of a locally priced retailer than do intraformat
competition; the crucial difference laying, ofcourse, in the dependent variable. It’s very likely that a store is viewed as relatively expensive when there’s a multitude of HDs nearby.
However, when a locally priced retailer is surrounded by traditional supermarkets, the hypothesis is that customers view the retailer of focus as inexpensive relative to the competition.
H5: Intraformat competition is expected to have a greater negative effect on Share-of-Wallet than Interformat competition
10
2.2.2 Amount of Competition
For a retailer to survive in the long-run, it has to somehow be able to attract customers in it’s nearby trading area. Moreover, for any given trading area there is a limit as to the amount of customers a supermarket is able to attract; how big the piece of the pie one is able to eat. Following this line of reasoning, the Share-of-Wallet will linearly decrease as the amount of competition increases. However, since our retailer of focus is locally priced, price image is hypothesized not to be affected by the amount of competitors in the nearby trading area. Generally, a locally priced retailer will by definition fall on the lowest end of a price spectrum. The relativity component of price image therefore does not come into play and price image should theoretically not be affected, however many competitors enter the trading area under investigation.
H7: Amount of Competition is negatively correlated with Share-of-Wallet H8: Amount of Competition is not correlated with Price Image
2.2.3 Distance to Competition
Distance is a major factor in transactional utility (Bell & Tang, 1998). This is one of the key antecedents of the aforementioned monopolistic position a supermarket enjoys; consumers tend to trade in monetary gains for a decrease in traveling distance relatively quickly (Hoch, Kim, Montgomery, & Rossi, 1995). Following this reasoning, intuitively, one would say that the closer the competition, the higher the magnitude of the effects felt of said competition because the overlap in households served is the greatest with close proximity of both parties. Contrarily, Vroegrijk et al. (2013) found an inverted U-shaped effect of distance with the entry of an HD; losses to competitive Hard Discounters were greatest for the
incumbent when the distance was of intermediate proportion. Close proximity gave way to one-stop shopping where the HD and TS compliment one another in their serving to the neighborhood. The distance between the focal retailer and it’s competition is therefore expected to heterogeneously affect Share-of-Wallet, where the distance to a competitor TS is expected to be linearly, positively correlated with Share-of-Wallet (of Jumbo) whilst the distance to a nearby Hard Discounter is hypothesized to form a U-shaped effect.
H9: When there is a Hard Discounter nearby, the effect of Distance to said HD on Share-of-Wallet of the focal retailer will form a U-shape.
H10: When there is a Traditional Supermarket nearby, the effect of Distance to said TS on Share-of-Wallet of the focal retailer is positive and linear.
H11: The negative effect of interformat competition on price image is strengthened when said competition is closerby.
11
2.4
Assortment composition
Another major store-related component able to drive the overall store price image is the composition of the assortment. Scholars have been able to conjecture price image on the basis of the size of the assortment: the amount of individual stock-keeping units (SKUs) a retailer holds (Chernev, 2003; Chernev & Hamilton, 2009; Iyengar & Lepper, 2000). The theory underlining this finding, is that retailers holding large assortments are able to enjoy economies of scale and are therefore able to offer lower prices than their smaller-sized counterparts (e.g. convenience stores) (Hamilton & Chernev, 2013).
Similarly to size, the distribution of the assortment is assumed to drive the price image of a store. To elucidate; the shares of NBs versus PLs a store sells is expected to be a proxy of the price level a store holds and consequently be able to drive the price image. Moreover, the share of national brands and the share of private labels are hypothesized to be positively and negatively correlated with price image, respectively, but plausibly mediated by the average price level a retailer holds.
H13: The amount of SKU’s a retailer holds is positively correlated with their price image.
H14: The share of national brands in a retailers’ assortment is negatively correlated with their price image.
H15: The share of private labels in a retailers’ assortment is positively correlated with their price image.
H16: The effects of the share of NBs/PLs in a retailers’ assortment is mediated by the average price level a retailer holds.
2.5
Control Variables
As mentioned before, a large set of variables are already controlled for, such as mass media efforts, promotional activities and general service levels, because this study focuses on one national retailer and therefore aforementioned variables can be assumed to be equally projected in the price image across the individual stores. There are, however, some
demographic variables and store characteristics which could influence the price image of the individual stores, but are not of primary interest in this study. It is important to control for these influences as they could otherwise tamper with the results.
12 Chernev, 2013) as it can be a signal of large sale volumes for which volume discounts could be passed on to the customer.
Bare in mind that some of these control variables might cause collinearity issues. For example, the size of the store and the amount of stock keeping units a store sells might very well be part of the same latent factor and be linearly correlated. In the same vein: age,
education level and income are plausible causes for multicollinearity issues. Regardless, such issues can later be resolved; for now it’s important to create the most comprehensive model
13
3. Methodology
This section is devoted to the methodology in this study. First, the datasets used to gather the information on which this empirical study is built are discussed, after which each individual variable and their operationalizations are outlined to shed light on the way each construct was measured.
3.1
Data Collection
This research is conducted with support of data on the Dutch grocery retailing market covering a period of 31 months from June 2017 until December 2019. The pricing variables, as well as the assortment composition variables are gathered through data by Superscanner.nl. The dataset of Superscanner.nl contains daily obtained ‘crawling’ data on all individual stock-keeping units for each individual grocery store within the Netherlands. All data is aggregated on a monthly level to match the aggregation levels of other datasets used.
Information on the competitive environment of each individual Jumbo is gathered through the use of the EFMI Supermarket Database. This database contains information on the location of each individual grocery store within the Netherlands, as well as some
information about the store characteristics. Spatial data is calculated with the use of the ZIP codes of each individual store.
Data concerning the dependent variables, as well as the control variables age, income, education level and household size comes from the EFMI Shopper Monitor. The EFMI Shopper Monitor is an extensive dataset based on monthly surveys held within the population of the Netherlands concerning their grocery shopping behavior and attitudes. Because there is only one retailer within the top five biggest supermarket retailers in the Netherlands
(Distrifood, 2020) which prices locally, Jumbo, only those respondents who purchased (a share of) their groceries at Jumbo during a given timeframe are included.
3.2
Measurement
14
Type Variable Operationalization Dataset Min Mean Max
Dependent Price Image “Assume a full shopping cart of groceries will cost €100,-. Some are more expensive; some are less expensive. What will that shopping cart cost, according to you, at Jumbo?”
EFMI Shopper Monitor
35 98 (19,7) 250
Share-of-Wallet "Can you estimate what percentage of your total grocery purchases last month you purchased at Jumbo?" EFMI Shopper Monitor 0 40,7 (32,4) 100
Pricing Average Price Level
Average price for all SKUs in the product categories in a month (in €)
Superscanner.nl 2,13 2,20 (0,1)
2,30
Price Dispersion Average absolute deviation in the price for all SKUs in the product categories in a month (in €)
Superscanner.nl 1,03 1,07 (0) 1,12
Assortment Composition
Amount of SKUs The number of SKUs in the product categories in a month
Superscanner.nl 17749 18780 (726,7)
21946
Share of PLs The number of PL products in the product categories divided by the number of SKUs in the product categories in a month (in %)
Superscanner.nl 24,69 25,40 (0,3)
26,73
Share of NBs The number of NB products in the product categories divided by the number of SKUs in the product categories in a month (in %)
Superscanner.nl 73,27 74,6 (0,4)
75,31
Competition Type of Competition
Dummy variables for the presence of a competitor HD (1) or TS (2) as one of the two closest supermarkets from the respondent.
EFMI Supermarket Database 0 = No; 1 = Yes Amount of Competitors
The total number of supermarkets (HD & TS) with which the focal retailer has to compete in an area of 1000 meters around the respondent. EFMI Supermarket Database 0 3,63 (3,7) 31 Spatial Distance to Competition
Distance between the postcodes of the respondent and each supermarket in meters. EFMI Supermarket Database 22,83 1124,41 (919,70) 9260,42
Control Gender Gender of the respondent EFMI Shopper Monitor
1 = Male; 2 = Female Age Age of the respondent EFMI Shopper
Monitor
18 47,99 72 Education Level Education level of the
respondent
EFMI Shopper Monitor
1 = No education; …; 8 = Academic level
Household Size Household size of the respondent
EFMI Shopper Monitor
1 2,19 10 Income Income of the respondent EFMI Shopper
Monitor
1 = <€1000,-; …; 7 = >€6000,- Store Size Store size of the closest
Jumbo from the respondent in km².
EFMI Supermarket Database
493 1363 3350
15
4. Descriptive Statistics
In this section various descriptive statistics are outlined so as to get a clear view of the data and the relationships between different variables.
4.1 Preliminary Analysis
As mentioned, the price image of Jumbo was measured by asking how much a full shopping cart costs at Jumbo, given that on average a full shopping cart costs €100,-. Preliminary analysis shows respondents estimating the costs as high as €140130,-, which is not only a statistical outlier but also a illogical deviation from the mean. Removing outliers by rule of thumb (first quartile – 1,5*IQR; third quartile + 1,5*IQR) meant omitting any
responses under €73,50 or above €118,-, which would be too critical. Therefore, it’s rather arbitrarily chosen to omit any responses that forgo the confines of being three times cheaper (<€33,-) or more expensive (>€300,-) than the hypothetical market average, omitting 112 of the total of 3343 observations.
4.2 Pricing
Figure 2 displays the boxplots of the price images of the focal traditional supermarkets and hard discounters to form a first view of the manner in which consumers perceive their prices. In alignment of common perceptions, Albert Heijn scores highest with a median of €110, followed by Plus, Jumbo, Aldi and Lidl scoring €102,-, €100-, €85,- and €85,-, respectively.
16 Zooming in on Jumbo, one of the main interests is, of course, the effect of the chain’s different price levels per store on each stores’ price image. The ConsumentenBond (2020) released a study on the different price levels Jumbo holds and categorized each individual Jumbo store in being either in the Low, Middle or High segment. Furthermore, additional analysis on the prices of 256 National Brands at 677 different Jumbo locations in a period of 21 weeks (September 2019 – February 2020) shed light on the average prices each Jumbo asks for their products. The NBs used in this analysis ranged over 20 product categories and the basket analysis differentiated the distribution of categories along the turnover shares per category per Jumbo; i.e. if one location acquired 10% of their revenue from selling alcoholic products, 10% of the basket used in analysis for that Jumbo consists alcoholic products. Jumbos categorized by the Consumentenbond to hold the low price level averaged prices of 2.15, while Jumbos holding a middle price level averaged at 2.17 and high price levels having an average of 2.30. Moreover, the variation of the average prices within these categories was found to be microscopic to the extent that it’s negligible in further analyses.
Interestingly, as figure 3 shows, the price levels the different Jumbos hold seem to have no effect on the price image of Jumbo. This notion was confirmed by a one-way ANOVA and subsequent TukeyHSD test, which showed no significant deviations in the pairwise comparisons of the means between the groups. To conduct this analysis, each respondent was linked to the Jumbo that was closest to them, with this Jumbo, whether it being Low, Middle or High priced, acting as their point of reference. This seems to be first evidence supporting hypothesis 1, indicating counter-intuitive non-correlation of price on price image of the different stores. These findings are in line with prior studies recognizing that price is not the sole driver of price image (Hamilton & Chernev, 2013; Kavilanz, 2011). It must be emphasized again that these findings are not generalizable to pricing strategies across supermarket chains or formulas; this study focuses on the effects of pricing of one locally priced retailer on it’s overall price image.
17
4.2.2 Share-of-Wallet
In contrast to the perceived prices of Jumbo, the Share of Wallet does seem to vary across the different price level Jumbo holds. As can be seen in figure 4, Jumbo’s holding a low average price level acquire 40% of the Share of Wallet of respondents, while the middle and high priced Jumbos only make out 30% of the median distributed expenses of
respondents.
Figure 4: Boxplots of SOW across Jumbo Categories
4.2 Competitive Environment
With use of the EFMI Supermarket Database all competitive environments were able to be mapped for each individual Jumbo location. This granted insights in the types of supermarkets each Jumbo location had to compete with.
4.2.1 Amount of Competition
One of the most important components of a competitive environment is the amount of competition with which one has to battle for turnover. It’s highly likely that the amount of competition severely impacts the share-of-wallet of Jumbo, but the impact on price image is uncertain. The amount of
competition was measured by counting the amount of
supermarkets in an area of three different radiuses around each respondent, namely an area of 1km, 2,5km and 5kms around the respondent. The boxplots in figure 5 display the outcomes.
18 Furthermore, a distinction was made between all supermarkets, competitor Traditional Supermarkets and competitor Hard Discounters. As can be seen from figure 6 & 7, although the scales differ in each of the three boxplots, the distribution of amount of competition over the three radiuses are relatively similar.
The competitor variable also granted additional insight in the pricing strategy Jumbo follows. Table 2 provides the mean amount of competitors each of the three kinds of Jumbos had to compete with, from the perspective of the consumer. There are some noticeable deviations in the average amount of competitors across the differently priced Jumbos. It seems that Jumbos facing less competition (are able to) follow a high pricing strategy whilst Jumbos facing more competition (need to) follow a middle or low pricing strategy.
Mean amount of competitors across the differently priced Jumbos
N Mean Standard Deviation
Low 424 4,06 3,01
Middle 1229 4,79 4,52
High 1490 2,56 2,66
Table 2: Amount of Competitors per Jumbo Category
To statistically test this, a one-way ANOVA was conducted, which resulted in highly significant results. The subsequent Tukey HSD test also showed highly significant differences in all three of the pairwise comparisons of the means. It was found that the normality
assumption of the one-way ANOVA was violated, however, a Kruskal-Wallis rank-sum test, which does not require the data to be normally distributed, resulted in similar output. These results show the effect of Jumbos competitive environment and the benefits of a local pricing strategy; when there are few competitors around, Jumbo is able to ask vastly higher prices (as much as 8% more, as discussed later on) for their products, whilst in a competitively dense environment the retailer is able to compete by making use of their low (or middle) pricing strategy.
19
4.2.2 Distance to Competition
Additionally, with use of the Haversine formula the distance to all the supermarkets from the perspective of the respondent could be calculated; table 3 provides a summary of the spatial variables generated and used in further analyses.
Variable Meaning Mean Min Max
DistanceNearestJumbo Distance between respondent x and the nearest Jumbo in meters
1573.20 5.001 19266.85 DistanceNearestTS Distance between respondent x and
the nearest AH or Plus in meters
1665.95 37.94 20581.12 DistanceNearestHD Distance between respondent x and
the nearest Aldi or Lidl in meters
1601.09 31.24 19872.16 DistanceNearest
Supermarket
Distance between respondent x and the nearest supermarket in meters
583.45 5.001 7085.66 DistanceSecondNearest
Supermarket
Distance between respondent x and the second nearest supermarket in meters
1043.58 22.83 9260.42
Table 3: Operationalizations of Spatial Variables
The various spatial variables allowed for some continuous variables to be plotted against each other to see in what way one variable impacts the other. In Figure 8 Price Image was plotted on the distance to the nearest hard discounter, traditional supermarket and Jumbo. One important similarity between the graphs is the dip in price image when distances
approach 1000 meter. This seems to instigate a notion that another mechanism is at play here, which is not yet accounted for in the current analyses.
20 Another interesting insight is given by calculating the relative distances between each respondent and the nearest Jumbo and nearest Hard Discounter/Traditional Supermarket. The relative distances are calculated by dividing the distance to the nearest Jumbo for each
consumer by the distance to the nearest HD or TS for that consumer. With these calculations, values between 0 and 1 can be interpreted as closer proximity to a HD/TS for each consumer, while values above 1 are indicative of the responder being closer to a Jumbo than to a HD/TS.
Figure 9: Relative Distance to a Hard Discounter Figure 10: Relative Distance to a Traditional Supermarket As can be seen from figure 10, when the distance to a Jumbo is in close proximity with the distance to a Hard Discounter, a definite rise in the perceived prices of Jumbo is
noticeable. Moreover, as figure 9 shows, when the distance to a Jumbo is comparable to the distance to a competitor Traditional Supermarket, a small dip can be seen, indicating that Jumbo is perceived as being cheaper in that case. This can be seen as first evidence
concurring with Hypothesis 11 and 12, indicating that interformat competition has a positive effect on price image whilst intraformat competition negatively affects price image.
4.2.3 Share of Wallet
Figure 11 shows the linearized effects of the distances to the nearest HD, TS and Jumbo, respectively, on the Share-of-Wallet of Jumbo. The slopes are in alignment to the expectations; increased distance to a competitor Hard Discounter or Traditional Supermarket increases the Share-of-Wallet of Jumbo while increased distance to the nearest Jumbo decreases the Share-of-Wallet of Jumbo.
21
4.3 Assortment Composition
To get a first glance at the assortment compositions of Jumbo, the number of Stock-Keeping Units and the respective Private Label and National Brand shares were calculated for each of the three Jumbo categories. This preliminary analysis makes one doubtful whether any significant correlations can be found on the basis of the composition of the assortment across Jumbo locations. As could be suspected a priori, neither the number of stock-keeping units nor the shares of Private Labels or National Brands show noticeable deviations from one another across the pricing categories of Jumbo stores; it seems that, unlike their pricing strategy, Jumbo follows a rather unilateral assortment strategy.
Low Middle High
Min Mean Max Min Mean Max Min Mean max
# of SKU’s 17905 18956 21946 17949 18849 20311 17749 18696 20311 PL Share 24.84 25.46 26.73 24.83 25.43 26.19 24.69 25.37 26.19 NB Share 73.27 74.54 75.16 73.81 74.57 75.17 73.81 74.63 75.31
Avg Price 2.15 2.15 2.15 2.16 2.17 2.18 2.28 2.3 2.31
22
5.
Linear Models
In order to get a joint and summarized view of the interplay between variables, both Price Image and Share-of-Wallet of Jumbo have been modelled with use of the variety of variables available. This section is devoted to these models; the conceptual model and the hypothesized links therein are challenged in a statistical manner to see what can be perceived as the real drivers of price image and share-of-wallet of a locally priced retailer.
5.1 Price Image
Table 5 displays a first attempt to capture price image with use of all variables
hypothesized to have an effect on the construct in the conceptual model. Some imperfections are deliberately left in, which will be resolved in a second model so as to take the reader into the gradual process of model building.
From a bird’s eye view it’s quite interesting to see that the competitive environment shows much more correlation with the price image of the locally priced retailer than the variables on which the retailer actively has power over, such as price and assortment. It’s interesting to see that the amount of competition increases the price image of Jumbo, regardless of the distinction between a TS or HD. However, the effect of nearby Hard
Discounters is shown to have a greater effect, which is in line with H6, stating that interformat competition has a greater effect on price image than intraformat competition.
OLS Price Image
Type Variable Estimate Std. Error P-value
Intercept 148,43 70,50 0,04*
Pricing Middle Priceline 0,47 1,19 0,69
High Priceline 1,48 3,45 0,67 Price Dispersion -39,14 42,62 0,36 Assortment Composition # of SKUs (in thousands) 0,96 1,66 0,56 PL share -1,25 3,28 0,70 NB share NA NA NA Competitive Environment TS -0,31 0,81 0,70 HD -0,51 0,89 0,57 Amount of TS nearby 1,65 0,60 0,01** Amount of HD nearby 2,32 1,09 0,03*
Spatial Variables Distance to Closest TS 0,17 0,22 0,44
Distance to Closest HD 0,61 0,27 0,03*
Interaction Effects Amount/Distance TS -2,60 1,21 0,03*
Amount/Distance HD -3,22 1,64 0,05*
Control Variables Gender 0,95 0,77 0,22
Age -0,13 0,03 0,00*** Education 0,86 0,25 0,00** Household size 1,67 0,32 0,00*** Income -0,37 0,25 0,14 Store Size 0,04 0,81 0,96 R² 0,04 N 3177 Adjusted R² 0,03 F-Statistic 6,64***
23 Moreover, the high P-values for the pricing variables, indicating non-correlation, are quite striking. In section 1 it’s hypothesized that there is no correlation between the average price level of a locally priced retailer because of the lowest price guarantee; no matter the level of average prices, the retailer will still be the cheapest in it’s local environment. This reasoning is of course not so much proven, but the counter-intuitive non-correlation of price on price image for a locally priced retailer is a novel finding within the bodies of literature on the way a price image is construed.
There are, however, some imperfections in the model displayed in table 5 which need to be resolved to increase the interpretability and certitude of the correlations. The NAs within the National Brand Share catch the eye, for example. Because the PL share and NB share are complementary, multicollinearity is a definite detrimentally impacting factor on the model. To resolve this issue, the impact of the assortment composition was captured by taking the
absolute figures of National Brands and Private Labels each Jumbo holds as variables. However, even with these numbers, VIF scores were too high. After comparing a couple of models based on the VIF- and BIC scores, the total amount of SKUs was the sole survivor of the assortment composition variables within the models. Furthermore it was found that price dispersion moved together with the price line categories; for the purpose of the parsimony of the model the price dispersion variable was therefore omitted.
Furthermore, a Breusch-Pagan test, checking for heteroskedasticity, was found to be highly significant. This indicates that the Ordinary Least Squares method (OLS), with which the model in table 5 is built, may not be the preferable alternative given the data. Table 6 therefore displays a model built with GLS (Generalized Least Squares), which was found to be more fitting to the data. Also, a QQ-plot of the residuals and subsequent Shapiro-Wilk test brought to light that the normality assumption of the linear model is violated. This issue was resolved by bootstrapping the linear model, which resulted in the same p-values and effects. Therefore, nonnormality can be assumed not to impact the interpretation of the outcomes of the linear model. Lastly a Durbin-Watson test was conducted to check for autocorrelation; this was not found to be impacting the model
GLS Price Image
Type Variable Estimate Std. Error P-value
Intercept 89,61 12,34 0,00***
Pricing Middle Priceline -0,02 1,06 0,99
High Priceline -1,88 1,07 0,08 Assortment Composition # of SKUs (in thousands) -0,25 0,62 0,69 Competitive Environment TS -0,50 0,80 0,53 HD -0,45 0,87 0,60 Amount of TS nearby 1,63 0,66 0,01* Amount of HD nearby 2,27 1,09 0,04*
Spatial Variables Distance to Closest TS* 0,15 0,23 0,51
Distance to Closest HD* 0,63 0,28 0,03*
Interaction Effects Amount/Distance TS -2,48 1,23 0,04*
Amount/Distance HD -3,60 1,61 0,03*
Control Variables Gender 0,92 0,75 0,22
Age -0,13 0,03 0,00***
24
Household size 1,78 0,33 0,00***
Income -0,34 0,25 0,17
Store Size* -0,06 0,80 0,84
Cox&Snell (pseudo R²) 0,10 N 3164
Nagelkerke (pseudo R²) 0,10 *in km(²) for interpretation purposes Table 6: Generalized Least Squares regression on Price Image
The significant interaction effects of amount of TS/HD and Distance to TS/HD create awareness about the joint effects of amount of competition and distance to said competition; in support of Hypothesis 11, we see that the negative effect of interformat competition diminishes as said competition is stationed further away (β: -3,60; p < 0,05). Hypothesis 12 expected a positive effect of intraformat competition to be dependent on the distance to said competition, however, since we have not found a positive effect of intraformat competition on price image we can’t claim to have found supporting evidence, although it was found that the effect of intraformat competition depends on the distance to said competition.
5.2 Share-of-Wallet
The second model which is estimated, tries to explain the effects all variables available have on the Share-of-Wallet of Jumbo. After reviewing the AIC, BIC and LogLikelihood scores of a multitude of models, it was found that OLS fit the data best. Furthermore, the model in table 7 was tested for autocorrelation, nonnormality and multicollinearity which was found not to be impacting the interpretability of the model.
OLS Share-of-Wallet
Type Variable Estimate Std. Error P-value
Intercept 94,78 23,64 0,00***
Pricing Price Image -0,14 0,03 0,00***
Middle Priceline -5,01 2,02 0,01* High Priceline -3,52 1,99 0,07 Assortment Composition # of SKUs (in thousands) -1,15 1,18 0,92 Competitive Environment TS -4,46 1,48 0,00** HD 1,91 1,56 0,22 Amount of TS nearby -1,66 0,75 0,03* Amount of HD nearby -0,74 1,00 0,46
Spatial Variables Relative Distance TS 0,09 0,13 0,48
Relative Distance HD 0,74 0,17 0,00***
Control Variables Gender -4,72 1,45 0,00**
Age -0,32 0,05 0,00*** Education -0,53 0,49 0,28 Household size -0,84 0,60 0,16 Income -1,99 0,28 0,00*** Store Size -0,71 1,56 0,65 R² 0,07 N 2333 Adjusted R² 0,06 F-Statistic 10,65***
25 At first glance, it seems that the variance in SOW of Jumbo is better explained by the variables at hand than the Price Image. As expected after the preliminary analysis in section 4.2.2, the different price lines Jumbo follows are (marginally) significantly correlated with the Share-of-Wallet of Jumbo. It seems that Jumbos following a low price line acquire more percentages of the total expenditures on groceries from customers than Jumbos following a middle (β: -5,01; p < 0,05) or high (β: -3,52; p < 0,10) price line.
It’s also interesting to see the different effects the presence of TSs vs HDs have on SOW. Concurring with Hypothesis 5, intraformat competition seems to have a greater effect (β: -4,46; p < 0,01) on SOW than interformat competition. Moreover, near-presence and amount (β: -1,66; p < 0,05) of Traditional Supermarkets both negatively affect SOW, whilst these variables concerning HDs show no significant correlations. However, the relative distance to a Hard Discounter was found to be significantly correlated with SOW; the closer a Jumbo is to a respondent relative to a Hard Discounter, the higher the SOW (β: 0,74; p < 0,001).
Furthermore, as hypothesized in section 2.2.2, the amount of competition is negatively correlated with Share-of-Wallet. However, it seems that Jumbo primarily competes with supermarkets offering the same service levels and quality; the effects of nearby Hard Discounters on SOW seem to be insignificant.
5.3 Hypotheses Tested
Table 8 gives a quick overview of the hypotheses tested in the earlier sections and whether or not there was found evidence to support the hypotheses.
Hypotheses Found evidence supporting the
hypothesis?
Average Price Level is not correlated with Price Image Yes Average Price Level is not correlated with
Share-of-Wallet.
No Price Dispersion is negatively correlated with Price
Image.
No Price Dispersion is negatively correlated with Share-of-Wallet.
No Intraformat competition is expected to have a greater
negative effect on Share-of-Wallet than Interformat competition
Yes
Interformat competition is expected to have a greater negative effect on the Price Image of a locally priced retailer than intraformat competition
Yes
Amount of Competition is negatively correlated with Share-of-Wallet
Yes Amount of Competition is not correlated with Price
Image
26 When there is a Hard Discounter nearby, the effect of
Distance on Share-of-Wallet will form a U-shape.
No When there is a Traditional Supermarket nearby, the
effect of Distance on Share-of-Wallet is positive and linear
Yes
The negative effect of interformat competition on price image is strengthened when said competition is
closerby.
Yes
The positive effect of intraformat competition on price image is strengthened when said competition is
closerby.
No
The share of national brands in a retailers’ assortment is positively correlated with their price image.
No The share of private labels in a retailers’ assortment is
negatively correlated with their price image.
No The effects of the share of NBs/PLs in a retailers’
assortment is mediated by the average price level a retailer holds.
No
27
6. Discussion
This study tried to explain the price image and share-of-wallet of a locally priced retailer by means of it’s local pricing strategy, assortment composition and local competitive environment. This section is devoted to the novel and confirmed findings contributing to the large bodies of literature on Price Image and Share of Wallet. The angle at which this study perceives price image -from the perception of a locally priced retailer- is new to the area and contributed to some interesting insights in the way a price image is construed.
The first and foremost finding concerns the counter-intuitive non-correlation of the average price level on price image. In this study, respondents of a large questionnaire concerning multiple aspects of grocery shopping were linked to the locally priced retailer which was closest to them. Because data was available on the pricing strategies of each
individual Jumbo, price image, a purely subjective construct, could be regressed on data of the actual price level, an inherently objective construct, of the closest Jumbo. It was found that no significant deviations of the price image of one another were found across stores, even though some stores ask on average as much as 8% more for the same array of products.
This is a novel and striking finding within the literature on price image. Most of the research concerning price image tend to homogeneously find a positive relationship between price and price image (Hamilton & Chernev, 2013; Feichtinger, 1988; Bell & Lattin, 1998). However, these earlier findings are the result of cross-retailer research. It’s interesting to see that when the subject of interest prices locally and within retailer analysis can be conducted, no such correlations are found. In the author’s opinion, there may be two reasons which may underlie these findings. One reason might be that the subject of interest in this study prices locally because it promises to be the least expensive in any given area. Therefore the relativity aspect of constructing a price image of a retailer lapses; no matter the height of the average price level, the retailer of focus will always be the cheapest in the area (if it upholds its promise). Another reason might be that mass marketing efforts and overall price-related communications are more prominent in the construction of a price image of a retailer, even to the extent that customers don’t notice deviations in the price levels across stores. Finding the root cause of the non-correlation might be an interesting subject for further studies.
Another interesting finding is the effect the competitive environment has on the price image and share-of-wallet of the different retailer locations. Concurring with Cleeren et al. (2010), it was found that intraformat competition has a greater negative effect on Share-of-Wallet than interformat competition has. In this study, the competitive environment was mapped by looking into the presence and amount of Hard Discounters versus Traditional Supermarkets nearby respondents. It seems that even though Hard Discounters can sometimes outperform Traditional Supermarkets in acquisition utility (Vroegrijk M. J., 2012), Hard Discounters are not likely to replace Traditional Supermarkets as their single store of choice. If spending ratios drop for a Traditional Supermarket, it is most likely the revenues turn up at another TS closeby, but not across format.
28 a multitude of Hard Discounters nearby; when the amount of Traditional Supermarkets nearby rises, so does the price image of the focal retailer, but not nearly as much as when the closeby competition consists of Hard Discounters.
The effects of assortment composition on Price Image and Share-of-Wallet have regretfully not been able to be properly researched. No significant correlations on both dependent variables have been found, but a lack of variety within the explanatory variables probably underlie the lack of findings. The theoretical background on which the hypotheses concerning assortment composition were based, underlined a link between the amount of SKUs a store holds and its price image; retailers holding large assortments were able to enjoy economies of scale and are therefore able to offer lower prices than their smaller-sized counterparts (Hamilton & Chernev, 2013). These findings were based on data concerning small convenience stores up to large discounters such as Walmart; it would be naïve to think the focal retailer in this research would show similar variance in it’s total offering of products. Future research could focus on finding the effects of assortment composition on price image and share-of-wallet of locally priced retailers, where the scales of focal retailers would need to be increased, so as to be able to generate variation across retailers.
From a managerial perspective the advantages of a local pricing strategy have been made abundantly clear. It seems rather illogical that the majority of retailers within the Netherlands (and in other Western countries, as well) choose to follow a uniform pricing strategy when there are such crucial differences in the (competitive) environments each store has to endure. Analogically; one would first check the destination of a vacation and then pack accordingly, but in grocery retailing it’s as if everyone packs the same outfit, no matter the location. This irrationality is underlined by the finding that the average price level per store does not have a significant effect on it’s price image; some stores could be asking up to 8% more for their products while their clientele is equally content with the prices paid.
With dynamic pricing on the rise (Dye, 2020; Duan & Liu, 2019; Elmaghraby & Keskinocak, 2003), the effects on price image are, again, uncertain. Following the findings of the current study, the effect of dynamic pricing on price image will be pretty much non-existent. However, for perishable products, the subject of the majority of the recent literature on dynamic pricing, decreasing prices in alignment with qualitative decay could be uplifting for ecofriendliness, customer satisfaction and subsequently the bottomline. The author does, however, think that the electronic price boards can create an overall awareness about the volatilty of the prices within a store, which might be harmful for ones image as customers might get suspicious.
29
7. Limitations
Although this study provided some interesting insights, there are some limitations which need to be discussed. First and foremost, the generalizability of this study is limited by the scope of the research, which consisted of one focal retailer. Because the larger
supermarket retailers within the Netherlands tend to follow a uniform pricing strategy, researching the effects of a local pricing strategy is made difficult by the slim amount of researchable contenders. Future research could focus on assessing the findings in this study and taking them abroad, where a larger pool of data on locally priced retailers may be available. Following up, the data used in this study consisted of only Dutch retailers, so caution may be needed when generalizing over the border.
Furthermore, promotional variables are left out of the scope of the current research but could have a significant impact on both price image and share-of-wallet. It would be
interesting to see the effects of mass marketing campaigns on price image and share-of-wallet. Additionally, the spatial variables are calculated by interpreting the distance between two geographical points as the crow flies. Doing this, obstructions such as waterways or housing blocks are not accounted for, which may result in small deviations between the real distances someone has to travel to the different stores and the distances calculated.
30
8. References
Alba, J. W. (1987). The Effects of Frequency Knowledge on Consumer Decision Making. Journal of Consumer Research, 14-25.
Alba, J. W., & Hutchinson, J. W. (1987). Dimensions of consumer expertise. Journal of Consumer Research, 411 – 454.
Alba, J. W., Broniarczyk, S. M., Shimp, T. A., & Urbany, J. E. (1994). The Influence of Prior Beliefs, Frequency Cues, and Magnitude Cues on Consumers' Perceptions of
Comparative Price Data. Journal of Consumer Research, 219-235.
Alba, J. W., Hutchinson, W., & Lynch, J. G. (1991). Memory and decision making. In H. H. (Eds.), Handbook of consumertheory and research (pp. 1-49). Englewood Cliffs, NJ: Prentice Hall.
Amine, A., & Cadenat, S. (2003). Efficient retailer assortment: a consumer choice evaluation perspective. International Journal of Retail & Distribution Management, 486-497. Bell, D. R., & Lattin, J. M. (1998). Shopping Behavior and Consumer Preference for Store
Price Format: Why “Large Basket” Shoppers Prefer EDLP. Marketing Science, 66–88. Bell, D. R., & Tang, W. H. (1998). Determining Where to Shop: Fixed and Variable Costs of
Shopping. Journal of Marketing Research, 352-369.
Berelson, B., & Steiner, G. A. (1964). Human behavior: An inventory of scien-tific findings. New York: Harcourt, Brace, & World.
Cassady Jr., R. (1962). Competition and Price Making in Food Retailing: The Anatomy of Supermarket Operations. New York: Ronald Press.
Chernev, A. (2003). When More Is Less and Less Is More: The Role of Ideal Point
Availability and Assortment in Consumer Choice. Journal of Consumer Research, 170-183.
Chernev, A., & Hamilton, R. (2009). Assortment Size and Option Attractiveness in Consumer Choice Among Retailers. Journal of Marketing Research, 410-420.
Cleeren, K., Verboven, F., Dekimpe, M., & Gielens, K. (2010). Intra- and Interformat Competition Among Discounters and Supermarkets. Marketing Science, 456-473. Davidson, S. (2010, 4 6). Geographic Price Discrimination. Retrieved from Cattalaxy Files:
http://catallaxyfiles.com/2010/04/06/geographic-price-discrimination/
Dellavigna, S., & Gentzkow, M. (2019, November). Uniform Pricing in U.S. Retail Chains. The Quarterly Journal of Economics, 134(4), 2011-2084.
Desai, K., & Talukdar, D. (2003). Relationship Between Products Groups’ Price Perceptions on Shoppers’ Basket Size, and Grocery Stores’ Overall Price Image. Psychology and Marketing, 903-933.
distrifood.nl/formules/artikel/2019/12/formulespiegel-de-grootste-31
supermarktformules-101129741?vakmedianet-approve-cookies=1&_ga=2.198367275.503162226.1585926109-1427000668.1585150200 Dobson, P. W., & Waterson, M. (2008). Chain-Store Competition: Customized vs. Uniform
Pricing.
Duan, Y., & Liu, J. (2019). Optimal dynamic pricing for perishable foods with quality and quantity deteriorating simultaneously under reference price effects. International Journal of Systems Science: Operations & Logistics, 346-355.
Dye, C. (2020). Optimal joint dynamic pricing, advertising and inventory control model for perishable items with psychic stock effect. European Journal of Operational
Research, 576-587.
Elmaghraby, W., & Keskinocak, P. (2003). Dynamic Pricing in the Presence of Inventory Considerations: Research Overview, Current Practices, and Future Directions. Management Science 49(10), 1287-1309.
Feichtinger, G. A. (1988). Optimal Price and Advertising Policy for a Convenience Goods Retailer. Marketing Science, 7 (2), 187-201.
Gourville, J., & Moon, Y. (2004). Managing price expectations through product overlap. J. Retail., 23-35.
Hamilton, R., & Chernev, A. (2010). The Impact of Product Line Extensions and Consumer Goals on the Formation of Price Image. Journal of Marketing Research, 51-62. Hamilton, R., & Chernev, A. (2013, November). Low Prices Are Just the Beginning: Price
Image in Retail Management. Journal of Marketing, 77, 1-20.
Hoch, S. J. (1999). The Variety of an Assortment. Marketing Science, 18 (4), 527-546. Hoch, S. J., Kim, B., Montgomery, A. L., & Rossi, P. E. (1995). Determinants of Store-Level
Price Elasticity. Journal of Marketing Research, 17-29.
Iyengar, S., & Lepper, M. (2000). When Choice is Demotivating: Can One Desire Too Much of a Good Thing? Journal of Personality and Social Psychology, 995-1006.
Kavilanz, P. (2011, March 8). Low, Low Prices: Target Beats WalMart. CNNMoney, 1. Safeway plc and Asda Group Limited; Wm Morrison Supermarkets plc; J Sainsbury plc;
Tesco plc. (2003). A Report on the Mergers in Contemplation. London: Competition Commission.
Sprott, D. E., Manning, K. C., & Miyazaki, A. D. (2003). Grocery Price Setting and Quantity Surcharges. Journal of Marketing, 34-46.
Van Heerde, H. J. (2008, October). Winners and Losers in a Major Price War. Journal of Marketing Research, 45, 499-518.
32 Vroegrijk, M. J. (2012). "Smart shopping": Implications of hard-discounters and
multiple-store patronage. Tilburg: CentER, Center for Economic Research.
Vroegrijk, M., Gijsbrechts, E., & Campo, K. (2013). Close Encounter with the HD: A MultipleStore Shopping Perspective on the Impact of Local Hard-Discounter Entry. Journal of Marketing Research, 606-626.
Vroegrijk, M., Gijsbrechts, E., & Campo., K. (2013). Close encounter with the hard discounter: a multiple-store shopping perspective on the impact of local hard-discounter entry. Journal of Marketing Research, 606-626.
