Do consumers benefit from national-brand listings by hard discounters?

Abstract

Hard discounters are moving away from an almost exclusive private-label (PL) focus by adding a select set of big-name brands to their otherwise scanty assortments. However, to what extent consumers benefit from this strategy remains unclear. This study quantifies the consumer-welfare impact of adding a select set of leading national brands to hard discounter Lidl’s assortment in the German market. We rely on the economic concept of compensating variation, which reflects the consumer-welfare effect in monetary terms and captures the impact on consumer prices while accounting for consumers’ possible appreciation of more convenient access to big-name brands. Although hard discounters argue that adding big-name brands to their assortment benefits consumers, our results reveal unfavorable price developments, not just at the hard discounter but in the entire market. Importantly, impoverished consumers are hurt more severely, with a monetary loss due to the price increase across the 18 categories in our sample that is about five times larger than for the more affluent.

Hard discounters like Aldi and Lidl have gained a firm footing in the European grocery market. In 2019, the hard-discount share was already above 20 percent in countries like the Netherlands (22 percent) and Poland (31.6 percent). The channel’s success is even more pronounced in Germany, where the format originated, and now already has a market share of over 40 percent. While previously thought to be only a Western-European phenomenon, it has recently become the fastest-growing store-based channel in other regions as well, such as Australia, Colombia, Turkey, and the U.S. (PlanetRetail RNG 2018). Aldi, for instance, is investing heavily to increase its U.S. footprint to 2,500 stores by 2022, while Lidl, the largest hard discounter in Europe and one of the most successful grocery retailers globally, is slowly but steadily increasing its U.S. market share after its 2017 arrival (Knudson & Vu, 2019).

Hard discounters’ traditional assortment strategy is to offer a limited product range, consisting of private labels (PLs) with a very low price positioning and, to a smaller extent, non-leading and low-priced NBs. Compared to conventional supermarkets, whose assortments of around 40,000 stock-keeping units (SKUs) predominantly consist of national brands (NBs), hard discounters typically carry less than 2,000 SKUs (Cleeren et al., 2010), with PLs accounting for more than 70 percent of their assortment (Steenkamp & Sloot, 2019). Lidl’s assortment in the U.S. is even 90 percent PL, a number that Aldi exceeds in most countries it operates in (PlanetRetail RNG 2018). By economizing on their assortment size, hard discounters sell higher volumes per SKU. As a result, their PLs are, on average, 50 percent less expensive than quality-equivalent NBs, and 30 percent cheaper than conventional retailers’ PLs (Steenkamp & Sloot, 2019). Their pricing muscle not only allows them to skillfully serve their traditional low-income shopper base but also makes the store attractive to the wealthier multi-store shopper. In 2020, over 60% of all U.K. households already patronized Aldi (PlanetRetail RNG 2021).

Policymakers increasingly express concerns about the continued proliferation of grocery retailers’ PL programs and fear that the PL rise will ultimately limit consumer choice when national brands (NBs) get pushed out of the market (European Commission, 2014). To facilitate consumer access to a wider variety of products and stimulate in-store competition, they pressure retailers to offer a more balanced assortment with both own and independent (manufacturer) brands (U.N. Conference on Trade and Development, 2016, p. 14). Not only do many consumers buy both NBs and PLs (Ailawadi et al., 2001), they also value the convenience of being able to purchase all items in a single store (Popkowski Leszczyc & Timmermans, 2001). With expensive transportation, increasing traffic time due to travel distance and congestions, shopping at different retailers has become a costly and time-consuming activity (Baltas et al., 2010) that many consumers no longer want to (or even can) undertake to acquire their desired brands. Even at adjacent stores, consumers face extra queuing time at the cash register and bear the efforts to bring the shopping basket from the first store to the car or to carry it around in the second store.

Given their narrow and undifferentiated product assortments, hard discounters were traditionally less conducive to one-stop shopping (Gijsbrechts et al., 2008). In response, several hard discounters have moved away from an almost exclusive PL focus by adding a select set of leading NBs to their otherwise scanty PL assortments (Dekimpe & Geyskens, 2019). Lidl was the first to explore this strategy at its German outlets (Lourenço & Gijsbrechts, 2013). Aldi, reluctant to deviate from its almost exclusive PL focus for a long time, has recently also reduced the proportion of its PL products in most European countries. For instance, in 2017, Aldi’s assortment in the U.S. included Always, Cheerios, Coca Cola, Crest, Duracell, Febreze, Gatorade, Haribo, KitKat, Pantene, Snickers, Sprite, Tampax, and Tide (Steenkamp & Sloot, 2019, p. 218). A similar situation is currently unfolding with hard discounters in several other markets.

However, to what extent consumers benefit from the addition of leading NBs to a hard discounter’s PL-dominated assortment remains unclear. Since leading brands are already available at other (sometimes adjacent) retailers, consumers could gain from increased access to variety, rather than from increased variety per se. It could therefore be argued that hard discounters’ clientele benefits from the convenience of not having to visit an additional retailer to buy their preferred popular NB. At the same time, the simplicity of hard discounters’ small, PL-dominated assortments may be compromised through the NB additions, thereby reducing consumers’ in-store shopping convenience. Moreover, the added convenience (if any) may come at an economic cost. Prices are known to be extremely low in a hard-discount setting. But how will prices evolve if hard discounters extend their price-oriented, PL-focused assortment with more expensive branded offerings? On the one hand, hard discounters’ NB listings may induce more intense competition in the store and the market. This competition may reduce consumer prices at the hard discounter or competing retailers. On the other hand, listing a more expensive NB may increase consumers’ reference prices in the category and allow the hard discounter to boost its other prices, alleviating the price pressure in the entire market.

Importantly, leading-NB listings by hard discounters may have a differential impact on cash-strapped shoppers. The economic importance of this customer group should not be underestimated—it is large and growing. For example, consumers who struggle to make ends meet make up 40% of U.S. households and represent $1.6 trillion in spending power (Springer, 2019). These consumers, who struggle to live within the limits of their incomes, represent a significant fraction of hard discounters’ core customer base (Springer, 2019). Impoverished shoppers tend to face more transportation difficulties and often do not have the same access as more affluent residents to large chain stores with broad assortments (Chung & Meyers, 1999). When popular NBs become available at hard-discount stores, which are also more likely to be located in less affluent neighborhoods (Zentes et al., 2017), cash-constrained consumers may benefit the most from this added accessibility. At the same time, if prices would increase, they may also be disproportionately hurt, given that sensible grocery prices are especially important to less affluent consumer groups (Chung & Meyers, 1999; Talukdar, 2008).

We study these issues in the context of 18 NB listings by Lidl, the leading German hard discounter that pioneered this strategy. These listings come from a variety of category types (food, personal care, household care). The brands included vary considerably in their national market share before their listing at Lidl (from 2 to 46%). Moreover, they exhibit considerable heterogeneity in their list price, not only in absolute terms but also in comparison to the PLs available at the discounter, and in terms of the price differential to conventional retailers. Interestingly, while many listings were priced at a lower level at the hard discounter, there were also several instances where the price charged at the discounter exceeded the average price paid at conventional retailers.

Drawing on the economic-welfare literature (e.g., Hausman & Leibtag, 2007; Nevo, 2012), we investigate whether and to what extent consumers obtain (must give up) utility from a hard discounter’s NB listing in terms of lower (higher) prices and better access to variety. The strategic assortment change by Lidl that took place initially in Germany in the years after 2001 represents a unique opportunity to determine the consumer-welfare impact of extending a limited hard discounter’s assortment with one (or a few) big-name NBs. Our focus on consumer well-being, with a special interest in the impoverished, is in line with the recent call in marketing to consider the broader impact of firms’ marketing actions outside the immediate impact on their bottom line (de Ruyter et al., 2022; Haenlein et al., 2022; Kohli & Haenlein, 2021).

While we find that consumers sometimes value the additional access to leading NBs at hard discounters, there is clear evidence of an unfavorable increase in PL prices, both at competing retailers and, to a lesser extent, at the focal hard discounter. The detrimental price effects are more pronounced among more cash-strapped shoppers. Still, the benefits of more accessible big-name NBs are occasionally appreciated, also by the impoverished, which may partly counter the unfavorable price developments.

Theory

Consumer welfare

While “inextricably linked” (Heins & Deeming, 2015, p. 13), the terms consumer welfare and consumer well-being have somewhat different connotations in different strands of literature. The transformative consumer research (TCR) movement, which is dedicated to understanding and improving consumer well-being, has defined it broadly as “a state of flourishing that includes health, happiness, and prosperity” (Mick et al., 2012, p. 6). More recently, there is growing recognition of marketing’s potential to benefit the world at large by also improving people’s lives rather than only firms’ performance (see, e.g., the 2021 special issue of the Journal of Marketing on Better Marketing for a Better World or the 2022 special issue of the Journal of the Academy of Marketing Science on Responsible Research in Marketing).

Over the years, several dimensions of consumer well-being have been identified. For example, McGregor and Goldsmith (1998) identify up to seven dimensions of well-being: emotional, social, economic, physical, spiritual, environmental, and political (for a similar taxonomy, see Stiglitz et al., 2009). Consistent with prior work on the repercussions for consumers of assortment changes (see Table 1, Panel A for a review), we focus on the economic dimension. This economic footing has not only gained traction within the marketing literature (see, e.g., Brynjolfsson et al., 2003; Israilevich, 2004; Kim et al., 2002) but has also laid the foundation for much legal work on the role of consumer welfare in competition law (see, e.g., Albæk, 2013; Daslakova, 2015).

In this economic tradition, the aggregate consumer-welfare change of an assortment adjustment has been defined as the difference in consumers’ expenditures before versus after the change needed to achieve the same utility level (see, e.g., Brynjolfsson et al., 2003, p. 1583). When prices of (some of) the incumbent products change following the addition/deletion of an offering, consumers’ expenditure adjustment can be attributed at least partly to these changed prices. However, consumers are sometimes willing to accept higher prices when they compensate for other non-financial benefits, like more choice or additional shopping convenience. Thus, absent any changes in the incumbents’ prices, expenditure shifts can be attributed to consumers’ (positive or negative) appreciation of the changed variety level resulting from that assortment change. Following the pioneering work of Hausman and Leonard (2002), various studies have decomposed the overall expenditure change into a price effect and a variety effect. A similar logic is adopted in the current study to decompose the expenditure change in a price effect and an access-to-variety effect, as the latter is changed when existing NBs become available in an additional (hard-discount) retailer.

Table 1 Literature review on the effects of retailer-assortment decisions on consumer welfare

Literature review

Our study contributes to two literature streams: (i) the economic-welfare literature and (ii) the literature on the increasingly popular hard-discounter format. Table 1 summarizes the extant empirical literature on the economic-welfare implications of retailers’ assortment decisions. The table shows how, from a more technical point of view, our modeling approach is well-established in both the economic and marketing literature. Notably, Table 1 also shows how our study differs from these previous applications from a substantive point of view.

First, while we focus on NB additions to a hard discounter’s assortment, several previous studies (Cohen & Cotterill, 2011; Kim et al., 2002; Tiboldo et al., 2021) have focused on the economic-welfare implications of assortment reductions. Given that consumers may react asymmetrically to positive versus negative assortment changes (Broniarczyk et al., 1998), generalizing from these studies may be tenuous. Second, while we focus on assortment changes involving leading NBs, some previous studies (e.g., Tiboldo et al., 2021) considered changes involving PLs. Also here, consumer reactions need not be similar (Bronnenberg & Wathieu, 1996). Third, while we focus on the economic-welfare implications of adding well-established NBs, previous studies (e.g., Hausman and Leonard 2002; Kim, 2004) considered the economic-welfare implications from adding new products to the available options. Because of that, any variety effects that we will infer can be interpreted as the welfare implications of giving consumers better access to existing options, rather than of extending the number of options in the market (e.g., Hausman and Leonard 2002; Kim, 2004). Fourth, while previous studies considered the welfare implications from assortment changes in conventional supermarkets (e.g., Kim et al., 2002) or through the online channel (Brynjolfsson et al., 2003), we focus on assortment changes in the hard-discount channel, which has not only a different business model (with a very different assortment composition and price positioning), but also a different clientele (with a larger fraction of the core customers that are budget constrained). Because of that, we will, unlike previous studies, explicitly distinguish between the welfare implications for impoverished versus more affluent consumers.

Our study also adds insights to the hard-discount literature, summarized in Table 2. While some studies have already evaluated the impact of NB listings by hard discounters, we take a vastly different perspective. Specifically, we focus on consumer welfare instead of discounter- or manufacturer-related metrics, such as share growth (Deleersnyder et al., 2007), brand and category sales (Deleersnyder et al., 2012), and discounter share of wallet (Lourenço & Gijsbrechts, 2013). We thereby extend the stakeholders studied (consumers versus NB manufacturers and retailers).

Table 2 Literature review on NB introductions at hard discounters

Consumers’ welfare may be affected in multiple ways. First, a hard discounter’s NB listings may have an impact on the prices charged for other products in the market, either at the hard discounter itself (the PLs it already sells) or in competing conventional retailers (where the prices of the focal NB, competing NBs, and/or PL products could change). Apart from these potential price effects, the addition of big-name brands can also benefit consumers who appreciate the convenient access to more variety at the hard discounter.

Price effects of a hard discounter’s leading-NB listings

Within-store price effect

A hard discounter’s category assortment consists mainly of PLs and, to a smaller extent, of non-leading and low-priced NBs. The listing of a leading, big-name NB may lead to increased PL prices at the hard discounter. The higher price of a leading NB may act as a reference, given that consumers are more likely to remember the prices of popular NBs (McGoldrick & Marks, 1986). This higher reference price may allow a hard discounter to somewhat increase its other (mostly PL-related) prices in the category, making hard-discount shoppers worse off.

Across-store price effect

The availability of the same NB across retailers makes price comparisons between retailers easier and increases inter-store price competition. This puts downward pressure on the price of the focal NB at competing retailers (Cleeren et al., 2010; Hausman & Leibtag, 2007). Multi-store shoppers that used to buy this NB at a conventional retailer may now buy it at the hard discounter or may be able to buy it at a lower price at this other retailer – in both cases, they benefit.

Alternatively, if a NB manufacturer starts collaborating with a hard discounter, competing retailers that have already listed the NB for a long time may retaliate by reducing their trade support for this NB (Steenkamp & Sloot, 2019), resulting in higher consumer prices at their stores. However, the attractiveness of a retailer’s assortment depends heavily on the availability (Broniarczyk et al., 1998) and price positioning (Lourenço et al., 2015) of a consumer’s favorite brand. Even though favorite brands vary by consumer, the leading NBs that hard discounters start to list are popular with many consumers – hence, it is less likely that competing retailers will drastically reduce their trade support in response to a hard discounter’s leading-NB listing (Van der Maelen et al., 2017). We therefore expect that a hard discounter’s NB listing will drive down that NB’s price at competing retailers. In a similar vein, we expect that a hard discounter’s leading-NB listing may also drive down competing NBs’ prices at other retailers (Bolton & Shankar, 2018).

Finally, by offering leading NBs, a hard discounter meets the needs of a larger consumer group, which may pressure other retailers to explore new PL opportunities to prevent losing shoppers to this price-aggressive rival. One reaction of these retailers is to extend their assortment with lower-priced economy PL lines to serve the most price-sensitive customers (Vroegrijk et al., 2016). Another reaction is to launch category-specific phantom brands with prices at par with hard-discount prices (Steenkamp & Sloot, 2019). Both reactions lead to lower PL prices, on average, at competing retailers.

In sum, we expect that the prices of a hard discounter’s incumbent products (PLs and NBs) may increase slightly, whereas NB and PL prices at competing retailers are likely to drop. The combined effect is an empirical issue. We therefore refrain from formulating a formal hypothesis on the combined (net) effect on prices.

H1

When a hard discounter lists a leading NB, consumers are harmed by unfavorable price changes (i.e., price increases) for incumbent products in its assortment.

H2

When a hard discounter lists a leading NB, consumers benefit from favorable price changes (i.e., price decreases) for the same NB at competing retailers (H2a), for competing NBs at competing retailers (H2b), and for the PLs at competing retailers (H2c).

Access-to-variety effect of a hard discounter’s leading-NB listings

Within-store access-to-variety effect

When a NB is introduced into an environment where few to no NBs are available, the perceived variety in this environment is expected to increase considerably, as a new differentiating attribute is brought into play (Van Herpen and Pieters 2002). Consumers likely benefit because they are more likely to find a product that matches their preferences, while they can remain flexible when their preferences are still uncertain.

On the other hand, the choice literature has shown that consumers do not always appreciate an increase in variety (e.g., Mantrala et al., 2009). Access to additional variety may increase the difficulty of consumers’ decision-making process, as they need to process new/more information (Huffman & Kahn, 1998). However, given that the demotivating effect of choice has been shown to occur especially with large assortments (Iyengar & Lepper, 2000) and that the number of choice options (i.e., different SKUs) at a hard discounter is highly limited, the latter effect is less likely to occur in the hard-discounter context. Overall, we therefore expect that a hard discounter’s clientele benefits from access to a leading NB.

Across-store access-to-variety effect

Multi-store shoppers, too, may benefit from a hard discounter’s leading-NB listing even if the brand is already available at other main retailers. Consumers may prefer different retailers because they offer different products (Gijsbrechts et al., 2008). In some categories or for some purchase occasions, consumers prefer the cheap PL offerings available at hard discounters. In other categories or for other shopping occasions, they may ‘trade up’ to popular NBs available at conventional retailers (Vroegrijk et al., 2013). If a hard discounter lists a leading NB, consumers may no longer need to sacrifice time and effort to travel around, which reduces the inconvenience or disutility of shopping at multiple stores. This option becomes especially relevant when hard discounters’ store density is high, as in most European countries (European Commission, 2014).

Chernev and Hamilton (2009) further show that consumers’ preference for a limited-assortment retailer (like a hard discounter) tends to increase especially when the attractiveness of the options in its assortment rises, which is the case when a hard discounter lists a leading NB. Accordingly, also multi-store shoppers may value a hard discounter’s leading-NB listing, even if they already have access to the NB, at a comparable price, at other retailers.

In sum, we expect that consumers benefit from increased access to the leading NB in the market. We therefore hypothesize:

H3

When a hard discounter lists a leading NB, consumers benefit from the additional access-to-variety.

Do price and access-to-variety effects differ for the impoverished?

Our arguments so far pertained to the ‘average’ consumer, who has long been the focus of consumer protection laws. More recently, authorities have turned their attention to ‘particularly vulnerable’ consumers (Waddington, 2013), i.e., consumers who are more susceptible to economic harm because of characteristics that limit their ability to maximize their utility (Smith & Cooper-Martin, 1997). A key indicator of vulnerability is consumers’ income. We define impoverished consumers as consumers whose income falls below the poverty line; their limited financial resources make them unable to obtain the goods and services needed for an adequate standard of living (Hamilton & Catterall, 2005). We continue by theorizing how the expected price and access-to-variety effects may differ for the impoverished.

How does the price effect differ for the impoverished?

On average, the impoverished buy more PLs because they are cheaper (Dubé et al., 2018) and often do so at a hard discounter (Steenkamp & Kumar, 2009). Moreover, despite limited choice, hard discounters tend to carry some price-oriented NBs in several categories that are affordable alternatives to the leading NBs for low-income consumers. Since a hard discounter may (slightly) increase the prices for its established PL and/or budget NBs following a leading, more expensive NB listing (as argued above), we hypothesize:

H4

When a hard discounter lists a leading NB, impoverished consumers are harmed more by unfavorable price changes (i.e, price increases) than more affluent consumers.

How does the access-to-variety effect differ for the impoverished?

Consumers face a ‘cost conflict’ when visiting one store would entail lower fixed but higher variable costs (e.g., a conveniently located neighborhood store with higher prices), while visiting another would entail higher fixed but lower variable costs (e.g., a further located store with lower prices). Gijsbrechts et al. (2008) have shown that, under such conditions, systematically patronizing both stores on separate shopping trips – a phenomenon known as multi-store shopping – provides a compromise solution for the consumer.

However, multi-store shopping is likely to be less attractive for the impoverished than for the more affluent. Apart from their limited budgets that make shopping at full-service retailers unappealing, they also face more transportation and time constraints. A nation’s impoverished face more mobility barriers. This reduces their access to affordable goods (Talukdar, 2008), especially since their neighborhoods are less attractive to retailers (Alwitt & Donley, 1997), resulting in fewer competitors, higher prices, and a smaller assortment of goods. In addition, the impoverished spend more time commuting than the more affluent, as they tend to live in areas with lower access to work opportunities (Shen, 2000) and rely more upon slower modes of travel, such as walking and transit (Clifton, 2004) since they often do not own a car. As a result, the impoverished engage more in one-store shopping.

In sum, especially impoverished consumers may benefit from a hard discounter’s leading-NB listings, as these listings make a hard discounter—which is more likely situated in an easily accessible neighborhood (Zentes et al., 2017, p. 35) and in a low-rent district (Steenkamp & Kumar, 2009, p. 90) – more attractive as a single store-of-choice.

In contrast, accessibility is typically less of a problem for the more affluent. As the well-off are more likely to live in areas with higher store density and more often own their transportation, they can stop in stores on their way to/from work, making multi-store shopping not only feasible but also less time-consuming. Thus:

H5

When a hard discounter lists a leading NB, impoverished consumers benefit more from the additional access-to-variety than more affluent consumers.

Empirical setting

Lidl was among the first hard discounters to add big brand names to its assortment. Since 2001, it has listed a limited number of leading NBs in some of its core categories (ascentialedge.com). This strategy considerably increased the number of stores where consumers could purchase those NBs. To illustrate, in 2007, Germany’s top-three conventional retailers collectively operated 4,044 stores (Rewe: 3,170; Kaufland: 525; Real: 349). When Lidl listed a NB, it became available in an additional 2,962 stores.

We analyzed four-weekly GfK household panel data across multiple product categories from January 2003 until July 2008. This period covers the initial developments in Lidl’s branded activities in Germany, a strategy that it expanded later to other countries, in Europe and overseas. Importantly, Lidl’s practice has recently been copied by hard discounters around the world (e.g., BIM in Turkey and DIA in Spain), making Lidl’s take-off stage especially informative to countries where the hard-discount format is still in a nascent (like Australia, Croatia, Finland, and Italy; BCG, 2017) or early-expansion (like the Czech Republic, Romania, and Switzerland; BCG, 2017) stage. In this respect, PlanetRetail RNG (2016) has urged to draw lessons from Germany to prepare for the rise of hard discounters in other regions.

The data cover consumer purchases made in the five leading German grocery banners with a national market share of more than five percent during the data period: Aldi, Lidl, Kaufland, Real, and Rewe.¹ Data were available at the regional level, with Germany divided into six regions. All five retailers operate in each of the six regions. Even though there are no substantial differences in Lidl’s store density across the regions (as measured by the average number of inhabitants per Lidl outlet), the number of inhabitants per region differs. Therefore, we perform all analyses on a per-household basis.

We aggregated the data across package sizes and variants to the brand level, using the procedure outlined in Pauwels and Srinivasan (2004). For each category, we determined the top-five NBs in terms of their value sales across the five largest retailers in Germany over our data window. Next, we evaluated if Lidl listed any of these leading NBs during the data period (see, e.g., ter Braak et al., 2013 for a similar approach). We identified 23 grocery categories where Lidl listed a top-five NB during our data period. We excluded five categories where Lidl listed in rapid succession multiple variants of the same NB and/or multiple top-five NBs, leaving us with 18 categories that provide a clean setting to analyze the economic-welfare effects of a specific listing. As shown in Table 3, these listings differ considerably along multiple dimensions, such as the nature of the category and the NB’s market share before its listing.

Table 3 Descriptive statistics by product category

Methodology

Micro-economic foundation

As Brynjolfsson, Hu, and Smith (2003, p. 1582) put eloquently, the welfare approach that we adopt reflects how consumers ‘vote with their dollars,’ and therefore their appreciation of the studied assortment change. Following the seminal work by Hausman (1981) and Hausman and Leonard (2002) and later applications in economics (among others, Bokhari & Fournier, 2013 or Kuchler & Arnade, 2016) and marketing (see, e.g., Brynjolfsson et al., 2003 or Kim et al., 2002), we express consumers’ economic-welfare change as the difference in the consumer expenditure function in the category with and without the NB available at the hard discounter, while holding consumer utility constant at the post NB-listing level.

This difference is referred to as the compensating variation (CV) since it reflects how much consumers need to be ‘compensated’ to be as well off without a change as with a change (Nevo, 2012). In our setting, the change relates to a hard discounter’s NB listing. Formally:

$$\Delta\;\mathrm{Consumer}\;\mathrm{Welfare}=\mathrm{CV}=\mathrm{e}\left({\mathrm p}_{\mathrm{i},\mathrm{m},-},{\widetilde{\mathrm{p}}}_{\mathrm{NB},\mathrm{d},\mathrm{m}},{\mathrm{u}}_{+}\right) - \mathrm{e} \left({\mathrm{p}}_{\mathrm{i},\mathrm{m},+}\;{\mathrm{,p}}_{\mathrm{NB},\mathrm{d},\mathrm{m}},{\mathrm{u}}_{+}\right),$$

(1)

where ${}_{{\mathrm P}_{\mathrm i,\mathrm m,-}}$ and ${}_{{\mathrm P}_{\mathrm i,\mathrm m,+}}$ are vectors of prices of the I incumbent offerings (i = 1,.., I) in the category in the M regional markets (m = 1,.., 6) with the NB absent from (-) or present in ( +) hard discounter d’s assortment. ${\mathrm p}_{\mathrm{NB},\mathrm d,\mathrm m}$ is the observed price of the newly-listed NB by hard discounter d in market m. ${\widetilde{\mathrm p}}_{\mathrm{NB},\mathrm d,\mathrm m}$ is the ‘virtual price’ of that NB, defined as the price that sets its demand at hard discounter d in market m to zero. The incumbent offerings i include all products available in the market before the hard discounter’s NB listing. These involve all PLs and NBs already sold at the hard discounter and competing retailers. Note that also the focal NB is part of the established offerings because it was already available at most, if not all, retailers in the market before it got listed by the hard discounter. ${\mathrm u}_+$ reflects the utility level consumers achieve after the hard discounter listed the NB. Whereas the first term on the right-hand side of Eq. 1 represents consumers’ ‘virtual expenditures,’ i.e., consumers’ category expenditures in case the NB would not have been available at the hard discounter, the second term captures consumers’ actual (observed) expenditures when the hard discounter lists the NB. Keeping utility constant at the post NB-listing level $\left({\mathrm u}_+\right)$, the difference between the two expenditure functions represents the (monetary) change in consumer welfare that can be attributed to the hard discounter’s NB listing.

Empirical derivation of the economic-welfare effect

The change in consumers’ economic welfare due to the hard discounter’s NB listing can be expressed as (we refer to Hausman, 1981, pp. 664–669, for a formal derivation; see also Hausman & Leonard, 2002, p. 261)²:

$$\mathrm{CV}=\frac1{1+\widehat\delta}\left[{\widehat{\mathrm P}}_\_\left({\mathrm p}_{\mathrm i,\mathrm m,-},{\widetilde{\mathrm p}}_{\mathrm{NB},\mathrm d,\mathrm m}\right)\cdot{\widehat{\mathrm Q}}_-\left({\mathrm p}_{\mathrm i,\mathrm m,-},{\widetilde{\mathrm p}}_{\mathrm{NB},\mathrm d,\mathrm m}\right)-\mathrm P\left({\mathrm p}_{\mathrm i,\mathrm m,+},{\mathrm p}_{\mathrm{NB},\mathrm d,\mathrm m}\right)\cdot\mathrm Q\left({\mathrm p}_{\mathrm i,\mathrm m,+},{\mathrm p}_{\mathrm{NB},\mathrm d,\mathrm m}\right)\right],$$

(2)

with $\widehat\delta$ the (estimated) price elasticity for the category and all price symbols defined as before. The first term between the square brackets, ${\widehat{\mathrm P}}_\_\left({\mathrm p}_{\mathrm i,\mathrm m,-},{\widetilde{\mathrm p}}_{\mathrm{NB},\mathrm d,\mathrm m}\right)\cdot{\widehat{\mathrm Q}}_-\left({\mathrm p}_{\mathrm i,\mathrm m,-},{\widetilde{\mathrm p}}_{\mathrm{NB},\mathrm d,\mathrm m}\right)$, captures the estimated category expenditures in the absence of the NB from the hard discounter’s assortment, i.e., the virtual category expenditures. The second term between the square brackets, $\mathrm P\left({\mathrm p}_{\mathrm i,\mathrm m,+},{\mathrm p}_{\mathrm{NB},\mathrm d,\mathrm m}\right)\cdot\mathrm Q\left({\mathrm p}_{\mathrm i,\mathrm m,+},{\mathrm p}_{\mathrm{NB},\mathrm d,\mathrm m}\right)$, captures the actual (observed) category expenditures. Equation 2 is derived under the compensated-demand principle that keeps consumers on the same indifference curve as prices vary. To calculate the CV, we must quantify the different components in Eq. 2. Since we observe the actual category expenditures in the data, we only need to (i) derive the virtual category expenditures in the absence of the NB at the discounter (i.e., the virtual price index ${{\widehat{\mathrm P}}_{-}}$ times the virtual quantity demanded ${\widehat{\mathrm Q}}_-$), and (ii) estimate the category-price elasticity $\widehat{\mathrm\delta}$. These components are derived empirically from a two-equation demand system that is not only consistent with consumers’ utility-maximizing behavior, but also captures the two stages in consumers’ typical budgeting process (Nevo, 2012): a category-demand model, which estimates the market demand for the product category, and a market-share model that reflects the substitution between alternative offerings in the market.

We estimate both the category-demand and market-share models on aggregate-level data. While the underlying demand specification and its associated properties originate at the level of the individual utility-maximizing consumer, Deaton and Muellbauer (1980) have shown that these properties transfer to aggregate-level data in our two-equation system. Therefore, “the aggregate-level demand can be treated as the demand of a representative consumer” (Hausman & Leonard, 2005, p. 282), so that “the welfare of the representative consumer (i.e., from the estimated demand system) is equal to the true consumer welfare, i.e., the aggregation of the welfare over individual consumers” (ibidem, p. 283). For a similar argumentation in the marketing literature on the consistency of the adopted demand system with individual consumers’ utility-maximizing behavior, we refer to Israilevich (2004, p. 147).

Category-demand model

We use a log–log specification for the category-demand model:

$${\mathrm{logQ}}_{\mathrm m,\mathrm t}={\mathrm\beta}_0+{\mathrm\delta\mathrm l\mathrm o\mathrm g\mathrm P}_{\mathrm m,\mathrm t}+{\mathrm\beta}_1\mathrm l\mathrm o\mathrm g\mathrm C\mathrm a\mathrm t\mathrm A{\mathrm{dv}}_{\mathrm t}+{\mathrm\beta}_2{\mathrm{logProm}}_{\mathrm m,\mathrm t}+{\mathrm\beta}_3{\mathrm{logPricDisp}}_{\mathrm m,\mathrm t}+{\mathrm{\Lambda Z}}_{\mathrm m,\mathrm t}+{\mathrm\varepsilon}_{\mathrm m,\mathrm t},$$

(3)

where ${\mathrm Q}_{\mathrm m,\mathrm t}$ is the category demand (in volume) in market m (m = 1,.., 6) in period t (t = t₀,.., t_T), t₀ is the four-weekly period in which the hard discounter listed the leading NB, and T is set to two years (26 periods) after that listing.³${\mathrm P}_{\mathrm m,\mathrm t}$ is the category-price level in the market. In line with previous research (e.g., Israilevich, 2004), we use the Stone index to capture ${\mathrm P}_{\mathrm m,\mathrm t}$. The Stone index is a weighted average of the prices of the focal NB (NB), competing NBs (if any; CNB), and PLs (PL) within hard discounter d, and of the same three brand types across (the combined set of) competing retailers c. We use static weights to weigh the six prices in the Stone index, with the weights represented by the average value market share ${\mathrm s}_{\mathrm i,\mathrm r,\mathrm m}$ (see Hausman & Leonard, 2002, p. 248) of the corresponding brand type i in retailer r and market m in the first two years after the hard discounter’s NB listing. Specifically:

$$\log\;{\mathrm P}_{\mathrm m,\mathrm t}=\sum\nolimits_{\mathrm i=1}^3\sum\nolimits_{\mathrm r=1}^2{\mathrm s}_{\mathrm i,\mathrm r,\mathrm m}\log\;{\mathrm p}_{\mathrm i,\mathrm r,\mathrm m,\mathrm t}$$

(4)

with index i representing the brand type (i = NB, CNB, or PL), r the retailer (r = d or c), m the market, and t time.⁴ If no competing NBs were available at the hard discounter in the post-listing period, or if the hard discounter removed the competing NBs when listing the focal NB, the price index is based on five instead of six prices.

To obtain better estimates of the focal $\widehat{\mathrm\delta}$ parameter, we account for the advertising (Nijs et al., 2001), promotional intensity (Nijs et al., 2001), and price dispersion (Zhao, 2006) in the category. ${\mathrm{CatAdv}}_{\mathrm t}$ is the advertising in the category in the country (in Euros, obtained from AC Nielsen), while ${\mathrm{Prom}}_{\mathrm m,\mathrm t}$ captures how many of the five leading brands in the category run a price promotion in a given market. In line with Gedenk and Neslin (2000), a brand is defined as running a price promotion when its price is at least five percent lower than its average price over the estimation window. ${\mathrm{PricDisp}}_{\mathrm m,\mathrm t}$, in turn, is measured as the coefficient of variation across the top-five brands within the category. ${\mathrm Z}_{\mathrm m,\mathrm t}$ is a vector of control variables, which contains (i) a trend variable, (ii) two harmonic variables to parsimoniously capture possible seasonal fluctuations (Hanssens et al., 2001, p. 46), (iii) a Christmas dummy variable, and (iv) region-fixed effects to capture time-invariant differences in market demographics and consumer preferences. Finally, ${\mathrm\varepsilon}_{\mathrm m,\mathrm t}$ is an error component that is assumed to be i.i.d. normal. Income is not included in the category-demand equation, as the income elasticity can be ignored for fast-moving consumer goods that constitute only a small fraction of consumers’ disposable income (Brynjolfsson et al., 2003, p. 1584).⁵

Firms often do not set prices randomly, but consider consumer response, competition, and/or other unobserved demand shocks (Rutz & Watson, 2019). To correct for the potential endogeneity of the price-related variables (price and promotional intensity), we rely on the Gaussian copula approach. This instrumental-variable-free approach controls for the correlation between the potentially endogenous regressors and the error term by adding copula terms to the regression equation for each of these regressors (Falkenström et al., 2021; Park & Gupta, 2012).⁶ A copula term is a nonlinear transformation of the endogenous regressor (see Rutz & Watson, 2019, p. 490 for further details), and has the advantage of not requiring additional data. In the category-demand model, we first augment the model with both copula terms for price and promotion intensity. Bootstrapped standard errors allow for assessing whether these additional terms are statistically significant and endogeneity problems are present (Hult et al., 2018; Papies et al., 2016). In line with the recommendations of Papies et al. (2016), we retain the statistically significant copula terms (based on bootstrapped standard errors, derived from 250 random samples with replacement) and then re-estimate the model.

Market-share model

We follow Hausman and Leonard (2002) and Israilevich (2004) by relying on an extended version of the ‘Almost Ideal Demand System’ (AIDS) to model consumers’ buying behavior for the different offerings available at the hard discounter and competing retailers in the category:

$${\mathrm{ms}}_{\mathrm j,\mathrm m,\mathrm t}={\mathrm\gamma}_{0,\mathrm j}+{\mathrm\gamma}_{1,\mathrm j}\log\frac{{\mathrm Y}_{\mathrm m,\mathrm t}}{{\mathrm P}_{\mathrm m,\mathrm t}}+{\textstyle\sum_{\mathrm i=1}^6}{\mathrm\alpha}_{\mathrm i,\mathrm j}{\mathrm{logp}}_{\mathrm i,\mathrm m,\mathrm t}+{\mathrm\gamma}_{2,\mathrm j}{\mathrm{logNBAdv}}_{\mathrm t}+{\mathrm\gamma}_{3,\mathrm j}{\mathrm{logCAdv}}_{\mathrm t}+\mathrm\Lambda_{\mathrm j}^{'}{\mathrm Z}_{\mathrm m,\mathrm t}+{\mathrm\zeta}_{\mathrm j,\mathrm m,\mathrm t}\;,$$

(5)

with ${\mathrm{ms}}_{\mathrm j,\mathrm m,\mathrm t}$ equal to the category value share of offering j (j = NB at d, NB at c, CNB at d, CNB at c, PL at d, PL at c) in market m at time t (t = t₀,.., t_T) and ${\mathrm p}_{\mathrm i,\mathrm m,\mathrm t}$ the corresponding prices. ${\mathrm Y}_{\mathrm m,\mathrm t}$ equals the product category’s value sales in market m at time t, and ${\mathrm{P}}_{\mathrm{m},\mathrm{t}}$ is defined as before. ${\mathrm{NBAdv}}_{\mathrm{t}}$ is the focal NB’s advertising in the country, while ${\mathrm{CAdv}}_{\mathrm{t}}$ captures competing NBs’ advertising. We further include the same vector of control variables as in Eq. 3.⁷ We estimate the market-share equations based on data for the first two years after the hard discounter’s NB listing and do not impose restrictions. As the same explanatory variables appear in every equation, there is no efficiency gain from a SUR estimation, allowing us to proceed with OLS for estimating the focal equation, i.e., pertaining to the NB’s share with the hard discounter.

Neither the hard discounter nor its competitors may set retail prices arbitrarily. Rather, they could select prices with the potential to maximize their category market shares. Similarly, brand manufacturers may anticipate how consumers will react to their prices and adjust them accordingly. To account for possible endogeneity in the six brand/market-specific price variables in our market share equation, we implement the same copula-based correction approach as with the category-demand model.⁸

Calculating the economic-welfare effect from the estimated models

We first calculate the virtual price (${\widetilde{\mathrm{p}}}_{\mathrm{NB},\mathrm{ d},\mathrm{m},\mathrm{t}}$) that sets demand for the focal NB within the hard discounter in market m at time t to zero (suggesting the absence of the NB from the hard discounter’s assortment), while keeping the prices of the incumbent offerings at their current levels. We do so using the relevant estimates of Eq. 5. Next, we compute the virtual price index $({\widehat{\mathrm P}}_-)$, by replacing the NB’s actual price in Eq. 4 with its virtual price. At the same time, we replace the five actual prices of the incumbent offerings with their respective extrapolated (or predicted) prices in the absence (-) of the NB from the hard discounter’s assortment, to arrive at ${\widehat{\mathrm{P}}}_{-}$.⁹ Once the virtual price index ${\widehat{\mathrm P}}_-$ is determined, we include its (log-transformed) value in Eq. 3 to predict the virtual quantity (${\widehat{\mathrm Q}}_-$) demanded in the category if the NB had not been listed.

With the components obtained above, we calculate the virtual category expenditures by multiplying (i) ${\widehat{\mathrm P}}_-$, the virtual (Stone) price index with (ii) ${\widehat{\mathrm Q}}_-$, the virtual quantity demanded in the category, had the focal NB not been listed by the hard discounter. By subtracting the actual (observed) category expenditures from the virtual category expenditures and dividing by the category elasticity $\widehat{\mathrm\delta}$ + 1, we arrive at the CV (Hausman & Leonard, 2002).

Separating the economic-welfare effect into a price effect and an access-to-variety P effect

We separate the economic-welfare effect in Eq. 2 into a price effect and an access-to-variety effect. To determine the latter, the prices of incumbent offerings are fixed at their observed levels after the NB’s listing (${\mathrm p}_{\mathrm i,\mathrm m+}$), thereby removing any effect due to potential changes in incumbents’ prices from the economic-welfare effect in Eq. 2. Hence, the access-to-variety effect (AE) is given by (for the micro-economic foundation, we refer to the Appendix):

$$\mathrm{AE}=\frac1{1+\widehat{\mathrm\delta}}\lbrack{\widehat{\mathrm P}}_+\left({\mathrm p}_{\mathrm i,\mathrm m,+},{\mathrm p}_{\mathrm{NB},\mathrm d,\mathrm m}\right),{\widehat{\mathrm Q}}_+\left({\mathrm p}_{\mathrm i,\mathrm m,+},{\widetilde{\mathrm p}}_{\mathrm{NB},\mathrm d,\mathrm m}\right)-\mathrm P\left({\mathrm p}_{\mathrm i,\mathrm m,+},{\mathrm p}_{\mathrm{NB},\mathrm d,\mathrm m}\right),\mathrm Q\left({\mathrm p}_{\mathrm i,\mathrm m,+},{\mathrm p}_{\mathrm{NB},\mathrm d,\mathrm m}\right)\rbrack$$

(6)

Equation 6 captures the difference in consumers’ category expenditures when they would not have had access to the NB in the hard discounter’s assortment (at a price ${\widetilde{\mathrm p}}_{\mathrm{NB},\mathrm d,\mathrm m}$) versus when they do have this access possibility (at the observed price${\mathrm p}_{\mathrm{NB},\mathrm d,\mathrm m}$). The calculation of the access-to-variety effect is like that of the CV. The only difference occurs in the first term between square brackets: instead of using (predicted) prices of incumbent offerings in the absence of the focal NB in the virtual price index, we use their actual prices as observed when the NB is listed by the discounter. The price effect (PE) is then reflected in the remainder of the CV: PE = CV—AE.

To calculate the various effects, we compare the actual expenditures by consumers with their virtual expenditures, for a full year after the NB listing. We use a one-year window, as this is an often-used time span to evaluate the market success of new products in the grocery industry (see, e.g., Ernst & Young/ACNielsen, 2000; Lamey et al., 2018). Since the effects are based on estimated functions, we rely on bootstrapping to obtain the corresponding standard errors (using 250 random samples with replacement and winsorizing at 1% and 99% to account for outliers; see Ette & Onyiah, 2002).

Empirical results

The case of the Danone listing

To illustrate our approach, we first describe the listing of Danone (marketed as ‘Dannon’ in the U.S.) in Lidl’s yogurt category. This brand and category were also used by Kim et al. (2002) when they considered the economic-welfare implications of removing one of five popular Dannon flavors from a conventional retailer’s assortment in the U.S. At the time of Lidl’s listing of Danone, Lidl sold ten different PLs in the yogurt category, as well as four price-fighter NBs. The Danone listing marked the category’s first listing of a premium-priced big-name NB by one of the world’s top producers of dairy products (Danone has consistently appeared in Interbrand’s yearly Top 100 Best Global Brands ranking since 2002; www.interbrand.com).

Descriptives

In Table 4, we report Danone’s market share and price for each of the six regions in our data set. Note that Danone is highly valued in region 6, the former East-German region, where its market share is almost double its share in the other regions. Danone is about three times as expensive as the PLs in Lidl’s assortment. Interestingly, despite Lidl’s low-price focus, Danone’s price is slightly higher at Lidl than at competing retailers.

Table 4 Descriptive statistics for Danone by region

Price effect

Table 5 (column 2) reports the price effect of Lidl’s Danone listing for each of the six regions in our data set. Lidl’s Danone listing has a detrimental impact on prices, with yogurt prices in the market increasing corresponding to a monetary equivalent value of -€0.18 (p < 0.01), on average, across the regions. To investigate whether these effects differ for the impoverished, we separated ‘impoverished’ and ‘more affluent’ households in our dataset based on each household’s monthly disposable income.¹⁰ The impoverished (more affluent) group consists of households with an income below (above) €1,500, which approximates Germany’s 2005 poverty threshold of €1,545 (for households with two adults and two children under the age of 14; see www.wsi.de/verteilungsmonitor).¹¹ The price effect after Lidl’s Danone listing is more detrimental for the impoverished (-€0.30, p < 0.01 versus -€0.17, p < 0.01; Table 5, columns 4 and 6).

Table 5 Economic-welfare effects after the Danone listing by Lidl

Access-to-variety effect

We find a positive access-to-variety effect (AE = €1.07, p < 0.10). Thus, the average household appreciates the convenience of now having access to Danone in Lidl’s 2,500 + outlets. The access-to-variety effect for households below the poverty threshold is €3.93 (p < 0.10) compared to €0.76 (p > 0.10) for other households, indicating that especially more impoverished people appreciate the access to Danone yoghurt at Lidl.¹²

While the beneficial variety effect was expected (H3), we find evidence of an overall harmful price effect. More variety at the hard discounter comes at the cost of higher prices, which hit the most vulnerable consumers hardest.

Does the harmful price effect generalize across categories?

To uncover whether the harmful price effect generalizes to other NB listings, we analyzed an additional 17 listings by Lidl in a wide variety of product categories (including food, beverages, household care, and personal care) between 2003 and 2006.

Descriptives

Table 3 provides overall statistics for each category where Lidl listed a leading NB during our observation window. The listed NBs differ widely. For instance, the NBs’ market shares in Germany before Lidl’s listing vary between 2.77% (Milka cakes) and 46.01% (Coca Cola), and they are 19% (Eduscho coffee beans) to over three and a half time (Nivea deodorant) more expensive than Lidl’s PLs. Interestingly, even though the listed NBs are often considerably cheaper at Lidl than at competing retailers (e.g., Bübchen from Nestlé or Procter & Gamble’s Meister Proper), they can also be more expensive (e.g., Meggle butter or Danone yogurt as mentioned before). Importantly, Table 3 shows that, except for Lipton ice tea, the national market shares of all NBs increased after Lidl listed them.

Price effect

We find a negative price evolution for the incumbent offerings in 9 out of 18 categories, with an average yearly monetary equivalent value per household of -€0.18 (Z = 5.20, p < 0.01) (see Table 6).

Table 6 Economic-welfare effects after the NB listings by Lidl^a

Triangulation

To support the validity of our PE findings, we estimate a structural-break time-series model to examine the changes in category prices following a hard discounter’s NB listing. In so doing, we use an alternative dependent variable and an alternative approach that involves very few modeling assumptions.

For every category where Lidl listed a NB, we regress the category price index on (i) a set of ‘break’ variables that allow the growth path of the price series to change after Lidl’s NB listing, (ii) Z_m,t (the same vector of control variables as used in Eqs. 3 and 5), and (iii) the lagged dependent variable to allow for inertia in the price setting and reduce any serial correlation in the residuals. This trend-break model, estimated for each of the 18 categories, takes the form:

$$\log\;{\mathrm P}_{\mathrm m,\mathrm t=\;}\left[{\mathrm\gamma}_{1\cdot}{\mathrm{TB}}_{\mathrm t}+{\mathrm\gamma}_{2\cdot}\;{\mathrm{dSTEP}}_{\mathrm t}+{\mathrm\gamma}_{3\cdot}\;{\mathrm{dPULSE}}_{\mathrm t}\right]\;+\mathrm\rho\cdot\log\;{\mathrm P}_{\mathrm m,\mathrm t-1}\;+\mathrm\Lambda'''_\cdot{\mathrm Z}_{\mathrm m,\mathrm t}+{\mathrm\omega}_{\mathrm m,\mathrm t}\;.$$

(7)

The dependent variable, $\log\;{\mathrm P}_{\mathrm m,\mathrm t}$, is the Stone index in market m (= 1, …, 6) and period t (as in Eq. 4). In Eq. (7), the terms between brackets represent post-introduction ‘break’ dummy variables that change once the focal NB enters the assortment of Lidl. The trend break (TB) takes the value of the trend after the break (and zero before), the step dummy variable (dSTEP) takes the value of one after the break (and zero before), and the pulse dummy variable (dPULSE) equals one in the period of the break only (and zero in all other periods).¹³

If the hard discounter’s NB listing has an adverse effect on the prices that consumers pay in the category, this is reflected in a positive trend break, which is consistent with an upward change in the original (positive or negative) growth pattern of the price series after the break date. Indeed, when the initial price evolution changes upward, prices will rise compared to the original prices, making consumers worse off. Consistent with the adverse price effects in our welfare analysis (Table 6, column 3), a meta-analytic test confirms that prices in the market go up after a leading NB’s addition to Lidl’s assortment (meta-analytic Z = 24.22, p < 0.01). Overall, we replicate the adverse price effects in our welfare analysis with this alternative approach, which increases the confidence in our findings.

Which prices are driving the adverse price effect?

The negative price effects reported in the welfare analysis and the trend-break analysis above reflect the net change in all prices following Lidl’s NB listing. In our theorizing, we pointed out that a hard discounter’s NB listing can lead to price changes in (i) its competing PL offerings and (ii) its competing NB offerings, but can also trigger price reactions from (iii) the focal NB, (iv) other NBs, and (v) PLs at competing retailers. To obtain more insight into what drives the unfavorable price effect, we each time fix the price of the focal NB at the hard discounter to its virtual price ${\widetilde{\mathrm{p}}}_{\mathrm{NB},\mathrm{d},\mathrm{m},\mathrm{t}}$ (that sets the NB demand within the hard discounter to zero) in the calculation of the virtual price index ${\widehat{\mathrm{P}}}_{-}$, and replace one-by-one the observed price of an incumbent offering with its prediction.¹⁴ The corresponding virtual price index and virtual quantity result in a new calculation for the CV, where only a single price (out of the five incumbent prices) is allowed to change. Accordingly, we obtain to what extent the price effect is due to price changes in each brand type-retailer combination as the difference between the new CV and the AE determined by Eq. 6.

A decomposition of the price effect reveals an increase in PL prices in the year following the hard discounter’s NB listing, both within the hard discounter (average PE for the hard discounter’s PLs = -€0.03, meta-analytic p < 0.01) and even more at competing retailers (average PE for other retailers’ PLs = -€0.10, meta-analytic p < 0.01).¹⁵ The prices of the (cheaper) secondary NBs at Lidl, when they were available, also increased slightly (-€0.02, meta-analytic p < 0.01). The prices of the focal NB and competing NBs at other retailers, on the other hand, show no consistent evidence of a price increase in the year following Lidl’s NB listing (meta-analytic p’s > 0.05). Possibly, if a hard discounter moves away from a pure price focus by also carrying more expensive big-name NBs, the extreme pressure on grocery prices in the market is alleviated, and the prices at the lower end of the spectrum increase. Hence, not only the hard discounter tends to raise its prices. Also competing retailers seem to use the hard discounter’s NB listing as an opportunity to increase their PL prices (Zhu et al., 2011).

How does the price effect differ for the impoverished?

So far, the reported effects pertain to a ‘representative’ consumer and were derived from all purchases made by the full panel of GfK households in Germany (Hausman & Leonard, 2005, p. 282; see also Deaton & Muellbauer, 1980 and Israilevich, 2004). To investigate whether and how these effects differ for the impoverished, we estimate our models based on purchases from only those households below or above the poverty threshold. Table 6 (columns 5–8) portrays the results. In the exemplar yogurt category, the price effect was more detrimental for the impoverished (-€0.30, p < 0.01 versus -€0.17, p < 0.01). The same patterns can be observed across all categories: as expected in H4, the price effect is more harmful (t = -2.16, p < 0.05) to the impoverished (PE = €-0.69, p < 0.01) than the more affluent (PE = -€0.15, p < 0.01).¹⁶

Do consumers appreciate access-to-variety?

Consistent with the Danone case, Table 6 (column 4) shows that consumers often appreciate having access to the leading NB in the hard discounter’s assortment, reflected in an average yearly monetary value in the product category equivalent to €0.81 per household. A meta-analytical test supports the positive access-to-variety effect predicted in H3 (Z = 5.99, p < 0.01). At the same time, these results do not hold for all NBs, with only seven out of 18 categories showing a significantly positive access-to-variety effect.

Furthermore, in the exemplar yogurt category, the access-to-variety effect for households below the poverty threshold is €3.93 (p < 0.10) compared to €0.76 (p > 0.10) for other households, indicating that especially less affluent people value the added access to variety for yogurt. The same pattern is observed across all categories: the access-to-variety effects are significantly larger (t = 4.38, p < 0.01) for the impoverished (AE = €10.76, p < 0.01) than the more affluent (AE = €0.31, p < 0.01), supporting H5.¹⁷ For the more affluent, accessibility is less of a problem. Such households tend to already have a wider selection of retailers in their neighborhoods (Alwitt & Donley, 1997). In addition, they are more likely to possess their own transportation (Clifton, 2004), making multi-store shopping easier.

Robustness checks

We run several sensitivity analyses on our overall and group-level results and perform a face-validity test. We summarize the results in Table 7.

Table 7 Robustness checks

Sensitivity analyses for overall results

First, we test the sensitivity of our results to a semi-log (linear-log) specification for the category-demand equation. Second, we estimate our models using all available post-listing data rather than two years. Third, we allow for a copula term for $log Ym,t/Pm,t$ in Eq. 5 to account for potential endogeneity in category sales (see, e.g., Israilevich, 2004). Our analyses show robust findings in all instances.

Sensitivity analyses for group-level results

First, when we vary the income thresholds for impoverished households from €1,500 to €1,750 and €2,000, the welfare effects are largely similar, although the effects become (as expected) somewhat smaller in absolute size as the income threshold increases. Second, we consider three, rather than two, income groups, and compare the economic-welfare effects for the impoverished with a middle-income group (monthly income between €1,500 – €2,500) and a higher-income group (monthly income above €2,500). The results indicate that the average economic-welfare effects per household from the inclusion of a NB in the hard-discounter assortment are indeed largest for the impoverished group (PE = -€0.69; AE = €10.76), compared to middle-income households (PE = -€0.09; AE = €0.37), and smallest for higher-income households (PE = -€0.08; AE = €0.22). As expected, also the average category price elasticity underlying these effects is highest (most negative) for the impoverished (-0.46), indicating they are most sensitive in their category purchases to price changes. As a comparison, the average category price elasticity for middle-income households is -0.38, while it is only -0.12 for high-income households who hardly adjust their demand when prices change.

Discussion

Hard discounters predominantly sell PLs. Aldi and Lidl, the world’s two largest hard discounters, are also the number one and two sellers of PL grocery products worldwide. Even so, hard discounters are gradually extending their scanty PL assortments with a few big-name, leading NBs. Lidl has been the first to do so in the European market.

We quantified the consumer-welfare impact of leading-NB listings by Lidl in 18 product categories in the German market. We relied on the economic concept of compensating variation, which reflects the consumer-welfare effect in monetary value and captures not only the impact on prices but also accounts for an access-to-variety effect. Although hard discounters argue that adding big-name brands will benefit consumers who appreciate the convenient access to more variety at the hard discounter, our results revealed that this comes at the expense of some unfavorable price developments, not just at the hard discounter but in the entire market.

Across the 18 categories, we observed a price increase equivalent to €3.22 per German household per year. As more hard discounters are adding leading NBs to more categories, households’ monetary losses from higher prices can rise substantially. When we decomposed the price effect into five components, we found that especially the prices at the lower end of the spectrum increased, primarily due to competing retailers raising their PL prices in the category following the hard discounter’s NB listing. Still, also the hard discounter increased its (PL and secondary-NB) prices, albeit to a smaller extent. Finally, we found that impoverished consumers were hurt more severely, with a monetary loss due to the price increase across the 18 categories in our sample that is considerably larger in absolute size than for the more affluent (€12.43 vs. €2.78).

Still, in some categories, there seems to be a silver lining in the form of increased access-to-variety that consumers appreciate. Our results support that consumers sometimes extract value from more convenient access to a leading NB in an otherwise limited, PL-dominated hard-discounter assortment. For an average consumer, across the 18 categories studied, this amounts to a yearly economic benefit of €14.57, a number that will become larger as NBs are introduced in more categories. While proponents may argue that these beneficial access-to-variety effects could counter the adverse price developments, we do want to point out that they are not consistent across the 18 categories. At the same time, the estimates also have higher uncertainty, making it hard to formulate general conclusions. Still, our findings hold important implications, not only for consumers but also for retailers.

Implications for consumers

The traditional absence of big-name brands from hard discounters’ assortments puts constraints on the consumption of variety, not necessarily because consumers cannot purchase these NBs budget-wise, but due to the inconvenience of having to travel around and acquire them elsewhere. With Lidl at the forefront, several hard discounters are gradually adding more leading NBs to their assortment. Such listings arguably bring along more choice and variety in a scanty PL assortment and may therefore allay consumers’ demand for more shopping convenience (Dekimpe et al., 2020; Grewal et al., 2020).

Although proponents advocate consumer utility will increase through its contribution to more in-store variety, our results reveal that this strategy has important adverse consequences on grocery prices. Not only the hard discounter’s prices but prices in the entire market tend to develop unfavorably when hard discounters carry more big-name brands. At the same time, the beneficial effects on consumers’ access-to-variety are not universally observed across all categories. Ultimately, consumers are not necessarily better off when hard discounters add NBs to their assortments.

Our results further reveal that the new strategy affects especially more cash-strapped shoppers, traditionally the hard discounters’ core customers. Extant research has shown that lower-income households pay more for groceries, as they do not have the same access to large chain stores and their lower prices as non-poor residents (Chung & Meyers, 1999). Stores, in general, are less likely to locate in poorer areas, and this is even more so for large chains—where assortments tend to be larger and prices lower—than for small or non-chain stores. In addition, the transportation infrastructure in these lower-income neighborhoods tends to be meager (Mendoza, 2011). These factors conspire to make the impoverished pay more for their groceries and/or have less variety to choose from. Unfortunately, hard discounters carrying more NBs does not seem to contribute to more equal prices for the poor. On the contrary, even though they may sometimes appreciate the more convenient access-to-variety, cash-strapped consumers will pay more for their groceries, thereby increasing the inequality gap. As the unfavorable price developments are most pronounced among the cheapest products in the market, the impoverished suffer even more to pay for their basic consumption needs when more hard discounters extend their assortments with expensive leading NBs. More generally, rather than only considering the welfare implications for the ‘average’ or ‘representative’ consumer, we advise policymakers to assess which groups are affected most (in our case, the impoverished).

Implications for hard discounters

Hard discounters owe their identity to their relentless low-price PL focus. By adding more well-known NBs to their assortment, they aim to offer shoppers more convenience yet move away from their original positioning. Inspired by Lidl’s pioneering strategy, traditional contenders such as Aldi and fast-growing challengers such as Biedronka (Jerónimo Martins’ hard-discount chain in Poland) and BIM (Turkey’s largest hard discounter) have also started to list a limited number of leading NBs. Industry sources (e.g., PlanetRetail RNG 2016) have repeatedly emphasized how insights from Lidl’s experience in Germany may provide valuable lessons to these discounters.

Our results imply that hard discounters’ decision to list big-name NBs may blur their key value proposition by moving away from rock-bottom PL (and secondary-NB) prices. This may not only reduce their differentiation from conventional retailers (BCG, 2017) but also affect their appeal to their core, more cash-strapped customers, who may switch to those hard discounters who stick to their original value position (e.g., the Russian discounter Mere, which has begun to challenge Aldi and Lidl in several Western-European markets). While this can be seen as another manifestation of the well-known wheel of retailing evolution, the NB-listing discounter should realize it puts unforeseen pressure on the format from both the top- and bottom-end of the competitive spectrum. To not put too much pressure on the format, it could avoid listing NBs or refrain from raising the prices of its incumbent (PL and second-tier NB) offerings. Either strategy would be welfare-creating for the more cash-strapped shoppers, which constitute a significant portion of hard discounters’ customer base.

Implications for conventional retailers

Conventional retailers often attribute the share they lose to hard discounters to the extreme price differential between both formats (Vroegrijk et al., 2016). When hard discounters add popular NBs to their limited, PL-dominated assortments, conventional retailers fear losing appeal on that differentiating dimension where they still had a competitive advantage. Being unable to prevent the hard discounter from listing big-name NBs, we showed how conventional retailers take recourse to the price instrument. However, rather than reducing prices to improve their overall price image, conventional retailers typically opt to increase the price of their PL products, even more so than the hard discounters, which plays further into the hard discounters’ hands. This is a missed opportunity for conventional retailers to win defected customers back from the hard-discount format, especially the more cash-constrained ones who suffer the most from hard discounters’ own price increases.

Limitations and directions for further research

Several areas for future research remain open. First, our study considered multiple NB listings by one hard discounter, albeit an important one that pioneered the practice. The impact on consumer welfare was also derived from household purchases in the five main grocery chains in the German market. Future research could establish to what extent our results hold for other hard discounters, when accounting for all competing retailers (also smaller ones and non-grocery chains selling leading NBs), and in different countries.

Second, by considering 18 different listings, we could derive a first set of empirical generalizations on the sign and size of the economic-welfare effect. However, to limit the dilution of their business model, hard discounters have the incentive to extend only a select number of categories with NBs, and to identify those NBs that have the highest potential to enhance consumers’ variety perceptions. Future research could explore contingency factors on the size of the price and access-to-variety effects.

Third, following the listing of a NB, the question emerges how to set its marketing mix. For example, what would be the within-store (relative to the price of the PLs) and between-store (relative to the price of the same NB at conventional retailers) price gaps that result in a win–win situation to all three stakeholders: consumers (with a higher economic welfare), hard discounters (with a higher share of wallet) and NB manufacturers (with a higher market share).

Fourth, in the longer run, incumbents may not only react with their prices but also with their quality levels in response to a hard discounter’s NB listing. Pauwels and Srinivasan (2004) found indications that NBs tend to introduce higher-quality versions of their products when a PL enters their product category. In contrast, a hard discounter’s NB listing could damage NB margins, which, in turn, may induce NBs to reduce their costs by lowering product quality, an issue with clear welfare implications that would be worthy of more research.

Finally, in line with prior assortment research, we focused on the economic-welfare implications of NB additions, which allowed us to monetize the impact. Still, consumer well-being is a multidimensional concept of which economic or material well-being is just one dimension (Heins & Deeming, 2015). It would be helpful to supplement the current analyses with additional indicators that capture, for example, consumers’ overall satisfaction and quality of life to arrive at a more holistic representation.

Appendix

Micro-economic foundation of price and access-to-variety effect

The economic-welfare effect can be split into a price effect and an access-to-variety effect (see Hausman & Leonard, 2002 for a formal derivation):

$$\displaystyle\text{CV} = \underset{\text{price effect}}{\underbrace{\left[\mathrm{e}\left(\mathrm{p}_{\text{i,m},-},\widetilde{\mathrm{p}}_{\text{NB,d,m}},\mathrm{u}_{+}\right) - \mathrm{e}\left(\mathrm{p}_{\text{i,m},+},\widetilde{\mathrm{p}}_{\text{NB,d,m}},\mathrm{u}_{+}\right)\right]}} + \underset{\text{access-to-variety effect}}{\underbrace{\left[\mathrm{e}\left(\mathrm{p}_{\text{i,m},+},\tilde{\mathrm{p}}_{\text{NB,d,m}},\mathrm{u}_{+}\right) - \mathrm{e}\left(\mathrm{p}_{\text{i,m},+},\mathrm{p}_{\text{NB,d,m}},\mathrm{u}_{+}\right)\right]}}.$$

(8)

The price effect is the difference in category expenditures due to a change in the prices of the category’s incumbent offerings (including the focal NB at competing retailers) from $\mathrm{p}_{\text{i,m},-}$ (i.e., prices in the absence of the NB at the hard discounter) to $\mathrm{p}_{\text{i,m},+}$ (i.e., prices in the presence of the NB at the hard discounter). The price of the NB in the hard discounter is set equal to its virtual price $(\tilde{\mathrm{p}}_{\text{NB,m,d}})$ to reflect the NB’s absence from the hard discounter, while the utility is kept constant at u₊, the utility level consumers achieve in a market with the NB present at the hard discounter. If prices of incumbent offerings go down (at the hard discounter and/or at competing retailers), the discounter’s NB listing is welfare-enhancing (a positive price effect). Indeed, if prices are higher when the NB is absent, consumers must spend more to achieve a given utility level (the first term in the price-effect expression) than what they would have to spend when the NB is present (the second term in the price-effect expression). The resulting ‘saving’ leaves more money with the consumer and is thus increasing his/her economic welfare. In contrast, if prices of incumbent products increase, the consumer will be left with less money, making the listing welfare-reducing (a negative price effect).

The access-to-variety effect represents the value of having access to the leading NB in the hard discounter’s category assortment. It is the difference between a consumer’s category expenditures when the NB is present in the hard discounter’s assortment (at a price $\mathrm{p}_{\text{NB,d,m}}$) versus when it would not have been available (at a price $\tilde{\mathrm{p}}_{\text{NB,d,m}}$), while both the other prices and consumer utility are kept constant (at $\mathrm{p}_{\text{i,m},+}$ and u₊). The additional access is beneficial to consumers (and hence, their welfare increases) if their category expenditures to reach the utility level u₊ are lower with the NB present rather than absent from the hard discounter’s assortment, again leaving more money with the consumer.