Anything but hard discount: Storevorousness and the social stratification of grocery shopping venues in Italy (2014–2022)

Abstract

Quantitative research on the cultural stratification of food consumption has largely focused on what people eat rather than where they buy it. This study introduces the concept of “storevorousness” to examine the social stratification of grocery venues in Italy, analyzing nationally representative Household Budget Survey data from 2014 to 2022. Findings indicate that while supermarkets dominate everyday provisioning, traditional shops and street markets are eroding, and online or direct-from-producer channels remain marginal. Multivariate analyses reveal a strong wealth gradient: higher-expenditure households purchase significantly more items overall and distribute their shopping across a wider array of venues. Crucially, this storevorousness is highly selective. Affluent consumers exhibit an “anything but” pattern, deliberately avoiding hard discounts: they are less likely to enter them and buy fewer items when they do. Furthermore, while absolute gaps are largest in mainstream supermarkets and traditional stores, relative gaps peak in niche, symbolically charged venues like direct-from-producer and online. Together, these results suggest that classed distinctions in food consumption are enacted not only through what households buy, but through patterned engagement with specific kinds of retail space: being able to buy from anywhere, but not from everywhere.

“Look what I’ve done with this shop […] I’ve stocked it up with everything, I’ve spent millions on it, but it’s all top-quality stuff. Here: taste this prosciutto, see what it is—taste how sweet it is… Taste this ham! I told you, it’s sweet! It’s like sugar, Sergio—can’t you taste it?! And these olives? Here, try these olives. They’re Greek, you know… Greek! Come on, go on, they’re Greek! They’re good, right? How are they? Tell the truth.” — Mario Brega as Augusto in Borotalco (1982) ¹

Introduction

In this scene from the classic Italian comedy Borotalco, Mario Brega plays Augusto, a proud Roman local grocer, confronting his daughter’s timid fiancé Sergio (Carlo Verdone) with a barrage of offerings: sweet prosciutto, prized Greek olives, and the full sensory display of his well-stocked shop. It is not an invitation but a forceful performance of taste, abundance, masculinity, and class. His shop is more than a site of commerce: it is a social stage where status and tradition are asserted through food and its presentation. Implicitly, that staging also hints at a broader historical tension. As the supermarket format took hold in postwar Italy, it did more than reorganize provisioning; it transformed grocery shopping routines and triggered explicit reactions from small retailers, with merchants’ associations mobilizing around the defence of traditional stores (Scarpellini, 2007). These tensions were part of a broader transformation in food culture, as democratic ideals made overt hierarchy less tenable while criteria like authenticity, provenance, and personal expertise slowly came to organize judgements of taste (Johnston and Baumann, 2007). Arguably, the distinction at stake was not only about different goods (and prices), but about what those places meant for customers: the shopkeeper’s authority, the performance of expertise, and a promise of accountability grounded in identifiable people, set against the supermarket’s convenience and its more impersonal system based on the assurance of quality and safety (Everts and Jackson, 2009).

This paper begins from the premise that the place where groceries are purchased is at least as important, socially and symbolically, as the food itself (Johnston, 2008; Paddock, 2015). Even though many food shops now carry the very same Greek olives and sweet prosciutto once exclusive to local grocers, catering to consumers seeking novelty and authenticity claims (Gerosa, 2021; Grosglik, 2017), food venues continue to serve as powerful markers of status and identity. Arguably, one could leave both a supermarket and a food pantry carrying the same set of food items, yet the two experiences would signal profoundly different worlds of class, dignity, and necessity (De Souza, 2019).

A large body of sociological work shows that food tastes are socially stratified and have evolved over time. What people eat (Gronow and Holm, 2019), what they dislike and avoid (Lindblom and Mustonen, 2019; Oncini et al., 2023), where they dine (Warde et al., 2025), have all been the subjects of empirical scrutiny. At times, these studies have highlighted the rise of an omnivorous disposition among upper classes, often interpreted as a marker of cultural openness, or conversely as a way to draw symbolic boundaries and reinforce distinctions (Oncini and Triventi, 2021; Warde et al., 2007b). While considerable attention has been given to food practices and culinary preferences, the role of shopping venues has received far less consideration. This is a significant omission, particularly in light of the increasingly heterogeneous landscape of food retail.

In response, this paper extends the framework of cultural omnivorousness to the spatial dimension of food consumption by introducing the concept of storevorousness: the socially patterned breadth and selectivity with which households draw on diverse grocery venues. Storevorousness captures how households allocate routine provisioning across retail spaces, including the systematic combination of some formats and the avoidance of others. It is not simply ‘shopping more’: households with similar purchase volume may differ sharply if one concentrates spending in a single format while the other spreads it across venues. Just as the food omnivore negotiates symbolic hierarchies of taste, the storevore navigates hierarchies of place, embracing variety and patterned tolerance within boundaries of legitimacy and comfort. This conceptual extension offers a more complete understanding of how classed distinctions are enacted not only through what people eat, but also through where they procure it.

To do so, the paper focuses on the Italian context, using extremely detailed data from the Italian Statistical Institute (ISTAT) Household Budget Survey (2014–2022) to examine the social stratification of grocery shopping venues. The analysis traces the distribution and evolution of food purchases across seven types of retail outlets (traditional shops, street markets, supermarkets, hard discounts, department stores, direct-from-producer, and online sellers), capturing both their structural weight and symbolic significance. Specifically, the paper addresses three research questions: (1) Where do Italian households buy their groceries, and how are food items distributed across different venue types? (2) How have these patterns changed over the past decade (2014–2022), and specifically in the wake of COVID-19, and to what extent have any shifts persisted afterwards? (3) How are shopping venues, and the kinds of foods purchased therein, socially stratified? And particularly, is the cultural omnivore also a “storevore”, namely someone who draws from a wide and varied range of grocery venues? By shifting attention from food items to the places where they are bought, the analysis sheds light on the role of shopping venues in signalling membership and boundaries underlying grocery shopping routines, particularly in relation to innovative and potentially sustainable practices.

The next sections situate venue choice within debates on omnivorousness and patterned tolerance, with particular attention to the food domain. I then describe the HBS data and modelling strategy. The findings section first presents descriptive patterns and time trends, and then turns to multivariate analyses of overall purchasing volume, concentration, and venue- and item-specific acquisition. Finally, the discussion summarises the main findings, reflects on their implications for understanding food and cultural stratification, and outlines directions for future research.

Cultural stratification, omnivorousness and food consumption

The concept of cultural omnivorousness has played a central role in contemporary theories of stratification, particularly in understanding how upper-class dispositions may have evolved away from rigidly highbrow tastes toward a broader and more eclectic repertoire of consumption. Originally proposed as a corrective to Bourdieu’s (1984) thesis of cultural hierarchy and distinction, omnivorousness suggests that high-status individuals increasingly incorporate elements from both elite and popular domains, signalling not exclusionary refinement, but cultural versatility (Peterson and Kern, 1996). At the same time, research has made clear that this apparent openness is actually selective. The omnivore often does not consume indiscriminately, but exercises what Bryson (1996) termed “patterned tolerance”, namely a selective openness to certain items considered low-brow, paired with strong aversions to other ones.

Whether omnivorousness represents a genuine shift towards broader and more inclusive repertoires, or simply a rearticulation of the homology thesis under more flexible and cosmopolitan aesthetic rules, remains the subject of ongoing debate (Chan, 2019; de Vries and Reeves, 2022; Flemmen et al., 2019; Warde et al., 2007b). Concurrently, scholars have pointed to significant methodological challenges in how omnivorousness is measured and conceptualized (Johnston et al., 2019). In empirical research, much depends on how the boundaries between cultural forms are drawn, and whether one is measuring likes, dislikes, knowledge, participation, avoidance, or actual consumption practices.

The level of granularity at which cultural preferences are measured is particularly relevant here. Coarse-grained survey categories tend to show that elites maintain broader cultural repertoires than non-elites, spanning both high-brow and popular forms. This pattern underpins what de Vries and Reeves (2022) called the “weak” version of the omnivore hypothesis. Conversely, fine-grained measures and qualitative studies often reveal persistent symbolic boundaries, namely preferences and judgements that embrace certain popular genres while excluding others coded as “lowbrow” or associated with low-status groups, sometimes even adopting them only with ironic distancing (e.g. Peters et al., 2018). Such findings mean that the “strong” version of the hypothesis, which expects omnivores to refrain from systematic class-based exclusion, is rarely supported in empirical research.

These measurement challenges are especially pronounced in the field of food consumption, where cultural omnivorousness can encompass an array of overlapping and sometimes contradictory practices, preferences, and judgements. Food (and drinks) can be categorized in countless ways, based on cuisines, ingredients, brands, nutritional content, degree of processing, cultural provenance among others, and at levels of granularity ranging from very coarse (staples) to highly detailed (ingredients used for recipes, nutrients). Moreover, food consumption is not simply about what is liked or disliked, eaten (or avoided), but also about how, when, and with whom it is eaten—and crucially, I will argue, where it is purchased.

At this point, it is useful to distinguish cultural omnivorousness from cosmopolitanism. Omnivorousness concerns selective boundary crossing between “high” and “low” status forms, relying on justificatory repertoires that sustain symbolic boundaries even when framed as openness (Johnston and Baumann, 2007). Cosmopolitanism, by contrast, refers to openness toward the “other” across ethno-national boundaries (Regev, 2007). While the globalization and hybridization of cultural production and circulation expand the range of available items for omnivorous repertoires, cosmopolitanism is not constitutive of omnivorousness: one can be a cultural omnivore without being cosmopolitan, for instance by combining fine-dining versions of national dishes, regional specialities, and everyday convenience foods, while remaining largely indifferent to cuisines framed as “foreign.”

With this distinction in mind, existing literature does suggest a set of recurring themes that help to define the ideal-typical characteristics of food omnivorousness, particularly as a marker of upper class taste. What often characterizes this figure is a playful commitment to diversification, reflected in a wide-ranging repertoire of food types and practices (Johnston and Baumann, 2007; Johnston et al., 2019; Oncini, 2020; Warde et al., 1999); a curiosity about the unfamiliar and the non-local, often framed in terms of exoticism and sometimes connected to cosmopolitan sensibilities (Oleschuk, 2017); an experiential orientation, where novelty, innovation, and sensory discovery are prized (Warde et al., 2025); and a normative commitment to health-conscious eating and feeding practices (Wills et al., 2011). At the same time, this eclecticism frequently involves a form of selective traditionalism, or even gastronativism (Parasecoli, 2022), a valorisation of “genuine”, “authentic”, and “local” food practices, such as artisanal methods or regional cuisines (Gerosa, 2021; Yalvaç and Hazir, 2021). Ethical concerns are also often foregrounded, especially with regard to environmental sustainability, animal welfare, or support for local economies (Johnston et al., 2011; Kennedy et al., 2019). Finally, and perhaps most crucially from a sociological perspective, omnivorousness often entails subtle mechanisms of exclusivity and exclusion: a preference for high-quality, costly, or polished products or venues, coupled with a rejection of products and practices perceived as low-status, massified, or morally suspect, which underpins the “anything but” disposition (Bryson, 1996; Lindblom and Mustonen, 2019; Oncini, 2019; Yalvaç and Hazir, 2021). These orientations neither neatly co-occur nor appear uniformly across the middle classes (e.g. Flemmen et al., 2018), but taken together, they point to a set of class-coded dispositions that inform food choices and serve to distinguish the reflexive, upper-class consumer from others.

From omnivore to “storevore”

While place has occasionally surfaced in debates on cultural stratification and food culture (e.g., Cutts and Widdop, 2017), venues of food purchasing remain an under-examined yet crucial axis of cultural stratification in quantitative research. Unlike other dimensions of cultural consumption, where the place of consumption and the object are inherently tied (for instance, attending a concert to consume music or visiting a museum to experience art), food consumption is often characterized by a separation between the site of acquisition and the site of use: groceries are typically purchased in one venue and consumed elsewhere, making the place of purchase itself a distinct and meaningful marker of taste.

Qualitative research suggests that retail spaces are not neutral arenas but key sites where moral distinctions and status claims are made. Johnston’s (2008) study of Whole Foods Market shows how store design and curated variety invite shoppers to frame purchasing as an ethical and civic project, yet within marketized choice and a middle class ethos. Grosglik (2017) similarly argues that organic retail settings, from upscale supermarkets to box schemes and urban farmers’ markets, serve as resources for identity work, status display, and cosmopolitan orientations. In Turin’s Porta Palazzo market, Black (2012) shows that social boundaries are negotiated as much through venues as through food: for instance, Romanian meat stalls or Moroccan or Tunisian fruit stands can carry distinctive social meanings that organise hierarchies of legitimacy within the market. By contrast, research on US “dollar stores” (Shannon, 2021; Vargas, 2021) documents chronic under-investment and stigma in discount environments concentrated in poor and racialized neighbourhoods, conditions that both mirror and reinforce spatialized inequalities and shape middle class avoidance.

The symbolic salience of food shopping sites became particularly evident during the COVID-19 pandemic, when restrictions on mobility and fears of contagion triggered a marked rise in online platforms and delivery services (Dulsrud and Forno, 2024). For some, these digital intermediaries offered a practical response to lockdowns; yet the majority still preferred to rely on traditional grocery shopping sites, despite the risk of contagion and the other restrictions put in place by governments (e.g., Meister et al., 2023). These experiences during the pandemic likely left a trace, with online and hybrid formats becoming a more established, if unevenly adopted, part of everyday food provisioning.

With the notable exception of research on food deserts and retail access (Zhen, 2021), however, quantitative studies of food stratification have rarely focused on food shopping venues. Yet, the ideal-typical strands of food omnivorousness are not only embodied in the food consumed but also find spatial expression in the diversity of venues through which they are acquired. Experientialism and exoticism are enabled by ethnic stores, global e-commerce, and food delivery platforms, which can offer a vast array of unfamiliar or international products (Dulsrud and Forno, 2024; Zukin, 2008); selective traditionalism plays out in farmers’ markets, local grocers, alternative food networks, food trucks, and artisanal shops that emphasize provenance and craft (Dal Gobbo et al., 2022; Gerosa, 2021; Paddock, 2015); and ethical reflexivity is anchored in organic supermarkets, zero-waste stores, and direct-from-producer schemes that build moral narratives into shopping routines (Johnston and Szabo, 2011; Paddock, 2015). Perhaps most visibly, exclusivity and exclusion are enacted through venue avoidance: discount retailers and some large-scale supermarkets are negatively perceived not only for their assortments but for what they are taken to signify about production, labour, and value (Cairns and Johnston, 2015; Oncini, 2019).

Hence, if omnivorousness reflects an openness to diverse cultural repertoires, it could find its spatial counterpart in storevorousness, namely the capacity to navigate and pick from a plurality of retail spaces. At the same time, as patterned tolerance defines the omnivore’s cultural stance, such openness is never indiscriminate but selectively calibrated. As Johnston and Baumann's (2015: p. 146) interviews with foodies show, consumers do not move across stores interchangeably. Rather, they ascribe symbolic and ethical value to particular retail settings – farmers’ markets, co-ops, or Whole Foods – seen as spaces that take the worry away. In this sense, patterned tolerance in venue choice reveals how cultural openness operates within structured inequalities, reaffirming classed dispositions even in seemingly pluralistic consumption.

While existing survey data can only capture part of the diversity of food venues, they nonetheless make it possible to trace systematic differences in how social groups navigate and use them. The ISTAT Household Budget Survey is unusually detailed in this respect, offering distinctions across a wide range of outlet types. Although it cannot encompass the full spectrum of retail formats and emerging propositions, it nonetheless provides a rare opportunity to examine how venue choices are socially stratified and how these patterns have evolved over time.

Data and methods

Data

The analysis draws on data from the ISTAT Household Budget Survey (HBS) for the years 2014–2022. The HBS is a nationally representative survey of Italian households, administered via computer-assisted personal interviewing (CAPI). Designed primarily to monitor household expenditure patterns, since 2014 the survey has incorporated highly detailed information on the retail venues where specific products are purchased. This feature makes the HBS uniquely suited to examining the intersection of food purchasing and store selection, as well as changes in these patterns over time. The present study uses the full sample of 183,924 households with complete data on all variables considered. Sample sizes for each survey wave are reported in Table A3 of Appendix 1.

In section 5 of the HBS questionnaire, the household’s reference person records, over two consecutive weeks, the main purchase venues for each of 20 food products, including bread; pasta; biscuits, melba toast and snacks; fresh meat; frozen meat; cured meat; fresh fish; frozen fish; milk; cheese; yogurt; eggs; fresh fruit; fresh vegetables, tubers and legumes; dry or frozen vegetables, tubers and legumes; olive oil; mineral water; soft drinks; wine; and coffee (beans, ground or pods). For each product in any week, respondents may indicate up to two venues per item from the following: traditional shop, street market or mobile vendor, hard discount, hypermarket or supermarket, department store or chain store, direct from producer or other venue, and online purchase (see Table A1 in Appendix 1 for full details). A not purchased option is also available for items not bought during the reference period. This feature means that the HBS does not merely record isolated purchases, but offers a window into food shopping as a set of routinised practices (Warde, 2014), that is, regular and patterned behaviours shaped by household routines and broader dispositions, rather than as discrete, one-off decisions, self-reported “usual” behaviours (e.g., “where do you usually …”), or abstract preferences.

Analytical strategy

To address the research questions, and given the complexity of the data (20 food items by 7 possible venue types, with multiple responses allowed over 2 weeks), the information was transformed into the following variables:

DV1. Purchase of each food item in each store type (week 1 and 2 combined)

A binary indicator capturing whether a given product was purchased in a given venue at least once during the 2-week period. This reduces the complexity of the raw data and enables an initial visualisation of frequencies via heatmap.

DV2. Total number of purchases

The total count of product–venue combinations reported by each household over the 2 weeks, offering a measure of the overall purchasing activity (Table A2, Appendix 1). While this measure can be interpreted as a proxy for “storevorousness”, it does not account for the distribution of purchases across venues.

DV3. Concentration index (Herfindahl–Hirschman index, HHI)

The HHI measures, in this case, the degree of concentration in purchases across store types for each household (Table A2, Appendix 1). It is calculated as:

$HHI={\displaystyle \sum }_{\left\{i=1\right\}}^N{s}_i^2$

where $s$ is the share of purchases made in store type $i$ and $N$ is the total number of store types. The index ranges from 1/N (perfectly even distribution of purchases across all venues) to 1 (all purchases concentrated in a single venue type). Lower values indicate greater diversification, whereas higher values indicate greater concentration. Because it is based on shares rather than absolute counts, the HHI is not affected by the total number of purchases, and therefore does not penalise households that buy fewer items overall.

DV4. Number of purchases by store type

The count of purchases made in each specific store category, ranging from 0 to 40 (Table A2, Appendix 1).

Models

All models include a set of socio-demographic and contextual controls (see Tables A3 and A4 in Appendix 1 for descriptive statistics). Individual characteristics comprise the age, gender, and employment status of the household reference person. Household-level characteristics include type of family, household size, education level (measured using the dominance approach, i.e., the highest level attained by the reference person or their partner), geographical macro-area of residence, and size of the municipality. Familial economic resources are captured using total household expenditure (in deciles), which is generally regarded as a more reliable indicator of living standards than reported income, particularly in high-income countries where self-reported income tends to be underreported or volatile (Meyer and Sullivan, 2013; ONS, 2018).

Additional variables capture the household’s absolute poverty status, survey year, whether the household produces food for its own consumption, and the number of cars owned. While the HBS does not include spatial measures of proximity to different store types, the inclusion of municipality size, car ownership, and geographical macro-area provides partial control for retail availability and mobility constraints. Together, these controls mitigate, though they do not fully eliminate, the risk of conflating shopping preferences with the availability of particular store types.

The analysis proceeds in two steps. First, I present descriptive statistics in the form of a heatmap that summarises, combining the two reference weeks, the share of households purchasing each product in each venue type (DV1). Then, I examine trends in venue use over time by plotting the proportion of households that purchase at least one item in each store type as well as the average number of items purchased per venue (for buyers only) (DV4). Second, I turn to multivariate analysis. OLS regression is applied to the total number of items purchased across all venues (DV2); the HHI concentration index (DV3) is modelled with a one-inflated beta regression, given the spike at perfect concentration (HHI = 1, 19.1%). This two-part mixture treats boundary values separately and the interior differently: (i) a logit component models the probability of a boundary outcome (perfect concentration, HHI = 1), and (ii) conditional on 0 < HHI <1, a beta regression models the mean μ (logit link) and precision ϕ (log link) (Buis, 2020). I report average marginal effects for education and expenditure deciles on both the probability of perfect concentration and the conditional mean HHI. Venue-specific models (DV4) are then estimated using a two-step strategy: a logistic regression to capture whether households purchase at least one item in a given store type, followed by a negative binomial regression for the number of items purchased conditional on participation. Negative binomial regression is used due to substantial overdispersion in all venue-specific purchase counts (Long and Freese, 2014) (see Table A5, Appendix 1). Finally, product–venue variables (DV1) are modelled using logistic regressions, and the difference between the marginal effects for the highest and lowest expenditure quintiles is visualized in two heatmaps, one showing absolute and the other relative differences in predicted probabilities, providing a concise overview of how the likelihood of purchasing specific products in each venue varies across the social gradient.

Findings

Descriptive statistics

The heatmap (Figure 1) provides a first descriptive overview of the distribution of food purchases across different retail venues. The data appear extremely consistent and valid, as it clearly captures the expected segmentation of Italian food retail: different types of traditional stores (bakeries, butchers, fishmongers, greengrocers) emerge quite distinctly. Bread, for example, is still purchased primarily in bakeries (54.1%), while fresh meat also shows a strong presence in butchers (36.5%), and fresh fish in fishmongers (18.9%). Fruit (30.4%) and vegetables (28.3%) are likewise well represented in traditional outlets, confirming the enduring importance of small-scale shops in everyday grocery shopping.

Figure 1.

Heatmap of food purchases by retail venue and product (Italy, 2014–2022). Note. Cells show the share of households purchasing at least one item in a given venue for each product category. N = 183,924.

Street markets play a visible but more circumscribed role, particularly for fresh products. Around 16.9% of households report buying fruit at markets and 16.1% report buying vegetables, while the figure is smaller for fish (6.2%). These results resonate with the long-standing role of open-air markets in Italy, often championed by Coldiretti, the main national farmers’ association, as symbols of authenticity and proximity between producers and consumers. Nonetheless, their use remains a minority practice compared to the other venues.

As expected, supermarkets and hypermarkets dominate grocery provisioning across almost all food categories. In most cases, between 45% and 70% of households rely on them: for example, 64.0% for pasta, 64.7% for biscuits, 68.8% for cheese, and 63.5% for milk. Even for fresh products like fruit and vegetables, supermarkets account for close to 60% of purchases, reflecting their function as one-stop shops that combine price competitiveness, assortment, and convenience. Department stores register only marginal relevance, with values rarely exceeding 1–1.3% across all categories. Given that they are typically defined as outlets offering a wide variety of product types or specialised chains focused on non-food categories such as sporting goods, electronics, drugs, or furniture, their negligible role in food retail is unsurprising. Hard discounts instead occupy a somewhat intermediate position: although they consistently trail behind supermarkets, they attract a non-negligible share of consumers, especially for packaged goods. Around 14.2% of households buy biscuits there, 13.9% cheese, and 13.2% pasta. Yet their penetration is systematically lower than that of supermarkets, and even for basic staples such as bread and milk, hard discounts reach only around 10–13% of households.

Purchases directly from producers remain limited overall, with shares typically below 3%. Some interesting peaks, however, appear: 3.3% of households buy olive oil directly from producers, 2.7% vegetables, and 3.5% wine, hinting at the presence of oil mills, farms, wineries that maintain direct connections with consumers. Despite these niches, however, direct-from-producer channels are far from mainstream in Italian food provisioning. Finally, internet purchases remain extremely residual, rarely exceeding 0.2% of households for most items. An exception is coffee (0.9%), likely linked to the widespread popularity of coffee pod subscription services. While online grocery shopping remains marginal, its distribution echoes that of supermarkets: a platform-based model that potentially allows access to all product categories (Oncini, 2023).

Trends over time

Figure 2 shows the decade-long trends in participation and the average number of items purchased across venues. Over time, grocery shopping patterns appear broadly stable, though with some notable shifts for several formats. Traditional shops and street markets have experienced a steady decline: the share of households buying in traditional outlets fell from 79% in 2014 to 63% in 2022, while street markets dropped from 33% to around 20%. Purchases directly from producers also declined, from around 16–19% in the mid-2010s to about 10% in 2022.

Figure 2.

Trends in venue use, 2014–2022: participation and average basket size. Note. Left panels: share of households that purchase ≥1 item in each venue (participation). Right panels: average number of distinct items purchased among users of each venue (basket size). N = 183,924.

Together, these trends reflect the continuing erosion of proximity stores, open-air markets, and short food chains, in line with data reported from trade associations (Confesercenti, 2024). Conversely, supermarkets and hypermarkets remain dominant, with participation consistently at or above 88% and the largest average baskets of all venues (around 17 items). The marked dip observed in 2022 may hint at a post-COVID rebalancing, but longer time series would be needed to confirm whether this marks a structural shift.

Other venues play more marginal roles. Hard discounts attract around 25–34% of households, but the intensity of use has increased over time. Internet purchases show the clearest COVID-related disruption: participation rose from under 1% in 2014 to a peak of 2.4% in 2020, with average baskets nearly doubling, before receding slightly yet remaining above pre-pandemic levels. The pandemic also left a clear imprint on offline provisioning: participation in street markets and direct-from-producer channels dropped sharply in 2020 (e.g., street markets fell from 26% in 2019 to 20% in 2020, and direct purchases from 13% to 9%), before partially recovering in 2021. Despite the vivacity of the Italian online food scene, from specialty e-commerce to delivery platforms (Oncini et al., 2020), online grocery shopping remains marginal.

Total volume and store concentration

Figure 3 shows how total purchasing volume varies across household resources and education. The panels display predictive margins and confidence intervals from the OLS regression on the total number of items purchased across the 7 stores, holding other covariates constant. Unsurprisingly, household resources emerge as a key driver: the predicted number of purchased items increases steadily with expenditure deciles, from about 18 items among the lowest-spending households to nearly 30 items among those in the top decile. This monotonic gradient suggests that wealthier families not only spend more, but also engage in more diversified purchasing activity, accumulating larger shopping baskets across venues. Education shows instead a different and somewhat counterintuitive pattern. The association is weak in magnitude but points to a modest negative gradient: households with higher educational attainment are predicted to purchase slightly fewer items overall, with averages declining from about 25.6 among those with only primary schooling to 23.5 among tertiary-educated households. While the effect size is negligible compared to that of expenditure, this finding may suggest that education is associated with a leaner or more selective purchasing style.

Figure 3.

Predicted total volume of purchasing across venues by household resources and education. Note: Adjusted predictions from OLS of the total number of items purchased across the seven venue types, shown over total expenditure deciles (left) and household education levels (right). N = 183,924. Full results are available in Appendix 2.

Figure 4 examines instead concentration in venue use with a zero–one inflated beta model for the HHI. The panels report predictive margins and confidence intervals for total expenditure and education level: the top panels show the model-implied probability of complete concentration—buying all items from a single store (HHI = 1), while the bottom panels plot the expected HHI conditional on diversification (0 < HHI <1). Net of overall purchasing volume, both parts of the models point in the same direction: wealthier households are markedly less likely to buy everything in one shop and display lower HHI values—the richer the household, the lower the index, consistent with a more storevorous grocery shopping style that draws on a wider mix of venues. Holding all covariates constant, the probability of complete concentration drops from about 26% in the lowest expenditure decile to 12% in the highest. Conditional on not being perfectly concentrated, the average HHI declines from roughly 0.64 to 0.59 over the same range. Interpreted as numbers-equivalent stores (1/HHI), this corresponds to an increase from about 1.56 to 1.69 effective stores: lower-spending households mainly rely on one primary outlet with a small top-up elsewhere, whereas higher-spending households spread purchases slightly more across additional venues. Across both sets of margins, education gradients are essentially flat.

Figure 4.

Concentration of venue use (HHI) by total expenditure deciles and education. Note: Top panels: predicted probability of complete concentration (HHI = 1). Bottom panels: expected HHI conditional on diversification (0 < HHI <1). N = 182,476 (1448 missing values for households not reporting any purchase). Full results are available in Appendix 2.

Anything but hard discount

Figure 5 reports predictive margins and confidence intervals from the two-step models applied to all seven store types (logistic regression for participation; negative binomial for the number of items conditional on participation), averaging over the other covariates. In line with patterned tolerance, an “anything but” pattern emerges. Across most venues, both the probability of entry and the number of items rise with total expenditure, with the sole exception of hard discounts. For discounts, participation falls from about 0.35 in the first decile to 0.25 in the tenth, and conditional baskets trace an inverted-U, peaking around 9.9 items at the third decile before declining to about 7.8 at the top. This means that affluent households are not only less likely to enter but, when they do, they buy less, pointing to selective use rather than differences in access.

Figure 5.

Two-part predictions for entry and conditional volume across venues. Note (i) logistic regression for participation (any purchase), (ii) negative binomial regression for volume conditional on participation. See Table A1 for samples sizes, Figures A3 and A4 for frequency distribution. Full results are available in Appendix 2.

Traditional shops, street markets, and direct-from-producer channels display an opposite and clearly nonlinear gradient: relatively flat at low expenditure, then steepening in the upper deciles. Entry into traditional shops rises from 0.64 to 0.80, with conditional volume increasing from roughly 5.7 to 9.1 items; street markets move from 0.20 to 0.33 in entry and from about 3.1 to 3.9 items among users; direct-from-producer shows the sharpest upscale tilt, with entry jumping from 0.06 to 0.22 and baskets growing from about 1.8 to 2.7 items. Hyper/supermarkets remain the backbone of provisioning: participation is already high and saturates early (0.86 in the first decile to about 0.96 by the tenth decile, flattening after the third and fourth deciles), while conditional volumes rise from roughly 12.2 to 19.4 items. Department/chain stores are niche but tilt upscale (from 0.03 to 0.10; volume from 3.0 to 4.8 items). Internet purchasing remains marginal yet concentrated at the top (from 0.006 to 0.042); among users, baskets are small and differences are hard to judge due to limited cases (about 1.7% of households buy at least one item online and only 0.6% buy more than two). Taken together, rising resources broaden venue repertoires everywhere except in the discount segment. Lower entry and lower conditional volume in discounts, alongside higher entry and volume in traditional, market, and producer channels (and, to a lesser extent, department/chain and online), indicate a storevorous yet selective shopping style among higher-spending households that is unlikely to be driven by store accessibility alone.

Absolute and relative differences by product and venue

Finally, the absolute (Figure 6) and relative (Figure 7) differences in total expenditure quintiles in the predicted probabilities of purchasing each item by venue reveals two complementary dimensions of social stratification in grocery shopping. The heatmap of absolute differences, expressed in percentage points, highlights where differences matter most in quantitative terms—namely where wealthier households contribute the largest gains in the overall probability of purchasing the item at least once in the 2-week period. The largest absolute gaps appear in hyper- and supermarkets, where the probability of purchasing almost all products is consistently higher among the richest households than the poorest. Traditional shops follow, with especially wide gaps for fresh meat (+14.3 pp), fresh fish (+15.1 pp), bread (+13.7 pp), fruit (+12.2 pp), and vegetables (+10.4 pp), likely confirming the enduring importance of butchers, fishmongers, bakeries and greengrocers as food sites of middle and upper classes. Street markets show pronounced differences for fresh produce (+8.6 pp for fruit, +7.2 pp for vegetables), while direct-from-producer outlets display stronger gaps for olive oil (+5.6 pp) and wine (+5.0 pp), reflecting the higher engagement of affluent households with short supply chains. In sharp contrast, hard discounts consistently record negative values, typically around −4 to −7 pp for fresh products, highlighting their clear lower-income orientation. Together, these findings reveal that affluence increases participation in almost all store types except hard discounts, confirming selective storevorousness: wealthier households shop more frequently, in more venues, but avoid those associated with economic necessity.

Figure 6.

Absolute differences (Q5 − Q1, percentage points) in the predicted probability of purchasing at least once in the 2-week period, by product and venue. Note. Logistic regression of any purchase (1 if household buys the item in the venue at least once over 2 weeks; 0 otherwise). N = 183,924. Cells show the difference in predicted probabilities (from logistic regression) between the highest (Q5) and lowest (Q1) expenditure quintiles, expressed in percentage points. Positive values (darker) indicate that affluent households are more likely to purchase the item in that venue; negative values indicate they are less likely. For example, +14.3 for fresh meat in traditional shops means affluent households are 14.3 percentage points more likely to buy fresh meat there than the poorest households.

Figure 7.

Relative differences (%) in the predicted probability of purchasing at least once in the 2-week period, by product and venue. Note: Logistic regression of any purchase (1 if household buys the item in the venue at least once over 2 weeks; 0 otherwise). N = 183,924. Cells show the relative difference in predicted probabilities between the highest (Q5) and lowest (Q1) expenditure quintiles, expressed as a percentage of the Q1 baseline. This measure emphasizes proportional rather than absolute differences in predicted probabilities. For example, a value of +400% means affluent households are five times as likely as the poorest to make that purchase.

The relative differences, expressed as percentages, shift attention from where inequality is most common to where it is most distinctive. Once predicted probabilities are rescaled to the lower-income baseline, some of the sharpest disparities emerge in more selective retail channels, especially direct-from-producer outlets and the internet. As for the former, high-income households are several times more likely than low-income ones to make a purchase: around 6.4 times as likely for cured meat, 5.6 times for olive oil, nearly 6 times for fresh meat, and about 3.9 to 4.5 times for vegetables and wine when buying directly from producers. Such purchases likely take place through wineries, oil mills, local farms, and other artisanal producers, or occasionally via specialty coffee roasteries, signalling both the economic means and the inclination to seek provenance, authenticity, and a more direct connection with producers. A similar pattern emerges online, where relative differences reach some of their highest levels: affluent households purchase coffee, olive oil, or wine at rates roughly 7 to 8 times as high, and buy frozen fish at nearly 5 times the rate. Although online grocery shopping remains relatively rare in Italy, these patterns likely capture highly selective behaviours, such as ordering premium wines through specialised platforms rather than routine provisioning. By contrast, relative gaps in traditional shops and supermarkets are generally more modest, while in hard discounts they are negative, with affluent households roughly half as likely to buy staples such as fresh meat or bread, and about one-third less likely to buy pasta. Taken together, these results suggest that absolute differences capture the material weight of inequality across mainstream venues, whereas relative differences reveal its more symbolic edge: the retail spaces where affluence is most sharply expressed through selective producers’ circuits and curated online platforms.

Discussion and conclusions

Despite long-standing recognition that consumption unfolds within specific material and institutional settings, the venues of grocery shopping have remained largely peripheral in quantitative analyses of cultural stratification. Classic and recent analyses of food stratification have concentrated on dietary content, taste hierarchies, culinary preferences, and commensality patterns, while the spaces through which such distinctions are enacted have received only limited attention.

This article begins to fill this gap by focusing on the Italian context and by drawing on detailed data from the HBS to examine how social stratification shapes everyday food provisioning. The analysis reveals that Italian grocery shopping remains organised around a remarkably stable hierarchy of venues. Hyper- and supermarkets continue to dominate as the core infrastructure of food purchasing, serving nearly all households across the income distribution. Traditional specialty shops and open-air markets persist as important sites for fresh products (bread, meat, fish, fruit, and vegetables), particularly for wealthier households, possibly suggesting that what were once routine neighbourhood shops have increasingly become spaces of distinction. In contrast, direct-from-producer and online platforms remain niche and marginal, appealing only to a narrow segment of affluent households.

The low prevalence of these latter channels carries important implications in light of the growing literature on short food chains, sustainable food platforms, and retail innovation (Dal Gobbo et al., 2022; Dulsrud and Forno, 2024; Oncini et al., 2020). Far from the optimistic portrayals that cast farmers’ markets, direct sales, or digital platforms as transformative engines of a sustainable food transition, the Italian case offers a more sober, if less inspiring, picture. These venues remain peripheral to the weekly provisioning of most households, reflecting both structural constraints in retail organisation and enduring class-based differences in access, routines, and meanings attached to food purchasing. Instead, these circuits are used sporadically and primarily by affluent consumers. What emerges, therefore, is a realistic image of their current role in the daily lives of Italian families, and of the limited capacity they presently have to drive large-scale shifts toward more sustainable, local, or circular food systems. Their social and economic reach remains narrow and high-end, and their contribution to the sustainability transition, while symbolically charged, is materially constrained by their scale and their user niche.

In line with literature on cultural omnivorousness, the multivariate analyses reveal a subtle but consistent expansion of venue repertoires with rising household resources. Wealthier families are not only able to buy more, but also to distribute their purchases across a wider range of outlets, exhibiting what can be termed a storevorous pattern of consumption. This broadening of engagement is nonetheless selective. These households show greater participation and larger baskets in almost all venues except one: the hard discount. Here, participation declines steadily with expenditure, and conditional quantities suggest the existence of an inverted-U, peaking among lower-middle groups and falling among the richest. In other words, affluent families are both less likely to shop in discount outlets and, when they do, they purchase fewer items, hence suggesting deliberate avoidance rather than differences in access or need. Education, by contrast, plays a surprisingly minor role: once expenditure is accounted for, differences by educational attainment are negligible. This finding may indicate the primacy of material over cultural resources in structuring grocery venue choice. Possibly, two distinct pathways of stratification operate in food consumption: one driven by economic capacity, shaping where households can afford to shop; and another, more culturally inflected, influencing what kinds of products they choose within those venues and how they are used and talked about. Unlike dining out, concert attendance, or museum visits, grocery shopping is perhaps an invisible, backstage practice; cultural capital may matter more for public consumption where it can be displayed and recognized, while venue choice may be more straightforwardly shaped by economic constraints. Alternatively, the finding may indicate that when household resources are measured through detailed expenditure data, rather than self-reported income or occupational categories, the economic dimension of class becomes more visible, revealing how practical affordability, rather than cultural capital alone, shapes access to different stores.

This “anything but” pattern closely parallels the selective openness observed in other domains of cultural and food consumption (Bryson, 1996; Lindblom and Mustonen, 2019). As Cairns and Johnston (2015) show, middle-class consumers’ aversions to discount retailers stem not from price or quality concerns but from maintaining moral and aesthetic boundaries. One of their interviewees vividly captures this stance: “There’s a Food Basics near me that I would never go in because it’s a hole… I am not going to go in there and have that sort of experience. It’s disgusting.” Similar affective boundaries surface in Italy, as one respondent explains: “I don’t go to discount stores, because I trust my supermarkets… you don’t know what you’re buying in discount stores” (Oncini, 2019: p. 25). Such classed emotions of discomfort and disgust translate directly into venue choices, allowing privileged consumers to assert moral distance from spaces coded as abject while affirming their entitlement to pleasurable, curated environments.

While the HBS offers unusually rich data on grocery venues, two main limitations should be acknowledged. First, although the typology of stores is detailed by survey standards and much more precise than the “how often” questions, it remains imperfectly granular. The traditional store category, for instance, merges diverse outlets (bakeries, butchers, fishmongers, and greengrocers) and does not allow distinguishing between ethnic or organic-only retailers. Finer retail categories would be needed to assess the extent to which venue repertoires map onto cosmopolitan or gastronativist orientations. Similarly, the direct-from-producer and online options encompass heterogeneous forms of exchange, from farm-gate sales to e-commerce platforms that can offer very diverse ranges of products (Oncini et al., 2020). These ambiguities, however, likely yield conservative (lower-bound) estimates of social differentiation, since finer distinctions, such as organic boutiques or ethnic specialty shops, would probably reveal even sharper gradients in storevorousness. Similarly, the 20 food items included in the HBS cover a broad but still partial range of household groceries: while they capture the staples of everyday provisioning, they exclude certain categories, such as ready and takeaway meals, that also form the basis of many families' eating practices.

Second, the absence of direct measures of proximity or retail density limits the ability to separate preference from opportunity structures. However, the inclusion of municipality size, car ownership, and regional controls provides reasonable proxies for spatial access, and the persistence of strong gradients across all models suggests that the observed patterns are not reducible to geography alone. This is particularly evident in the case of hard discounts, where participation declines and purchase volumes follow an inverted-U pattern across expenditure deciles, indicating that differences are not driven by availability alone. Indeed, evidence from the United States shows that even in spatially dispersed, car-dependent contexts, improved physical access to different outlets has negligible causal effects on diet quality and health outcomes (Zhen, 2021). If distance and store density exert little influence on food purchasing in such environments, their explanatory power in Italy, where retail geographies are far denser and travel distances substantially shorter, should be even weaker. The persistence of class gradients in store choice, therefore, likely reflects differentiated consumption dispositions and symbolic boundaries rather than mere opportunity structures.

Future research could build on these findings in several directions. First, it would be valuable to apply the same, or a closely similar, survey structure used by ISTAT in the HBS to other national contexts, collecting biweekly records of food purchases by venue. The relative simplicity of the instrument, its proven reliability, and the wide diffusion of budget surveys make it a promising model for comparative research on the social organisation of grocery shopping. Alternatively, the store-choice module alone could be revised, expanded, and replicated on smaller population samples across countries to generate comparable data at lower cost and effort. Comparative data would also help establish whether the pattern identified here is distinctively Italian or more general, especially in light of cross-national differences in the balance between home cooking and eating out (Warde et al., 2007a). While the specific role of traditional shops likely reflects Italian food culture, the broader mechanism of storevorous diversification combined with selective avoidance of discount formats is plausibly generalizable to contexts where similar retail hierarchies structure everyday provisioning.

Repeating the survey could also allow researchers to complement these findings with geo-referenced data on store locations to disentangle dispositions from opportunity structures more precisely, and possibly analyse their interaction. For instance, it is plausible that storevorousness increases, net of economic resources, when households reside in areas with denser and more diverse retail provision. Conversely, it is possible that hard discount avoidance is attenuated among affluent households residing in areas where they are easily accessible. Combining such spatial retail information with measures of political orientation would also open a promising line of inquiry into how venue repertoires intersect with identity politics. In contexts where food becomes a vehicle for claims about authenticity and belonging, venue choice may reflect not only classed distinction but also the politicization of “local” food and the boundary drawing it can enable (Parasecoli, 2022). This suggests testable questions about whether political orientations and residential sorting shape openness toward ethnic groceries, foreign-owned chains, or online platforms, and whether some forms of localism translate into the avoidance of retailers coded as culturally “other”.

Finally, it is possible to imagine other domains of cultural consumption where the place of purchase matters at least as much as, or even more than, the good acquired, albeit with different temporal horizons. In clothing, for instance, the social meaning attached to venues such as fast-fashion chains, vintage shops, luxury boutiques, online stores, or personal tailors often exceeds that of the garments themselves, possibly revealing different lines of social differentiation. Alternatively, home furniture offers another illustrative case: designer studios, second-hand platforms, or local artisans often embody contrasting aesthetic and ethical orientations, from thrift and sustainability to craftsmanship and exclusivity. Exploring these domains could help assess whether the mechanisms observed in grocery shopping are part of a broader pattern through which venues express and reproduce social differences.

Writing about Manchester in the 1840s, Engels (2007) observed that working-class consumers were largely confined to small, often unscrupulous shopkeepers who sold adulterated and inferior goods. The material conditions have shifted dramatically since then: food safety regulation, cold chains, and mass retail have altered what reaches consumers and how. Yet one dimension of Engels’ observation persists. Even where goods themselves have converged, the relationship between consumers and food intermediaries continues to mark social position. The grocer’s performed authority, the supermarket’s impersonal assurance, the farmers’ market’s promise of directness and authenticity, the hard discount’s efficiency: each signals something about who the buyer is and where they stand.

By centering venue choice in the analysis of food consumption, this study reveals that intermediary relationships remain a powerful yet largely invisible dimension of stratification. The concept of storevorousness extends cultural omnivorousness theory into the spatial domain, demonstrating how class distinctions operate through the selection of retail relationships as much as through taste. Understanding these spatial dimensions of consumption offers a more complete picture of how inequality manifests in everyday provisioning, where the choice of intermediary subtly demarcates who belongs where.

Supplemental material

Supplemental Material - Anything but hard discount: Storevorousness and the social stratification of grocery shopping venues in Italy (2014–2022)

Supplemental Material for Anything but hard discount: Storevorousness and the social stratification of grocery shopping venues in Italy (2014–2022) by Filippo Oncini in Journal of Consumer Culture.

Supplemental material

Supplemental Material - Anything but hard discount: Storevorousness and the social stratification of grocery shopping venues in Italy (2014–2022)

Supplemental Material for Anything but hard discount: Storevorousness and the social stratification of grocery shopping venues in Italy (2014–2022) by Filippo Oncini in Journal of Consumer Culture.