Quality competition on markets: a socio-economic account

Abstract

1. Introduction

The rise of the ‘new’ economic sociology, which began in the early 1980s, has long focused attention on market valuation processes. Classic studies have shown that social relations affect valuation outcomes (Baker, 1984; DiMaggio and Louch, 1998), that professional audiences shape and stabilize evaluative orders (Zuckerman, 1999; Hsu et al., 2012), that status serves as a proxy for quality (Podolny, 1993; Benjamin and Podolny, 1999; Lynn et al., 2009; Malter, 2014), that valuations are negotiated outcomes of political struggles (Fligstein, 2001), and that product valuations are often based on contingent and even contradictory evaluative orders (Espeland and Stevens, 1998; Beckert and Aspers, 2011; Beckert and Musselin, 2013). Together, these studies have made a strong case that real-world markets operate by logics that are more complex than the basic neoclassical account of price formation as the meeting point of supply and demand suggests.

However, even as sociologists have shed light on the many factors that affect valuation outcomes, this proliferation of findings has not been accompanied by commensurate formal, integrative modeling of markets. Since the early days of the field, when observers noted a healthy engagement with economist conceptions of markets (Granovetter and Swedberg, 1992), little progress has been made in formulating genuinely sociological alternatives to what has been described as the ‘hollow core’ of the neoclassical account (Lie, 1997; Krippner, 2001). One notable exception is work by Harrison White, whose socio-economic models of production engage directly with established economic theory (White, 1981b, 2002b). White reinterprets economists’ efforts to model forward-looking uncertainty and incomplete information (Knight, 1921; Akerlof, 1970; Spence, 1973) against the sociological precept that actors adjust their own actions to observed behaviors of their peers. In White’s deductive account, markets stabilize as a consequence of the profit-seeking behaviors of ‘tangible cliques of producers observing each other’ (White, 1981b, p. 543). By establishing conditions under which profit-maximization induces producers to select themselves into non-overlapping and therefore uncontested market niches, the socio-economic models present a self-contained argument for the etiology of production markets.

While widely acknowledged as a major contribution to the sociology of markets (e.g. Swedberg, 1991; Fligstein and Dauter, 2007; Collins, 2013), White’s socio-economic models proper have been remarkably little engaged with to date. Although important work has built on White’s presentation of production markets as role structures (for many, see Baker and Faulkner, 1991; Podolny, 1993; Fligstein, 1996; Zuckerman, 1999; Kennedy, 2005), few studies draw explicitly on the typology of markets that White has developed and presented. This lack of engagement with White’s market typology is particularly surprising in light of its noted overlap with the French Économie des Conventions (Favereau et al., 2002; Favereau and Lazega, 2002; White, 2002a; Diaz-Bone, 2010). Despite this interest, studies of specific industries through the lens of White’s account (e.g. Biencourt and Urrutiaguer, 2002; Chiffoleau et al., 2006; Éloire, 2010) have mostly remained without follow-up.

In part, this lack of engagement can surely be attributed to the unwieldy nature of White’s model: Literal application would require fitting a bewildering array of parameters to data in ways that are not only challenging to realize but whose results are also difficult to report on (e.g. cf Leifer, 1985 for one early exploration of the parameter space). Moreover, White’s formal apparatus builds on the premise that profit maximization under demand uncertainty compels mutually observant producers to select themselves into uncontested quality niches. As White put it: ‘The key feature of my approach is this: Firms can observe only volumes and payments, not qualities or their valuations, and they act on the basis of these observations, thereby reproducing the observations’ (White, 1981b, p. 520f., emphasis in original). While an elegant starting point for engaging with economists’ accounts of markets, this profit-maximization assumption places an empirically unrealistic computational burden on producers, who must base their production decisions on the integral of a twice-exponentiated production function (see White, 2002b, pp. 40–43 et passim). And yet, White’s framework seems especially valuable for an economic sociological perspective on markets, for it opens economists’ accounts to distinctly sociological insights into the workings of real markets.

To render the socio-economic models productive for further enquiry, this article turns away from the etiology of markets, so central to White’s original account, to focus instead on how different production regimes incentivize producers to create products of specific qualities. This shift in perspective allows a subtle reinterpretation of the profit-maximization assumption as a practical precept by which to anticipate (other) producers’ behaviors (e.g. Durkheim, 1982 [1895], p. 68). Instead of describing individual producers’ motivations, profitability delineates a producers’ maximal scope for economically viable action. Specifically, the following account distinguishes two levels of incentives on the basis of their time-scale of variation (see Doehne et al., 2024b, for a discussion of time-scale separation as a general, analytical approach). The resulting model distinguishes between the individual producers’ short-term incentive to preserve their reputation for quality and a long-term collective incentive to stabilize a discernible and hierarchically graded conception of product quality. Whereas the former predicts how high-quality producers will respond to a quality-related innovation, thereby potentially legitimizing it in the eyes of others, the latter predicts a quality-related innovation’s overall uptake among the producers of a defined production regime. By identifying how the individual and collective incentives align (or not), the model suggests testable hypotheses about the locus of quality-related innovations in market-driven production regimes.

I test these hypotheses on a narrowly defined quality-related decision faced by winemakers as they bring their product to market: whether or not to close their premium wines under cork or screwcap. In considering this choice, winemakers confront a tradeoff between reducing the incidence of cork-related sensory defects on the one hand and alienating buyers who associate screwcaps with inferior quality on the other. Drawing on the model of quality competition, I hypothesize that and how adoption depends on the winemakers’ position in the quality order of the production regime they are a part of. Analyzing data on 52 880 German wines made by 1028 winemakers operating in three production regimes (commercial, cooperative and elite), I find that German winemakers have, indeed, broadly aligned their behaviors with the model predictions. This suggests not only that their individual behavior is shaped by broader, social contexts, but also that accounting for quality competition stands to enrich our understanding of markets: one and the same mechanism sustains multiple possible outcomes. I conclude with a discussion of four general implications for future socio-economic inquiry into markets.

2. Seeking closure(s): a producers’ perspective on product quality

Sociologists have long argued that quality evaluations on markets are a contingent outcome of social processes. A recurring theme of this literature is that market participants must overcome initially intractable problems of valuation before transactions can take place (Podolny, 1994; Espeland and Stevens, 1998; Beckert, 2009, 2020). To resolve these problems, they rely on institutions, quality schemas and judgment devices that guide and govern their behaviors (Eymard-Duvernay and Eymard-Duvernay, 1989; Callon, 1998; Favereau and Lazega, 2002; Callon et al., 2007; Karpik, 2010; Diaz-Bone, 2018). Markets for fine wines, for example, are steeped in regional appellations that encode elaborate certification processes (Stanziani, 2004; Zhao, 2008; Fourcade, 2012; Carter, 2018), status orders that connect products to their makers (Benjamin and Podolny, 1999; Malter, 2014), and medial discourses that are informed by the verdicts of expert wine critics (Roberts and Reagans, 2007; Hay, 2010; Hsu et al., 2012; Rössel et al., 2018). In short, the criteria by which products are valued on markets are historically evolved, socially conditioned and liable to change over time.

This insight that product valuations are shaped by social processes challenges the basic assumption that prices are neutral mediators of supply and demand. Instead of reflecting the relative demand for specific products, as the canonical neoclassical account assumes, they are inherently shaped by social processes (Baker, 1984; Chiffoleau and Laporte, 2006; Beckert, 2011; Slater and Tonkiss, 2013; Eloire and Finez, 2021). Or, to put it strongly, price is both an economic and a cultural category within the structure of market relations (Slater, 2002). By extension, price depends on perceived quality in the eyes of buyers; and that, in turn, depends on social context. Thus, in effect, prices embody not only products’ relative scarcity but also reflect their social valuation, i.e., their qualification (for many, e.g. Appadurai, 1988; Çalışkan and Callon, 2009; Callon and Law, 2005).

While sociologists have long emphasized the social embeddedness of buyers’ valuations of products, focusing on how tastes, preferences and consumption are shaped by social process and culture, a predominant emphasis on buyers’ valuations arguably neglects the producers’ perspective. This oversight results in a one-sided understanding of producers as mainly responsive to consumer preferences: other aspects being equal, they aspire to make products that best meet consumers’ definitions of quality. This idea resonates with everyday assumptions about quality; after all, we are all seasoned buyers and consumers of products. However, such a buyer-centric view leaves unclear the conditions for market-endogenous change to occur; if and when it does. To overcome this impasse, this article proposes a producer-centric definition of quality and ensuant quality competition on markets. To develop the producer’s perspective in a generalizable way, this article builds on a basic premise of White’s reasoning. Famously, White (1981a,b) argued that although buyers’ reasons for valuing one product over another are inscrutable to the producer, the producers’ ulterior motivation is to secure their economic viability. Interpreting their competitors’ behaviors through this lens allows producers to ease demand uncertainty and adjust their behaviors accordingly. As each producer is observed by every other producer, the choices producers make jointly enact and reproduce the production regime that they are a part of.

In the following, I extend White’s original insight to formulate a producer-centric model of quality competition on markets. This model treats the ‘producer’ as a class of social actor whose behaviors can be predicted on considerations of cost and revenue implications. It thereby leverages White’s emphasis on profit-seeking to shift attention away from the traditional view of firms as price-takers who respond passively to changes in supply and demand. Instead, asserting a firm’s profit orientation emphasizes the interplay between prices, output volume and returns to scale, and socially embedded conceptions of product quality shaping production regimes. In some regimes, for instance, increasing output volume promises economies of scale but lowers buyer-perceived quality. In others, a quality-centered strategy requires costly investments in research and development, potentially leading to higher production costs but also commanding price premiums.

Modeling effects of output volume and quality on profitability can be approached with a level of abstraction, allowing researchers to hypothesize producers’ behaviors in response to marginal changes in the parameter space. The model can be adapted to empirical settings by establishing a ranking of producers on quality that correlates with buyers’ valuations on the one hand and with producers’ unit costs of production on the other. In practice, defining this quality order of production requires making assumptions about how producers’ revenues and costs would vary at different output and quality levels. In competitive settings, such information is usually kept confidential for strategic reasons, making any profitability argument based on individual-level calculations problematic. Rather than build on unreliable estimates of individual producers’ cost and revenue functions, a more effective approach is to infer generic cost- and revenue functions from overall market characteristics. These functions, in turn, can be interpreted under the assumption that producers act as economic agents who adjust their behavior to predicted demand at different quality levels (e.g. Leifer and White, 1987; Chiffoleau et al., 2006; Éloire, 2010; Doehne, 2022). The resultant model predicts producers’ attitudes toward different quality-related courses of action.

As illustration and test, I develop the following account around a specific quality-related innovation that has taken hold in the wine industry in recent decades: the screwcap for premium wines. Winemakers are modeled as producers, that is, as social actors who consider the cost- and revenue implications of their behaviors under conditions of quality competition. Where appropriate, I have relegated formal derivations to technical appendices.

3. Quality competition in the wine industry

3.1 Introduction to the case study: a twist on wine bottle closures

The 20th century was a period of remarkable advancements in winemaking technologies that have led to substantial improvements in the overall quality of wines being made around the world (Loubère, 1990; Halliday and Johnson, 1992). The introduction of refrigerated fermentation tanks allowed winemakers to precisely control the conditions during wine development, modified clonal grape varieties and yeasts provided unprecedented control over natural factors influencing the wine’s character, and increasingly sophisticated filtration, fining and flotation technologies effectively eliminated unwanted deposits, resulting in biochemically stable and technically pure wines (Robinson, 2006, p. 174f.). Each of these innovations are quality-related in the sense that they had a profound impact on both the products being made and on the market positions of the winemakers who embraced them, setting them apart from those who resisted change. Each innovation was also accompanied by intense industry discourse. Yet despite the significance of these advancements, none stirred as much controversy across the industry as the innovation at the center of the following case study: the Stelvin, a long-skirted screwcap for premium wines.

Featured on the covers of industry and trade journals, the screwcap for premium wines is among the most contentious innovations in the wine industry of recent decades (Taber, 2007). Its uptake has sparked intense, industry-wide debate around the role of oxygen in wine development post-bottling, the true incidence of cork-related defects, the environmental impact of synthetic closures and consumer perceptions. Central to this ongoing discourse is the tradeoff faced by winemakers between mitigating cork-related defects on the one hand and addressing buyers’ adverse perceptions of screwcaps on the other: Winemakers who opt for traditional cork closures risk an estimated 2–5% of their product being adversely affected by cork taint and related defects (Simpson, 1990; Pereira et al., 2000). Used on table wines since the 1930s, the chemically inert screwcap eliminates cork taint and related wine defects (such as bottle variation and flavor scalping). Screwcaps are easier to open, they do not require corkscrews and can be resealed; a feature valued by bars and restaurants. Apart from these technical benefits, screwcaps are also cheaper than high-quality cork closures. However, consumers associate screwcaps with low-quality wines.

To the winemaker, the choice of bottle closure thus presents itself as a tradeoff: each individually has a strong economic incentive to adopt a closure that is both cheaper than high-quality natural cork and eliminates cork-related wine defects. However, the screwcap also marks a departure from the culturally laden quality expectations of buyers, and switching closures risks foregone sales. This dual meaning of the bottle closure as both a material aspect of the product packaging and as a symbolic marker of quality creates a unique testing ground for a socio-economic model of quality competition. In the sense of Berger and Luckman, the closures used on particular wines establish ‘enduring indices of the subjective processes of their producers, allowing their availability to extend beyond the face-to-face situation in which they can be directly apprehended’. (Berger and Luckmann, 1966, p. 34). Amidst industry-wide contention, a winemaker’s decision to use one closure on a particular bottle of wine, and by implication not the other, is an endorsement of one of two mutually exclusive value propositions being brought to market. Before considering the winemaker’s tradeoff in detail, the following section first considers how quality competition shapes a producer’s scope for economically viable action.

3.2 Quality and the producer’s scope for economically viable action

From Equations (1) and (2), White has established market conditions under which profit-maximizing selects mutually observant producers into non-overlapping quality niches (see  Appendix A; White, 2002b for an extensive treatment). For each producer, the volume of product sold and revenue generated by each competitor become actionable signals for the key unobservable in White’s model: buyers’ relative valuations of product quality. While elegant, this approach imposes an unrealistic computational burden on producers, who must solve twice-exponentiated functions that integrate the volumes shipped and revenues generated by their competitors (cf White, 2002b). But what if matters are not actually this complicated? What if producers interact with one another and observe and infer meaning from the behaviors of other market participants: critics, buyers, consultants, and others? Assuming that producers do have some sense of how the quality of their product compares to that of their competitors, we can consider how different scenarios will play out in a simplified version of White’s original model.

3.3 The incentive structure of the market

ROI is a multiplicative function of how buyers’ valuations and producers’ costs relate to output volume and quality. The exponents on output volume y and quality n_p identify settings in which it is beneficial for producers to pursue growth- or quality-focused strategies, respectively. As the saturation parameter $γt+1$ connects the operationalization offered in Equation (3) to White’s formal model, it is worth elaborating briefly: White internalizes $γ$ without a time index as part of producers’ profit-maximizing calculus. This ensures finite and actionable proscriptions, (cf White, 2002b, chapter 6), but it also implies that producers perfectly accommodate for the production schedules of their competitors. A more realistic approach is to acknowledge that producers will overproduce and compete for their competitors’ market shares if conditions are right. This is captured by treating $γ$ as an ex-post markdown on buyers’ valuations, that is, $γt+1$⁠. For a given regime, $γt+1$ stabilizes as a discount of revenue on output volume, $R=r·yγta·npb=r·yγt·a·npb$⁠. This simplifies the model while preserving its basic logic (cf  Appendix B).

Overall, the producer’s incentive to make high-quality products depends on how returns to quality (⁠$b/d)$ and returns to scale (⁠$a/c)$ relate to one another at time $t$⁠. Any combination of values for a/c and b/d can be represented as a point on a plane with a/c on the abscissa and b/d on the ordinate. This is the market plane as White has presented it (e.g. 1981; 2002). As total costs and revenues both increase with output volume, the plane is bound to the left by a/c ≥ 0. While price increases with quality, $b≥0$⁠, the cost of producing quality, d, depends on what buyers value. For example, if buyers value the social confirmation of products that are also bought by many other buyers (as example, White (1981b) discusses disposable diapers), economies of scale can make it cheaper to produce high quality. As a result, d can take on values smaller than zero, and b/d can be positive or negative.

Four boundaries, at $b/d=0$⁠, $a/c=b/d$⁠, a$/c=1$ and $a/c=1/γt+1$⁠, define regions of the market plane that differ in terms of the incentives they offer producers. Each region implies a distinctive regime, or set of logical rules, which drives the actions of profit-seeking producers. White has given names to six regions: ordinary, unraveling, crowded, explosive, trust and paradox (White, 2002b). The scope conditions for sustained and profitable action have been developed elsewhere (Leifer, 1985; Leifer and White, 1987; Bothner and White, 2001; White, 2002b). Here, I consider how these regimes differ in terms of the emphasis they place on quality competition and how quality competition in turn shapes the locus of quality-related innovations.

For a given production regime, the intensity of quality competition can be read off its position on the vertical axis of the market plane, $b/d$⁠. The canonical scenario of pure competition obtains for diminishing returns to scale (⁠$a/c<1$⁠) and $b=0$⁠: buyers do not remunerate differences in quality. The producer’s sole incentive is to minimize production costs at requisite quality (see Burt, 1992, pp. 198ff. for an insightful discussion). The higher a regime locates on the $b/d$ axis, the greater the producers’ incentive to compete on quality (other aspects equal). The hatching in Figure 1 maps the intensity of quality competition: Regimes that locate in regions with diagonal hatching are marked by intense quality competition; regimes in regions with vertical hatching prioritize scale-based production. Specifically, if a regime locates above the main diagonal, that is, $b/d>a/c$⁠, then the incentive to compete on quality is greater than that to maximize output volume, or scale. This is the case for the unraveling and crowded regimes.

Figure 1

The market plane.

Note: Regions with diagonal hatching exert an overriding collective incentive to compete on quality; regions with vertical hatching exert a collective incentive to compete on price. Dark regions identify regimes on which absolute profits increase with marginal increases in quality; light regions identify regimes on which cost-efficient production is paramount (for $γt=1$ and $r,q>0$⁠). Returns to quality and returns to scale are both increasing in the cross-hatched crowded region, incentivizing producers to overproduce and compete for market share.

Open in new tabDownload slide

The shading of regions in Figure 1 projects these inequalities onto the market plane: dark shading identifies regions in which absolute profits increase with marginal increases in quality. High-quality producers have a strong incentive not to jeopardize their quality position n_p in unraveling, trust, or paradox regimes. These are settings in which producers’ incentives to produce quality are broadly aligned with buyers’ conceptions of quality: Other aspects being equal, high-quality producers will act to preserve established quality expectations. By comparison, producers in ordinary or crowded regimes are less focused on quality. Instead, in these regimes, profitability hinges on producers realizing economies of scale.

Table 1 summarizes how the different production regimes drive quality competition. It identifies a collective, regime-wide incentive to compete on quality and an individual, producer-level incentive to endorse (or reject) quality-related innovations. The former predicts overall uptake whereas the latter predicts whether high-quality or low-quality producers will be the first to adopt. Together, these two characteristics predict the quality-related behaviors of producers conditional on (a) the production regime they are a part of, and (b) their position in the respective quality order of production. This sets up the general framework for modeling the uptake of quality-related innovations in defined production regimes. Next, I apply this framework to the winemaker’s choice of wine bottle closure.

3.4 The winemaker’s choice of bottle closure (hypotheses)

To a winemaker, the choice of bottle closures comes as a stylized tradeoff: on the one hand, a producer who switches from cork to screwcap stands to eliminate cork taint and other cork-related wine defects. In doing so, however, that winemaker risks being downgraded in the eyes of consumers if other winemakers do not switch to screwcaps. So when, if ever, will a winemaker take that crucial first step of bringing their wines to market under screwcap? The answer to this question depends on the production regime: In some settings, buyers hardly remunerate differences in quality anyway. In others, profitability depends on being perceived as a maker of high-quality products. The relation between the production regime and the individual winemaker’s decision situation can be broken down into three parts: (a) Would buyers accept the screwcap if it is broadly adopted, or would they turn to alternative offerings instead? (b) Would widespread screwcap adoption favor large producers over small producers? (c) Would widespread screwcap adoption favor low-quality producers over high-quality producers or vice versa? The answers to these three questions shed light on the locus of quality-related innovations in production regimes.

The first question addresses the possibility that buyers will simply exit the producer’s market in search for substitutes that better meet their quality expectations. For example, they may purchase a wine from a region that continues to use corks (e.g. France or Italy), or they may turn to a different product entirely (e.g. chocolates, flowers or beer). The resulting decline in demand triggers a crowding-out effect as firms are forced to offload their surplus production at a discount, $γt+1<1$⁠. This amounts to an outward shift of the boundary separating a crowded from an explosive production regime: winemakers find themselves competing viciously for shares of a potentially much smaller market. Indeed, the first attempt to introduce screwcaps on premium wines arguably failed for exactly this reason. From 1977 to 1982, 20 million bottles of expensive Australian wine were brought to market under screwcap, a step lauded by industry commentators at the time (Rubin, 1980). Despite favorable media coverage, consumers rejected the closure, few of the wines were sold, and by the mid-1980s, all winemakers had reverted to cork (Marks and Mortensen, 2003). The lesson learned by the innovating firms at the time was that unless producers can prevent buyers from exiting their niche, for instance by convincing them of the benefits of screwcaps, the innovation cannot take hold.

The second question concerns how widespread screwcap adoption would affect returns to output volume. As each bottle sold requires one closure, one might expect large producers to benefit most from switching to screwcaps. On examination, however, the tin capsules atop most wine bottles hide inferior, fault-prone synthetic stoppers which cost far less than the premium long-skirted screwcaps that are used on premium wines (Hulot, 2009). Therefore, large producers who use cheap but fault-prone synthetics do not stand to save significantly more by switching to screwcap than their smaller competitors. At best, they can substitute a cheap screwcap for an inferior synthetic closure, thereby eliminating cork-related wine defects. For our purposes, the short-term effect of screwcap adoption on returns to volume (a/c) can therefore be treated as negligible. Effectively, the producer’s choice of bottle closure is grounded in quality deliberations, not quantity.

The third question considers how widespread screwcap acceptance would affect individual and collective returns to quality, as indicated by the regime’s position on the b/d axis of the market plane. Here, the quality tradeoff confronting winemakers fully comes into view: In principle, the inner lining of the industrially produced screwcap can be finetuned to mimic the seal of a natural cork of high quality but at a fraction of the price. Producers who use expensive corks could therefore realize substantial cost-savings by switching to screwcaps without compromising the sensory quality of their product. On the other hand, producers who use cheap but fault-prone closures (plastic stoppers, agglomerates, etc.) at the outset can improve the sensory quality of their wines at little to no extra cost. In sum, widespread screwcap acceptance lowers the cost of making wines of high quality for all producers, thereby lowering $d$⁠. Assuming, as argued above, that screwcap adoption does not significantly alter returns to output volume (⁠$a/c$⁠), this entails an upward movement along the b/d axis of the market plane (cf Figure 1). Any such movement that is driven by a reduction in the denominator d will be more pronounced the greater the numerator b is. Consequently, the collective incentive to endorse screwcap use increases with the overall intensity of quality competition, that is, the greater $b/d$ is at the outset. This predicts that

Hypothesis 1: Screwcap adoption on premium wines is higher in unraveling and crowded regimes than in ordinary and trust regimes.

While the collective incentive to endorse screwcaps depends on the regime’s position along the $b/d$ axis, this effect may yet be countered by the individual producer’s incentive. This in turn depends on a producer’s position in the quality order, $np$⁠. The inequalities in Equation (6) imply that buyers’ perceptions of quality matter particularly in unraveling, trust and paradox regimes, whereas quality is of subsidiary importance in ordinary and crowded regimes. The high-quality producers’ individual incentive therefore predicts that

Hypothesis 2_a,b: Screwcap adoption is higher among high-quality producers than among low-quality producers in (a) ordinary and (b) crowded regimes.

Hypothesis 3_a,b,c: Screwcap adoption is lower among high-quality producers than among low-quality producers in (a) unraveling, (b) trust and (c) paradox regimes.

In the following, I test these hypotheses with data on 52 880 German premium wines made between 2005 and 2015. The analysis proceeds in three steps: first, I assign each winemaker in the dataset to one of three distinct production regimes that characterize the German wine industry. Second, I fit statistical models to estimate the quality position of each winemaker based on the price and expert-evaluated sensory quality of their offering relative to their competitors’ offerings. Finally, I estimate the joint effects of the production regime and the producer’s position in the quality order of production on screwcap adoption.

3.5 The production of German premium wines

With 100 000 hectares (ha) of land under vine, German winemakers produce 8–10 million hectoliters (hl) of wine each year, placing the country eighth in global production by volume. However, this production does not cover domestic demand. With annual imports of 12 million hl, Germany is among the world’s largest markets for wine imports (Anderson et al., 2017). At the turn of the century, almost three quarters of these imported wines were being sold through Germany’s highly developed discount retailer infrastructure (Storchmann and Schamel, 2004). Selling wholesale at less than €2.00 on average (ca. 2.50 USD at the time), this great volume of cheap imported wines exerts perpetual downward pressures on prices. Overall, price competition is vicious and driven by market saturation and stagnating demand.

The official classification of wine quality does little to protect German winemakers from the pernicious effects of price competition (Weik, 2016). Codified into wine law in 1971, it distinguishes eight grades of quality based on the sugar content of the grape must from which a wine was made (in ascending order of quality: Tafelwein, Qualitätswein bestimmter Anbaugebiete (QbA), Qualitätswein mit Prädikat, Kabinett, Spätlese, Auslese, Beerenauslese, Trockenbeerenausles and Eiswein). The three highest categories identify specialty wines that are made only under particular circumstances. While any winemaker can make wines of the highest designation in principle, linking quality to an unambiguous metric favors large producers for whom it is easier to produce wine of requisite specifications at low costs (Halliday and Johnson, 1992).

Together, these factors denote a crowded regime on which producers compete for market share by underbidding and driving each other out of the market $(b/d<a/c<1/γ)$⁠. In line with this interpretation, the German wine industry has been consolidating for decades. From 1979 to 2020, the number of wineries has declined from 51 709 to 16 394 even as the number of large producers has doubled, and the landholdings operated by large producers have tripled (Table 2).

Historically, there have been two broad organizational responses to the ongoing industry consolidation: the emergence of large-scale winemaking cooperatives in the mid-19th century, and the founding of the elite Verband Deutscher Prädikatsweingüter (hereafter: VDP, or Elite), an association of premier wine estates, in 1910. Cooperatives and the VDP estates operate in regimes that differ markedly from the crowded conditions of commercial winemaking.

At the turn of the millennium, a third of German wine production by volume was being handled by around 200 cooperatives which process wines from grapes that have been grown by their members (Robinson, 2006). Founded for the purpose of bundling the activities of grape-growers whose landholdings were too small for independent operations, cooperatives must compromise between quality aspirations and cost-efficiency of production (Hanf and Schweickert, 2014; Santos-Arteaga and Schamel, 2018). Generally, this compromise tends toward cost-efficient production (Schamel, 2015; Frick, 2017). With returns clearly driven by scale rather than quality of production, cooperatives are firmly rooted in a trust regime (⁠$a/c > 1 > b/d$⁠) that offers consumers reliable quality at competitive prices.

At the upper end of the market, a group of premier wine estates differentiate their product on quality. Established as a union of four regional associations committed to making unchaptalized wines, the VDP is an umbrella association for around 200 premier German wine estates. It applies its own, privately maintained quality classification which resembles the terroir-based system of Burgundy, France, and which differs fundamentally from the official quality classification (Rössel and Beckert, 2013). Membership in the VDP is strictly regulated and requires adherence to self-imposed standards that exceed those prescribed by law, including yield restrictions and on-site inspections. VDP estates cultivate 5% of the area under vine, but self-imposed constraints limit output to 3%. Nonetheless, they account for around 7.5% of domestic production by revenue. This quality-focused strategy identifies VDP estates as firmly rooted in an unraveling regime (b/d > a/c < 1) that clearly emphasizes quality over output volume.

With three production regimes identified in the crowded, trust and unraveling regions of the market plane, we can use data on the bottle closures used on particular wines to examine how the producer’s screwcap adoption behavior differs by regime and quality.

4. Dataset

Data for the following analyses have been provided by Weinplus, one of Europe’s largest independent wine-testing organizations. Founded in 1998, Weinplus offers winemakers expert sensory evaluations of their wines. Wines are tasted blind and in flights of the same vintage, region, and style. Wine quality is scored on a scale from 50 to 100, with 5 points awarded for optics, 15 points for the nose, 25 points on the palate and 5 points on overall impression. Wines which score 80 points or more on quality are also advertised to a base of 65 000 website subscribers, including oenophiles and many restaurateurs. The anonymized dataset includes wines which scored fewer than 80 points.

In March 2005, weinplus began recording the closures used on incoming wines. In total, data has been collected on 53 840 red and white still German wines by 1227 producers that were tested from March 1, 2005, to December 31, 2015. As the following analyses infer the producer’s quality niche from the prices and sensory quality of producers’ non-specialty wines, I excluded 199 producers with fewer than ten such wines from the analysis. The final dataset contains 52 880 still red and white premium wines of vintages 1993–2014 by 1028 winemakers. It includes 195 of the approximately 200 member estates of the VDP (accounting for 4796 ha of land), 52 cooperatives (accounting for 9236 ha) and 735 commercial wineries (accounting for 9598 ha). Together, these wineries account for a quarter of the total area under vine in Germany; coverage of the premium wine industry is essentially complete.

5. Variables and measurement

5.1 Dependent variable

The dependent variable of interest is whether a particular wine has been closed under screwcap. Although the database also identifies synthetic stoppers (technical, agglomerate or plastic), at the point of sale, all internally sealing closures are usually concealed under a capsule and therefore not visible to the consumer prior to opening the bottle. To capture the winemaker’s decision as it presents itself at the point of sale, I binarized the data to distinguish between wines under screwcap and wines not under screwcap.

5.2 Main explanatory variables

The model of quality competition predicts that screwcap adoption will vary systematically by production regime and the producer’s position in the quality order of production. Three relevant regimes have been characterized in previous sections: a crowded regime of commercial producers, a trust regime of cooperatives and an unraveling regime of VDP estates.

Two wine-level measures lend themselves to assigning producers to positions in the quality order of production: the prices of their wines, as reported by the winemaker, and expert evaluations of the sensory quality of a producer’s portfolio of wines as established in blind comparative tastings. To control for differences in the types of wines made by different winemakers, I fit two multivariate regression models with random intercepts added to estimate producer-level variation on price and quality. Both models control for the wine’s vintage, the age when it was tested, its official quality designation, the main grape variety used, and its alcohol volume. To control for systematic variation in sensory quality assessments between experts and over time, I included crossed random intercepts for each testing event (2530 in total). Using the generated coefficients, I predicted the asking price and sensory quality of a hypothetical, standard-sized bottle of 2010 QbA Riesling wine of average alcohol volume for each producer. The inferred producer-specific price estimates for such a hypothetical wine range between €3.32 and €32.40 and between 75 and 88 points on quality. Figure 2 depicts each of the 1028 producers’ scores on either measure.

Figure 2

Predicted price and sensory quality of representative wines for each producer.

Note: Each point represents a producer.

Open in new tabDownload slide

In Figure 2, each point represents the price and quality level of the portfolio of non-specialty wines made by one producer. The nine-fold span on price seems empirically realistic and indicates that buyers are willing to pay substantial premiums for some producers’ wines. While it cannot be ruled out that some producers have submitted aspirational prices to weinplus in the hopes of signaling high quality (Askin and Bothner, 2016), the fact that weinplus communicates those prices to their readership creates an incentive to communicate realistic prices. Moreover, at r = 0.72, the correlation of producer-inferred quality by either metric is strong (far stronger, in fact, than the wine-level correlation between wine price and sensory quality, at r = 0.43). This supports the claim that producer’s quality niches inferred from price are related to the sensory quality of their wines. I take the ordering of producers on either metric as a proxy for their respective position in the quality order. To facilitate comparisons, I standardized both measures so that each producer’s score is scaled in standard deviations from the average score set to zero (z-scores).

5.3 Control variables

Producer-level differences in screwcap adoption rates can have many causes. For example, many advocates of natural cork concede that the screwcap is effective for white table wines that are not meant to be stored over long periods of time. Winemakers who specialize in such wines would appear more open to adopting screwcaps than they actually are, while specialists in age-worthy red wines would seem more conservative. To account for differences in types and styles of wines, all reported models control for the wine’s alcohol content, bottle size, main grape varietal, official quality designation, price, expert-assigned quality score and aging potential as inferred from the timespan between a wine’s vintage and the date at which each wine was tested.

6. Analysis

The following analyses assess the likelihood of a wine being closed under screwcap conditional upon its producer’s position in the quality order of production and the production regime in which it was made. The data are partly nested and not balanced over time: Each wine was brought to market by a particular producer, p, and in a particular year, t, but some producers submitted wines annually, while others are represented in the dataset only in three or more years. This dependence structure calls for a multivariate, multilevel, regression design with crossed random effects to control for clustering of observations at the level of producers and years.

7. Results

In the investigated period, the overall share of wines closed under screwcap in the database increased from 6.0% in 2005 (n = 2950) to 47.9% in 2015 (n = 2165). Table 3 presents bivariate correlations and other descriptive statistics. In line with the common narrative that screwcaps indicate industrially produced wines of inferior quality, screwcapped wines tended to score lower on quality-related variables (price, age at test and expert quality evaluation).

However, these descriptive statistics leave unclear how adoption varied by production regime and quality niche. To this end, Table 4 reports on two multivariate logistic regressions that consider different drivers of screwcap adoption.

Both models reported in Table 4 are identical with one exception: Model 1 infers producer quality from the markup on price that buyers are willing to pay for its product relative to the products of its competitors, while Model 2 infers producer quality from relative differences in the expert sensory evaluations of wines. All models control for the year in which a particular wine was brought to market, the main grape variety used in its making, and its official quality designation. Both models yield consistent outcomes, and both models were able to correctly predict more than 91% of closures used. As it generalizes to settings in which expert quality evaluations are not available, I focus the discussion of results on the price-based Model 1 while noting that Model 2 yields consistent findings based on wine sensory quality.

The first set of coefficients breaks down wine-level predictors of screwcap adoption. In line with the bivariate correlations, screwcaps are used on inexpensive white wines of modest quality and with limited aging potential. While consistent with the broader narrative that screwcaps indicate wines of inferior quality (Taber, 2007), this does not offer deeper insights into the hypothesized producer- and regime-level drivers of adoption.

At the producer level, the producers’ size of landholdings does not have a significant effect in either model. This finding is consistent with the observation that scale-based benefits to screwcap adoption are limited as large producers use cheaper, albeit fault-prone, closures from the outset. The coefficient on quality niche is significant and positive, indicating that high-quality producers were more open to screwcaps than low-quality producers. However, this finding must be interpreted with caution, as the reported models also control for wine price, which is strongly negative and correlated with producer quality. As robustness test, I therefore ran additional models without wine-level controls for price. Before turning to these models, however, it is instructive to first consider the main hypothesized effects related to quality competition.

In broad support of Hypothesis 1, both models indicate that, ceteris paribus, screwcap adoption was lowest in the trust regime of cooperatives and highest in the unraveling regime of VDP estates. While the difference in adoption levels is not statistically significant between the trust regime and the crowded baseline, producers in the quality-focused unraveling regime have broadly adopted screwcaps on premium wines. Substantively, this is plausible, as the problem of cork taint is particularly pernicious for elite winemakers whose success depends on their reputation, as is certainly the case for VDP wineries. Many of the VDP members rely on established relations with high-end restaurants and wine stores to sell their expensive product, and repeated complaints due to faulty cork quickly result in a delisting of the winery and foregone sales.

Notably, the interaction between quality niche and production regime suggests that the behavior of high-quality producers in trust and unraveling regimes differs systematically from that of producers in the baseline crowded regime. While compatible with the model predictions, as interaction effects, these findings must be seen relative to the baseline behavior of commercial producers in the crowded regime. To facilitate interpretation, Figure 3 depicts how the predicted probability of a standard wine of average quality being closed under screwcap varied by production regime and the producers’ position in the quality order of production in 2010.

Figure 3

Predicted screwcap adoption by quality position and production regime.

Note: Predictions for a median-priced (€7.50), standard-sized QbA Riesling of average sensory quality, aging potential, and alcohol volume in 2010. The producer’s position in the quality order of production has been transformed into standard deviations from the mean quality score (z-scores).

Open in new tabDownload slide

Figure 3 presents striking evidence in support of Hypotheses 2_b and 3_{a, b}: It suggests that high-quality producers in the crowded regime of commercial winemakers were more likely to use screwcaps than their lower-quality peers (Hypothesis 2_b) while high-quality producers in the unraveling and the trust regimes were markedly less likely to adopt screwcaps for their expensive than their lesser-quality peers (Hypotheses 3_{a, b}).

While consistent with the predictions of the model of quality competition, these findings may yet prove to be an artifact of the statistical design. After all, the producer’s quality niche is established on the basis of representative prices and both Models 1 and 2 include strongly negative wine-level covariates for price and quality; characteristics that VDP wines in particular score highly on. To rule out that the reported findings are artefactual, Table 5 considers three additional scenarios as robustness tests. To preserve space, control variables that are not relevant to the following discussion have been omitted (all are consistent across Models 1–5).

Model 3 is identical to Model 1 with one exception: it drops the interaction effect between quality position and production regime. It thereby identifies the collective incentive arising from quality competition, which remains positive and significant for producers in the unraveling regime but non-significant for the trust regime (relative to the crowded baseline). Model 4 drops the wine-level variables for price and sensory quality from Model 3. The producer-level coefficient for quality niche switches from positive to negative but again the collective incentive is robust: ceteris paribus, screwcap endorsement remains highest in the unraveling regime (and significantly higher than in the crowded baseline). Finally, Model 5 adds the interaction between production regime and quality niche to Model 4. Model 5 therefore differs from Model 1 only in that wine-level effects of price and expert quality score are not considered. Again, the interactions that test the hypothesized effects are robust: high-quality producers in the unraveling regime were markedly less likely to adopt screwcaps than their peers in other regimes. To be sure, an AIC comparison between Model 5 and Model 1 identifies price as an important predictor of screwcap adoption. But the main takeaway is that the hypothesized quality- and regime-level incentives are robust across model specifications.

Overall, both sets of analyses indicate that the likelihood of a wine being closed under screwcap depended not only on wine-level characteristics such as color, style, wine price, quality designation or aging potential, but also on the overall structure of the production regime in which it was made. Specifically, and in line with the hypothesized relations, the collective incentive to use screwcaps was highest among producers of the unraveling regime, where winemakers are already operating at the upper limit of what buyers are willing to pay. When accounting for the types of products made, the collective incentive to embrace screwcaps was lowest among the cooperative producers in the trust regime, where output volume and scale economies are the main drivers of profitability. In contrast, the individual incentive to preserve a high-quality reputation was most pronounced in the unraveling and trust regimes, whereas high-quality producers of the crowded regime have broadly embraced screwcaps. Across the three regimes, we thus observe broad support for the hypothesized effects of quality competition on individual winemakers’ behaviors.

Having reported these effects in isolation, it is instructive to consider how adoption in one regime may have affected adoption in the others. Although the three regimes are logically distinct, with producers assigned to one or another, the regimes do partly blend together at the point of sale. An isolated perspective that considers only the individual production regimes therefore seems incomplete. For instance, several larger VDP wineries distribute some of their affordable wines through retailers that also carry commercial wines and the products of select cooperatives. What might this imply for quality competition at the intersection of overlapping production regimes?

In general, credible challenges to established conceptions of quality are best issued by actors who are recognized for a high standard of quality by some measure of relevance. This points to the VDP as representative of the German wine elite. The VDP’s collective incentive is to support widespread screwcap adoption as a means by which to lower the spread on costs of quality, d. However, the high-quality producers’ individual incentive is not to jeopardize their position in the quality order. As they are best equipped to legitimize the screwcap in the eyes of buyers, their reticence may well override the collective incentive. A possible answer is obtained when considering the role of lesser-quality VDP producers in the diffusion of screwcaps. As Figure 3 indicates, the lowest-quality VDP producers occupy quality positions that are on a par with average commercial producers; their quality niches overlap. This likely exposes the lesser-quality VDP producers to quality competition from the crowded regime. At the same time, their membership in the VDP and the VDP’s reputation for uncompromising quality gives these producers the cachet to legitimize the use of screwcaps on their wines, thereby enabling high-quality commercial producers to adopt screwcaps on theirs. In line with this reasoning, the preceding analyses have identified the unraveling VDP-regime and the lesser-quality VDP producers in particular as the primary locus of screwcap adoption at the market interface. This observation supports a more general claim that overlapping quality orders will co-evolve (White, 1993). While anecdotal evidence is consistent with this observation, it remains speculative as data on the actual degree of regime overlap is not available. While the elaboration of this dynamic must be left to future research, it adds a dynamic dimension to quality competition based on the super-positioning of multiple quality orders of production.

8. Discussion

Overall, the extent to which German winemakers have adopted screwcaps aligns with predictions of the presented model of quality competition. In the period under investigation, the likelihood that a particular wine was brought to market under screwcap depended not only on that wine’s style, price, quality designation or region of origin, as the prevalent narrative suggests, but also on the context of the production regime in which it was made. This finding has implications on at least four levels. First, it sheds light on the diffusion of quality-related innovations that disrupt the established production order, highlighting the blending of production regimes and the interplay between stability and change as promising areas for future research. Second, it underscores the social embeddedness of the quality order of production markets in general and of unraveling production regimes in particular. Third, it emphasizes the value of refining our understanding of diverse social processes that influence the quality order of production, with potential applicability to other industries. Lastly, the typology of production regimes offers a refined basis for normative evaluations of how quality competition shapes market outcomes.

On the first point, the preceding analyses have identified an important locus of screwcap adoption among the lower-quality producers of the VDP, who have embraced the screwcap even as their high-quality VDP peers have remained reticent. This finding aligns with the presented model of quality competition: It identifies the unraveling regime as a setting in which the high-quality producers’ individual incentive to preserve the status quo conflicts with the collective incentive to lower the cost of producing high quality. In this constellation, quality-related innovations arise from among the lower tiers of the quality order of production, effectively eroding established conceptions of quality that high-quality producers are keen to preserve.

On the second point, the general observation that unraveling regimes incentivize high-quality producers to endorse innovations that undermine the status quo relates back to White’s formal account. In White’s original assessment, the unraveling region of the market plane could not actually sustain markets, as returns to quality motivate producers to encroach on the niches of their competitors. In his seminal Where do markets come from? (White, 1981b), for example, the unraveling region was rejected as a failed region incapable of sustaining markets as self-reproducing social structures. In the same vein, White (1981a, p. 45) singled out the prediction that markets would fail to realize in the unraveling region as one of the most surprising of his approach. Later, describing the ‘topology of reproducible structures’, Leifer and White (1987, p. 98) concluded of the unraveling region that ‘in these circumstances, we expect that markets do not appear’. However, this interpretation had changed by 2002, with White identifying particular markets as exposed to unraveling, such as Manhattan stores beset by the large supermarkets of suburban New Jersey (White, 2002b, p. 83) or the French Vins de Pays wine industry (White, 2002c). By this time, White had singled out unraveling regimes as a particularly promising context for sociologically informed enquiries into the social efficiency of markets (White, 2002b, p. 82).

More generally, acknowledging that markets may locate in one or any of the failed regions of the market plane draws attention to a basic economic-sociological insight: that markets are embedded into society more broadly. Market stability is hardly a pure outcome of the profit-maximizing behavior of producers, as White’s formal account requires. Instead, what appears impossible from a viewpoint that is predicated on producers’ profit-maximizing efforts may yet be sustained by other factors, such as commonly held narratives and cultural framings (White, 2000; Mützel, 2009; Godart and Claes, 2017), established customer relations (Chiffoleau and Laporte, 2006), conventions (Favereau et al., 2002), local social contexts (Doehne et al., 2024a), deep-seated principles of evaluation (Boltanski and Thévenot, 2006 [1991]; Diaz-Bone, 2005), state interventions (Fligstein, 2001; Doehne et al., 2023) or evaluative schemas that stabilize valuations of product quality at the producer’s level (Lynn et al., 2009; Hsu et al., 2012; Malter, 2014). Acknowledging that market participants compete not only on price but also on quality entails that producers actively shape buyers’ perceptions to improve their market position (Musselin and Paradeise, 2005; Beckert and Musselin, 2013; Musselin, 2018). The presented model operationalizes the time-varying outcome of such quality competition in the relative performance of producers and their changing positions on the quality order of production.

The task of disentangling the social processes that shape quality competition requires further investigation into the construction, stabilization and dissolution of production regimes and their underlying quality orders. This article has tested quality-related hypotheses derived from general assumptions about the cost structures and revenue function of producers operating in three distinct production regimes. The quantitative analysis supports the feasibility of this heuristic approach to locating production regimes on the market plane. However, it remains to be seen whether similar findings apply to other quality-related innovations in and beyond the winemaking domain. As such research develops, the analytical strategy employed in this article could be enhanced by ethnographic approaches that examine the narrative construction of the quality order and its negotiation. White’s elegant formal account rests dormant for the wrong reasons: its mathematical rigor comes at the expense of an unwieldy set of parameters needed to uphold the fundamental premise that profit-maximizing induces mutually observant producers to enact a self-sustaining market structure. For today’s world of increasingly singularized products, the challenge is to map the ever-evolving quality order of production onto the framework.

A third implication relates to producers’ incentive to make and bring high-quality products to market. As consumers, we are accustomed to considering quality as an inherently desirable property in a product, a view that also informs cognate concepts such as status. In general, having high status is considered preferable over having low status (Sauder et al., 2012; Piazza and Castellucci, 2014). The presented model of quality competition suggests a more nuanced account. While it may be beneficial for producers to be known for making products of the highest quality in some contexts and regimes, under other conditions, it can be far more profitable to bring large volumes of modest quality to market. A burgeoning literature is developing this insight (Kovács and Sharkey, 2014; Castellucci and Podolny, 2016). More generally, linking economic outcomes to positions in the quality order establishes an entry point for a nuanced assessment of the efficacy of market structures, the valuations that sustain them, and the dynamics of quality competition they foster.

Lastly, the model of quality competition presented here suggests that what is commonly thought of in terms of one monolithic institution, the market for a particular type of product, can in fact sustain numerous distinct production regimes that each incentivize production in fundamentally different, yet ultimately predictable, ways. Studies of markets and of market-related phenomena stand to be enriched by a correspondingly differentiated vocabulary; the market plane offers a basic taxonomy to which to turn for fitting adjectives. By drawing attention to the varied conditions under which producers are incentivized to endorse or discredit quality-related innovations, it is hoped that the framework laid out in this article will inform future research into how markets shape the behaviors of those who make and bring products to them.

Acknowledgements

I am indebted to Harrison White for his support and encouragement in early stages of this project. I thank Utz Graafmann of wein-plus for providing the anonymized dataset and the three anonymous reviewers for their careful and constructive engagement with the argument.

Funding

This research was supported by the FAZIT-Foundation.

References