Status, Quality, and Social Order in the California Wine Industry
Beth A. Benjamin
Over the last ten or fifteen years, sociologists have given considerable attention to the role of market ties as conduits for flows of information and resources. The basic argument is that market ties represent channels for the transmission of goods and services valued by market actors. Examples of this network-based approach include research on interpersonal networks and job search (Granovetter, 1974; Fernandez and Weinberg, 1997), the diffusion of information through interlocking directorships (Davis, 1991; Haunschild, 1994), and market models emphasizing how a firm’s position in the flow of payments and resources affects its autonomy (Burt, 1980, 1992). More recently, sociologists have suggested that ties are also important conveyers of identity in that an actor’s relations in a market influence how others perceive the actor (e.g., Baum and Oliver, 1991). One stream of research emphasizing the role of networks as conveyers of identity is research on the role of status in the marketplace (Podolny, 1993, 1994; Han, 1994; Stuart, Hoang, and Hybels, 1999). In this work, a market exchange is viewed, at least in part, as an act of affiliation whereby the average status of a market actor’s affiliates influences perceptions of the actor and thus the flow of payments, resources, and opportunities available to that actor.
In this paper, we seek to advance this line of research by examining status more completely. Specifically, our work extends previous research in three ways. First, we show that an actor’s status position influences the quality at which the actor chooses to produce, as well as the economic returns the actor derives from producing at a given quality. While prior research on market status has regarded the effects of status and quality on market outcomes as largely independent, in this study we examine how the affiliations constituting an actor’s status position actually affect the actor’s choice of quality and its subsequent returns. Second, previous research into status dynamics in both market and nonmarket settings has at best employed extremely imperfect correlates of quality (e.g., Podolny, 1993). In this study, we examine status relations in the California wine industry – a context that allows us to include more direct measures of past and present quality because of the substantial amount of time and attention the industry devotes to measuring quality differences across products. Finally, though previous research has revealed a certain amount of stability in status over time, to date there has been little if any research that has documented a mechanism by which this inertia occurs. In this study, we clarify a mechanism underlying the reproduction of the status ordering by demonstrating how an actor’s current affiliations affect and constrain returns to subsequent affiliations.
AFFILIATION, STATUS, AND QUALITY
It is well accepted that consumers’ expectations about the quality of a producer’s products determine – at least in part – the flow of payments and resources that the producer receives, but what determines such expectations in the first place? The quality of past offerings certainly provides one source of expectations. Economic models of reputation, for example, often emphasize the importance of past quality (e.g., Shapiro, 1983; Allen, 1984; Wilson, 1985), as do some sociological models (Raub and Weesie, 1990; Kollock, 1994). Expectations about quality also derive from affiliations that market actors develop through their exchange relations. As market actors enter into exchanges with other actors, they often become identified with one another. For example, when a firm enters into an ongoing exchange relation with an auditor, the auditor’s status affects how others perceive the firm (Han, 1994). Similarly, when a hospital establishes and publicizes relations with a well-respected agency or donor group, it increases the perceived quality of its services (Perrow, 1961). And when a young firm affiliates with a more established community group or other well-known organization, it may increase the perceived legitimacy of its activities and thus its chances of survival (Wiewel and Hunter, 1985; Baum and Oliver, 1991).
One conception of the market that draws attention to both past demonstrations of quality and affiliations as a basis for expectations is the status-based model of market competition (Podolny, 1993; Podolny and Phillips, 1996). According to this model, a market actor’s status has a dual foundation in both its past demonstrations of quality and the status of its exchange partners. The actor’s own status, in turn, has a positive impact on a number of market outcomes, such as market share (Podolny, Stuart, and Hannan, 1996), the spread between costs and price (Podolny, 1993), influence over the direction of technological innovation (Podolny and Stuart, 1995), and the reactions of the financial community (Stuart, Hoang, and Hybels, 1999).
Although this earlier research has acknowledged the dual influence of both past quality and status, for the most part these factors have been treated as independent determinants of market outcomes. In addition, status has rarely been examined with anything more than very indirect proxies for past quality. Podolny’s (1993) examination of the status-price relationship, for example, relied on measures of a firm’s past market presence to control for the effects of prior quality. Similarly, Podolny and Stuart’s (1995) examination of the effects of status on innovation used occurrence dependence terms to control for quality differences across inventions.
Whereas research on status in the market context has focused primarily on the independent effects of status and past quality, research on status dynamics in nonmarket contexts has moved beyond the study of independent effects to examine how these factors work together. This research has begun to focus on joint effects, looking at how status influences returns to past quality. A number of scholars have argued that higher status affiliations help to increase returns to a given quality of output. This proposition is perhaps most clearly articulated in the sociology of science. Latour (1987) and Camic (1992) both asserted that a scientist is more likely to receive favorable evaluations of a given quality of work to the extent that he or she is able to affiliate the work with the efforts of high-status others. Similarly, in sociological studies of education, Bowles and Gintis (1976) maintained that a child’s affiliation with a particular social class, rather than the child’s past educational achievement, shapes expectations about his or her intellectual ability. Finally, in community studies, sociologists have frequently observed that evaluations of an individual’s moral character have less to do with the individual’s prior acts of morality than with the status of the company he or she keeps (e.g., Elias and Scotson, 1965).
In these and other contexts, holding all else constant, the status of one’s affiliations should increase returns to an actor’s past demonstrations of quality in the market. Similarly, returns to quality in the market should decrease to the extent that the market actor does not affiliate with high-status others. There are two possible processes by which this might occur. First, status distinctions across actors may act as a screen or filter, drawing attention away from market actors with lower-status affiliates and toward actors with higher-status affiliates. If the likelihood of observing a market actor’s investment in quality increases with the status of the actor’s affiliates, then those with lower-status affiliations will generally find it more difficult to recoup a given investment in quality.
The claim that affiliations represent a screen or filter rests on the assumption that it is often easier to observe affiliations than it is to observe differences in quality. Such an assumption seems quite plausible but may not be valid in all cases. For instance, when evaluating job applicants, it is often easier to observe educational affiliations and the status of an applicant’s references than it is to immediately observe differences in individual performance. In the investment banking industry, it is easier to observe the status of a bank’s exchange partners than it is to discern differences in bank quality (cf. Podolny, 1994). And in selecting a daycare facility to care for one’s children, it may be easier to observe a facility’s affiliations with other parents or community groups than to observe the quality of daily care (Baum and Oliver, 1991). A second way that the status of an actor’s affiliates can affect returns to quality is by biasing evaluations of quality. Theorists from a variety of disciplines agree that perceptions are largely influenced by the context in which they are embedded (Lewin, 1935; Cohen, March, and Olsen, 1972; Pfeffer and Salancik, 1978; Granovetter, 1985; Ross and Nisbett, 1991). Social structures and patterns of relation direct attention, dictate the information on which we focus, and shape the meanings, attributions, and emotional responses that such information elicits (Asch, 1940). Empirical work supports these claims. Work in the area of construal and communicator credibility shows that arguments attributed to high-status actors produce greater attitude change than arguments attributed to low-status actors. Messages associated with high-status sources are attended to more closely, recalled more successfully, regarded as more accurate and reliable, and deemed more worthy of adoption than when these same messages are associated with low-status sources (Hovland, Janis, and Kelley, 1953). Extending this logic to the market context, claims of quality (either implicit or explicit) made by firms with high-status affiliations are more likely to be considered credible and trustworthy than similar claims made by firms with low-status affiliations. In essence, status distinctions may engender past and present beliefs about quality, which potential market participants not only use and are slow to revise but for which they are also willing to pay a premium.
In addition to considering how status affects returns to quality, it is also important to consider the extent to which the status-quality relationship may influence subsequent quality choices. That is, if the status of an actor’s affiliations affects returns to an actor’s investment in quality, then the status of an actor’s affiliations may also influence the quality level at which that actor chooses to produce. Notably, if high-status affiliations increase returns to past demonstrations of quality, then low-status affiliations will likely decrease returns and thus reduce the willingness of actors possessing them to produce high-quality products. If market actors with low-status affiliations bid for inputs that are necessary to produce high-quality goods, it is likely these actors will be outbid by their competitors with higher-status affiliations, since this latter group will be able to command higher prices for the goods that follow. As a result, low-status actors may face greater barriers and have less incentive to pay the costs associated with producing high-quality products, whereas high-status actors will have fewer barriers and thus greater incentive to produce higher quality. The above discussion leads to the following hypotheses on the status of a market actor’s affiliates:
Hypothesis 1: The status of an actor’s affiliates will have a positive effect on the actor’s rewards in the market.
Hypothesis 2: The status of an actor’s affiliates will increase returns to past demonstrations of quality.
Hypothesis 3: The higher the status of an actor’s affiliates, the more likely that the actor will subsequently choose to produce high-quality products.
Of course, if status enhances returns to quality, an obvious question arises: What prevents all actors from developing high-status affiliations and, in the process, diluting the status signal? In other words, what ensures that the pattern of status affiliations will remain in a stable equilibrium? For the status ordering to remain in equilibrium, those who have developed high-status affiliations in the past must somehow find it less costly or more rewarding to develop such relations in the future. Otherwise, status distinctions would cease to exist and would ultimately be irrelevant to market decisions and behavior. Two mechanisms may explain how status distinctions remain relatively stable over time even if all actors prefer to affiliate with high-status others. First, high-status firms have a vested interest in avoiding relations with low-status, low-quality producers because such relations threaten their own status. Firms with well-established high-status relations are likely to have more affiliative options from which to choose and thus are more likely to associate with others of similar status to preserve their relative advantage. Thus, one reason that low-status firms may find it more difficult or costly to affiliate with high-status firms is that high-status firms will actively avoid low-status affiliations. A second reason hinges on the previous assertion that the behavior of low-status firms is less likely to be noted or observed than the behavior of high-status firms. If there is a cost associated with affiliation, and if the actions of low-status firms are less likely to be noted, then low-status firms will obtain lower returns and be less likely to invest. Thus, if a status ordering is to exist as a tangible structure, a fourth hypothesis must also apply:
Hypothesis 4: The higher (or lower) an actor’s status, the greater (or less) the net benefit the actor will derive from subsequent high-status affiliation.
Stated another way, returns to high-status affiliation will be lower for actors that have not established high-status affiliations in the past. We tested our hypotheses by examining the affiliation patterns of California wine producers between 1981 and 1990, inclusive. We focused on the wine industry because it is an industry in which an elaborate status system has emerged and where expert ratings of quality are plentiful.
The California Wine Industry
During the period of our study, California winemakers owned approximately 88 percent of all wine acreage (Moulton, 1984) and accounted for 90 percent of all U.S. wine production (Manfreda and Mendelson, 1988). In 1990 alone, California wineries produced more than 320 million gallons of taxable wine (Gavin-Jobson, Jobson’s Wine Marketing Handbook [New York, 1991]). Because California wineries are central to the U.S. wine industry, they are an appropriate focus for the study of status and its effects on quality.
Few industries have as many publicly available evaluations of product quality as the wine industry. In fact, an entire profession has arisen to provide consumers with expert ratings of wine quality across a wide variety of products. Of course, the abundance of such ratings does not ensure their validity. If such ratings are truly to denote quality, two conditions must hold. First, there must be some convergent validity across independent evaluations. That is, wines considered high quality according to one set of evaluators should generally be considered high quality by other evaluators. The lack of such convergent validity implies the absence of a clear, linear metric for the assessment of quality. Second, convergent validity must derive strictly from the aesthetic properties of the wine itself and not from external cues denoting the regional origin, producer, or price of the wine. If convergent ratings were contingent on such external cues, then it would not be possible to assert that the perceived quality differences are in fact real differences inherent in the properties of the wine.
The evaluations of wine do in fact meet these two criteria. As with the rating of most products – especially aesthetic ones – there are sometimes discrepancies across evaluations. In general, though, discrepancies in wine ratings appear to be no more frequent or severe than discrepancies in evaluating product quality in any industry. As we demonstrate below, wine experts agree in their evaluations of wine much more often than they differ. As one well-known wine writer noted, “To the general public, ‘wine tasting’ has always been tinged with romance and obscurity. But serious professional wine judging is, in fact, a highly systematized procedure” (Roy Brady, as quoted in Thompson, 1984: 469).
The systemization of wine rating is due, in large part, to a variety of scientific methodologies, strict classifications, and traditional terminologies that wine experts have developed for the purpose of evaluation. “People who make wines, or blend wines, or sell wines need to make command decisions to determine differences among wines. They use … an array of sensory tests to reach conclusions: score cards, ranking, paired tests, and so on. And, if they are wise, they analyze the results by appropriate statistical procedures for their significance” (Amerine, 1984: 452). The existence of such rigorous standards of evaluation helps to ensure convergent validity. Moreover, because professional evaluations are typically performed under controlled conditions by panels of experts who are blind with respect to a wine’s identity and price, we can be relatively certain that the convergent validity we observe is not due to external earmarks such as a wine’s origin or price. Though there may be slight differences in opinion among individual rankings because of stylistic preferences, experts tend to demonstrate remarkable reliability overall, invariably agreeing on wines that are average, poor, or exceptional (Thompson, 1984).
The wine industry also provides a unique opportunity to measure the status of a firm’s affiliations. Wineries affiliate with formally recognized regions, or appellations, in their efforts to increase status and enhance their identities in the marketplace. Established in 1978 by the Bureau of Alcohol, Tobacco and Firearms (ATF), the American appellation system is a formally recognized governance system that dictates the viticultural designations that can be placed on a wine’s label. The intention behind the ATF system is to enhance the perceived value of American wine in the international marketplace by establishing the credibility of the appellation designations placed on its labels. According to ATF rules, a winery may place a “politically designated” region (i.e., a nation, state, or county) on its label if not less than 75 percent of the grapes that go into the wine originate from the designated area. The appellation system also governs the establishment and use of “viticultural areas.” A viticultural area (another appellation category) is defined as a delimited grape-growing region, distinguished by geographic features that set it apart from surrounding areas. Once an area is legally established, a producer may place an appellation designation on its label if not less than 85 percent of the grape that goes into the wine comes from the specified region. By 1990, 128 viticultural regions had been established throughout the United States, 62 of which were located in California (Figiel, 1991; U.S. Department of the Treasury, Code of Federal Regulations, “Laws and regulations under the Federal Alcohol Administration Act.” Title 27 [Washington, DC, various years]). Well-known examples of viticultural regions include Napa Valley, Sonoma Valley, and Alexander Valley. Together with the 58 politically designated state and county regions, California wine producers could choose among 120 different appellations during the period of our study.
As wineries publicly affiliate with various regions, a clear status or deference ordering emerges. Regions become subordinated to other regions as wineries within them engage in cross-regional affiliation. When a winery in one region – for example, Sonoma Valley – decides to designate another region on its label – for example, Napa Valley – the decision constitutes a clear act of deference, much like the decision of one firm to hire another’s employee or the decision of one graduate department to hire another’s student.
Contributing to the deferential significance of the act is the fact that wineries can exercise considerable discretion in choosing appellations. Wineries are not required to designate where their grapes originate, nor are they required to place an appellation designation on their label, though when they do designate a region on their label they must actually use grapes from that region in their product. Only if wineries choose to produce a varietal wine are regional designations mandated, but even then, wineries have considerable latitude in the designations they select, since appellations often overlap, making it common for a given product to be eligible for multiple designations at once. Nor do wineries have to be physically located within an appellation to place the appellation affiliation on the bottle. In fact, the practice of cross-affiliation is quite common, as wineries are required only to use the mandated percentage of grapes in their product to claim affiliation. Based on data discussed below, almost 50 percent of the bottles that use the Napa Valley designation originate from wineries located outside the Napa Valley region.
It is widely recognized that the motive force behind most viticultural affiliations lies in the perceptual impact that such affiliations have on consumers’ assessments of quality and the economic impact that such perceptions have on the winery’s ability to price its wine. According to the Wine Institute, the wine industry’s primary industry association, “viticultural area designation confers upon growers and wineries within such area a cachet of quality that may well raise the price of their grapes and wine” (Figiel, 1991: 32). By affiliating with a particular appellation, or deferring to it, wineries signal that their wine is of a quality consistent with the wine of other producers who also affiliate with that same appellation.
A particularly visible attempt at such an association is Gallo Vineyard’s efforts to associate with the Sonoma County appellation. Gallo is the largest producer of wines in the United States; in 1997, approximately one-fourth of the wine consumed in the United States was produced by the Gallo winery. At the same time, Gallo has traditionally been regarded as a maker of low-quality “jug” wines and, perhaps more notoriously, the Thunderbird brand. Yet throughout the 1990s, Gallo sought to improve perceptions of its wine, and in late 1997, Gallo began a large-scale advertising campaign in which it identified its new generation of products as “Gallo of Sonoma.”
When a winery places an appellation on its bottle to signify the geographical origins of its wine, the exchange relation through which the winery acquired its grapes becomes visible to consumers and takes on a dual significance. First, the exchange relation affects consumers’ perceptions of the winery. Over time, through multiple acts of affiliation, a winery acquires a distinct status or identity. Second, the exchange relation influences consumers’ perceptions of the overall status ordering of appellations. The more that wineries seek to visibly affiliate with a particular region, thereby showing a certain deference to the superiority of the region, the higher the region’s status becomes in the eyes of consumers.
The appellation system creates a framework by which wineries in one region defer to wineries in another. Together, these deference relations form a network of strategic association that creates meaning by directing constituents’ attention and shaping subsequent attributions (Granovetter and Swedberg, 1992). The deference ordering thus yields a context within which wineries may establish higher or lower status affiliations. A winery that continually affiliates with high-status regions, such as Napa Valley or Sonoma Valley, will have higher status than a winery that persistently affiliates with low-status regions, such as Ybarra County.
We expect appellations to vary by quality as well as status. An advantage of studying the influence of quality and status on rewards in the context of the wine industry is that the wine industry provides data on both quality ratings and affiliations over time, thereby allowing us to partial out the effects of estimated quality and to assess the independent effect of status. Thus, we are able to estimate more precisely the extent to which a winery’s history of affiliations produces market effects that are distinct from the effects of its past and present quality, as measured by experts who are blind with respect to the wine’s identity and origins.
In terms of our hypotheses, we expect a winery’s affiliations to have the following effects. First, controlling for past demonstrations of quality and other factors, wineries that have maintained high-status affiliations over time will be able to command higher prices for a given quality product. Second, returns to past demonstrations of quality will be greater for wineries that have affiliated with high-status appellations than for wineries that have affiliated with lower-status appellations, holding all else constant. Third, controlling for past demonstrations of quality and other economic factors, the status of a winery’s appellation affiliations will have a positive effect on the quality of the wines it subsequently produces: wineries with high-status affiliations will be more likely to produce high-quality wines, whereas those with low-status affiliations will be more likely to produce low-quality wines. Fourth, wineries will derive greater return to subsequent high-status affiliations (higher prices) to the extent that their past affiliations have also been of high status.
We examined the appellation affiliations and product quality of a sample of all California producers of red and white table wine over a 10-year period. The sample consists of all wineries whose products were listed in the Connoisseur’s Guide to California Wine between 1980 and 1991, inclusive. The Connoisseur’s Guide, published by C. Olkin & Company (Alameda, CA), is considered by many industry experts to provide the most comprehensive coverage of California wines available. It seeks to review a broad array of products generally available to consumers from each market segment and to provide information on retail price, quality, and availability. Moreover, unlike other publications, the guide does not solicit wine from individual wineries but, rather, tries to avoid upwardly biasing its selection by buying only finished products on the market. The guide is virtually the only publication that consistently reports retail price and quality data in conjunction with a wine’s appellation affiliation. Data include 10,079 products made by 595 wineries affiliated with 73 different appellations over the 10-year period.
Quality. We controlled for the quality of the bottle itself as well as past demonstrations of quality. Both measures of quality were derived from information contained in monthly issues of the Connoisseur’s Guide. The Connoisseur’s Guide rates wine on a four-point rating scale under peer group, single-blind conditions.(1) This means that all wines of a similar type (e.g., Chardonnay) are rated together and that raters are blind with respect to the wine’s producer and its price. We defined bottle quality as the rating that is reported in the guide for a given bottle of wine. A vineyard’s past quality was operationalized as the average quality of all bottles of wine reported for the winery over the three prior years, a moving three-year window.
Although interviews with winery owners and vitners suggested that the Connoisseur’s Guide provides quality ratings that are reasonable and well-respected within the industry, we assessed the reliability of our quality measure by comparing the Connoisseur’s ratings with the ratings of another wine publication, the Wine Spectator. The Wine Spectator’s Ultimate Guide to Buying Wine (New York, 1992) provides a compilation of the tasting results published in the Wine Spectator since 1980. The Spectator relies on two types of tasting for its quality ratings: (1) weekly blind tastings performed by the Spectator’s tasting panel and (2) special tastings of a particular type or vintage of wine conducted by the Spectator’s senior editors, frequently on location at the winery. To check the reliability of our quality measure, we randomly selected 198 wines rated by both guides.
Because tasters for the Wine Spectator rate wines on a 100-point scale and the Connoisseur’s Guide uses a four-point scale, the standard assumption of equal variation among the two samples did not hold. Thus, in lieu of a traditional rank correlation analysis, we performed a series of pairwise comparisons to determine the extent to which the two sets of ratings agreed in their relative rankings of the wines within the sample by considering all possible dyadic combinations of the 198 bottles. Since each bottle can be paired with 197 others, there are (198(*) 197)/2 = 19,503 such pairings. For each pair i and j, we coded the guides as disagreeing if one guide gave i a rating that was superior to j and the other guide gave j a rating that was superior to i. There was disagreement in only 21 percent of the cases.(2)
Status. Wineries acquire their status by affiliating with appellations of a given position in the overarching status hierarchy. The status hierarchy itself arises from the pattern of cross-regional affiliations that wineries engage in through the appellation designations they place on their labels. We used these cross-regional citations to determine the status ordering that existed during the 10-year period examined by our study. To assess the relative status of appellation regions in a particular year t, we constructed a relational matrix [R.sub.t], where each cell [r.sub.ijt] represents the number of wineries in geographical region i that use the appellation of geographical region j in year t. [R.sub.t] is a square, asymmetric n x n matrix, where n denotes the number of appellations. The main diagonal is set to 0. Having constructed the [R.sub.t] matrix, we then applied Bonacich’s (1987) c([Alpha], [Beta]) measure, a standard measure for relational data on status, to determine each appellation’s status relative to that of other appellations. The measure is formally defined as follows:
c([Alpha], [Beta]) = [Alpha] [summation of] [[Beta].sup.k] [R.sup.k+1] 1 where k = 1 to [infinity]
where c([Alpha], [Beta]) is a vector of status scores for the appellation regions, [Alpha] is an arbitrary scaling factor, [Beta] is a weight, and 1 denotes a column-vector of ones.
We set [Alpha] so that the highest status region in year t has a status score of 1. If [Beta] equals 0, an appellation’s status is strictly a function of the number of times that the appellation is used by wineries that are not physically located in that geographical region. If [Beta] [greater than] 0, an appellation’s status is a function of the number of times that the appellation is used by wineries that are not physically located in the geographical regions and of the status of the regions in which these outside wineries are located. The higher the status of the regions in which these outside wineries are located, the more that they enhance the status of the region whose appellation they are using. If [Beta] [greater than] 0, the conferral of status across regions is therefore transitive. As [Beta] increases, so does the transitive conferral of status, such that a region’s status is affected by the status of the regions that are subordinated to it through the cross-regional acts of deference. The upper bound on [Beta] is the reciprocal of the largest eigenvalue of the R matrix. For the analyses that follow, we set [Beta] at this upper bound. We considered a range of values for [Beta], and these alternatives produced no substantial differences in the status scores.
Calculation of status scores for appellations in year t requires complete information on all winery affiliations made during that year, in addition to the wineries’ geographic locations. While it is not possible to obtain full information on all wines produced within a year, it is possible to obtain complete information on all affiliations. The ATF requires all producers to obtain a Certificate of Label Approval if they intend to ship their products interstate or a Certificate of Exemption from Label Approval if they do not (U.S. Department of the Treasury, Code of Federal Regulations, “Laws and regulations under the Federal Alcohol Administration Act.” Title 27 [Washington, DC, various years]). Each year, wine producers register thousands of labels with ATF. ATF inspects all labels, determines whether they meet required guidelines, and then retains a final copy for its archives. Given the enormous amount of work involved in collecting even one year of data, we did not collect the labels for each year. Rather, we gathered data on the appellation affiliations of all California red and white table wines produced in each of four years staggered evenly throughout the 10-year sample period: 1981, 1984, 1987, and 1990. California wineries registered 22,027 labels across the four different years. Because many of these labels reflected different labeling designs for the same bottle of wine, however, we eliminated those labels reporting the same producer, varietal, vintage year, and appellation in a given year, reducing the total to 7,641 unique alignments across 73 designated appellations.
We determined the geographic locations for the various wineries using viticultural maps: Raven Maps and Images’ 1989 Vineyards and Wineries of California and Compass Maps’ 1992 Winery Guide to Northern and Central California (Modesto, CA). We verified locations by contacting by telephone all wineries still in existence, which allowed us to determine whether wineries had changed location during the period of study and assisted us in pinpointing exactly when and where such moves took place. Locations for wineries no longer in existence were verified through historical record. Together, the geographic location information and the ATF affiliation data enabled us to construct the [R.sub.t] matrices used in determining the status scores.
Table 1 presents representative status scores for a number of appellations in 1990. To validate the status scores derived from the cross-regional affiliation patterns, we surveyed seven industry experts in 1993 for their perceptions of appellation status at that time. We located these experts through officials at the Wine Institute, the wine industry’s major trade organization, which provided us with access to its Consumer Media Guide. Usually available only to the institute’s members, the Consumer Media Guide lists over 375 writers, editors, and reporters who regularly cover wine and the wine industry for newspapers, magazines, food and wine programs, and other consumer audience media throughout the United States. We selected approximately fifteen experts from the Media Guide who were well known for their writing on the industry, or associated with a well-known publication, and were familiar with both northern and southern California wine regions. Ten of the experts we contacted agreed to participate, and seven of these completed surveys. These experts consisted of wine writers, critics, and legal representatives active in establishing the appellation system and many of the viticultural areas within it.
Representative Status Scores 1990
Alexander Valley 1.0000
Sonoma County 0.7656
Russian River Valley 0.7052
Napa Valley 0.4491
Los Carneros 0.2637
Sonoma Valley 0.2489
Mendocino County 0.2163
Dry Creek 0.1927
North Coast 0.1461
Chalk Hill 0.0559
Sonoma Mountain 0.0519
Napa County 0.0413
Central Coast 0.0397
Howell Mountain 0.0391
Anderson Valley 0.0376
Sonoma Coast 0.0280
Stags Leap 0.0264
Sonoma County Green Valley 0.0254
Paso Robles 0.0224
Cienega Valley 0.0217
Knights Valley 0.0212
Monterey County 0.0210
San Luis Obispo County 0.0182
Northern Sonoma 0.0178
Santa Maria Valley 0.0170
Santa Barbara County 0.0167
Amador County 0.0157
Santa Clara County 0.0146
Edna Valley 0.0138
South Coast 0.0124
Santa Clara Valley 0.0116
Solano County Green Valley 0.0080
Lake County 0.0068
McDowell County 0.0067
Santa Ynez Valley 0.0059
Ventura County 0.0000
Mount Veeder 0.0000
Trinity County 0.0000
We first performed a principal component factor analysis on the ratings given by the experts to determine the aggregate level of agreement. The factor analysis yielded a single factor solution, with all experts loading between .72 and .90, which indicates a strong consensus among the raters. Having established interrater reliability, we calculated the mean expert rating for each appellation and then ranked ordered the mean ratings. The Spearman correlation between these ranks and the status order obtained from the relational matrix calculations on the 1990 ATF data was .81. These results suggest that perceptions of well-informed participants correlate strongly with the patterns of cross-regional affiliation we observe in our data.
Despite the high correlation, there are some features of table 1 that may surprise some individuals familiar with the California wine industry. First, there is a considerable break in status between the three highest-status regions and the rest. Second, there are some orderings that may seem non-intuitive. Napa Valley’s fourth-place ranking, for example, surprises some wine followers, who expect it to be higher. Others familiar with the industry, however, argue that the region’s lower-than-expected ranking can be attributed to the potentially diluting effect of numerous other appellations that have been established within its boundaries. As winery owners and grape growers have established smaller, more specialized subregions within the larger Napa Valley, many of these regions have seen their status grow, while Napa Valley’s status has diminished, a claim that is supported by our data.
Even with such explanations, it is impossible to refute all objections to any status ordering, since perceptions of status will never be completely uniform. Thus, we considered several factors in evaluating the strength and validity of our status scores. First, we believe the .81 correlation between the status ranking and the ranking of expert ratings provides strong support for the validity of our status measure. Even with some nonintuitive rankings, the experts surveyed show considerable agreement with the 1990 status ordering. Second, to the extent that status is improperly measured by our network measure, the error in measurement should only reduce, not enhance our ability to test and support our hypotheses. Thus, if our hypotheses do achieve significance, it would suggest that our status measure is acceptable. Finally, because there is no non-arbitrary way to adjust the status ordering to be perfectly consistent with all perceptual estimates, we performed one additional check in evaluating the robustness of our scores. We repeated our analysis of the effects of status on price using the mean ratings of the experts rather than the network scores. Since the experts provided ratings in 1993 only – while the network scores were compiled for 1981, 1984, 1987, and 1990 – we do not consider this analysis to be a rigorous test of our hypothesis, nor a definitive validation of our status measure; rather, to the extent that the results of this additional analysis remain largely consistent with our original one, the analysis provides another confirmatory check of the robustness of our measure and main results.
Using the status scores derived from the cross-regional affiliations, it is fairly straightforward to specify the status of a given bottle’s affiliation, as well as the average status of affiliations that a winery has maintained over time. A bottle’s status in year t simply derives from the most recently measured status score of the appellation with which the bottle is affiliated. For example, a bottle produced in 1982 is given the status of the bottle’s appellation in 1981, since 1981 is the most recent year for which we have information on the appellation’s status. Wineries produce multiple products each year. They may choose to use a mix of appellation affiliations across products or, alternatively, to affiliate with a single appellation for all products. As with the measure for past demonstrations of quality, we used a three-year moving window to summarize the average status of a winery’s appellation alignments, averaging the status of all appellation affiliations across all bottles produced by a winery over the three prior years. For example, if a winery produced a bottle in 1982 from an appellation with a status score of .023, another bottle in 1982 with a status score of .05, a bottle in 1983 with a status score of .10, and a bottle in 1984 with a status score of .08, its status in 1985 is (.023 + .05 + .10 + .08)/4 = .0633. According to our hypotheses, the status reflected by this three-year window will have an independent effect on price, will positively interact with the three-year quality window in affecting price, will positively interact with subsequent affiliations in affecting price, and will have a positive effect on the subsequent quality of wine produced by the vineyard.
Effects of Status on Price
If [P.sub.ijt] denotes the price at which winery j sells bottle i at time t, we model the determinants of price using the following equation:
[Mathematical Expression Omitted].
Here [q.sub.ijt] refers to the quality of bottle i sold by winery j at time t, [S.sub.ijt] refers to the status of the appellation listed on bottle i, [Mathematical Expression Omitted] refers to the average quality of winery j as revealed by its wine between t and t – k (excluding the quality of bottle i), [Mathematical Expression Omitted] refers to the average status of winery j’s appellation affiliations between t and t – k (excluding the appellation affiliation on bottle i), and x refers to a vector of control variables reflecting economic forces in the environment (e.g., annual demand for wine) and indicators of fixed and variable production costs (e.g., grape prices, vineyard acreage), and [U.sub.ijt] denotes an error term. This specification implies that the price that a winery commands for a particular bottle of wine is a function of factors at three levels: the level of the bottle, the winery, and the market environment. Hypotheses 1, 2, and 4 respectively imply that [b.sub.14] [greater than] 0, [b.sub.15] [greater than] 0, [b.sub.16] [greater than] 0.
Modeling the effects of past affiliation on a producer’s subsequent choice of quality is somewhat more complicated, because the quality of wine produced is ultimately a function of multiple choices that can be grouped into two categories: (1) choices about the cost or quality of raw materials used and (2) choices about the production process or the conversion of raw materials into the final product. Since the status of a winery’s affiliations can influence both access to and the ability to afford higher-cost raw materials, as well as decisions to engage in higher-quality production techniques and/or use higher-quality talent, we assessed the effects of past affiliations on quality choice in two stages.
First, we assessed the effect of past affiliations on the quality of the primary grape used in production, the raw material. We modeled this as:
[Mathematical Expression Omitted],
where c([g.sub.ijt]) is the cost of the grape used for bottle i produced by winery j in year t. Hypothesis 3 implies that [b.sub.22] [greater than] 0. The higher the status of a winery’s past affiliations, the more likely the winery will be to use more expensive (presumably higher-quality) grapes.
We then assessed whether past affiliations have an effect on wine quality, controlling for the raw materials used in the wine by modeling the perceived quality of a particular bottle of wine as a function of the past affiliations of the winery producing the wine. We expected that the effect of affiliations on final quality would be positive. Specifically, if
[Mathematical Expression Omitted],
we expect [b.sub.32] [greater than] 0. We then added two additional explanatory variables on the cost and status of the raw materials to this equation: the cost of the grapes used in the wine and the status of the appellation from which the grapes derived and with which the winery chose to affiliate. If the effect of past affiliations remains significant, then we can conclude that past affiliations affect not only the quality of inputs but also the quality of the process of converting those inputs to outputs. If the effect of affiliations becomes insignificant when these two variables are included, however, we would infer that the primary mechanism by which past affiliations affect quality is through the choice of higher- or lower-cost (and presumably quality) raw materials.
Control variables. Other factors besides quality may affect the price that a winery charges for its products. First, we expect demand to vary by varietal or the type of grape from which the wine is made. Chardonnay, Cabernet, and Merlot are all examples of varietals. There are 17 varietals or blends of varietals in our data. Some grapes, like Chardonnay and Cabernet, are in extremely high demand; others, like Syrah, are in much lower demand. To allow for these differences in demand across varietals, differences that may affect price, we included dummy variables for each varietal. The varietal information was obtained from the Connoisseur’s Guide. Second, we expected that demand would vary temporally, increasing or decreasing over time. To allow for such fluctuations, we included dummy variables denoting the year in which the bottle of wine was sold.(3)
Some industry experts assert that wine producers use a markup or cost-plus pricing strategy to set wine prices (e.g., Kramer, 1992) and apply a standard markup to the costs of producing the wine. Such explanations oversimplify the pricing decision, since most wineries consider many factors in setting their pricing policies and use a variety of strategies depending on their pricing objectives, but it would be difficult to deny that a producer’s costs are extremely relevant to price. As Stuller and Martin (1989: 39) noted, “Winery economics, as much as market considerations, are often the major factor in the price of a given wine. The quality of the raw materials and enological skill that go into a wine are critical if it is to successfully remain in a price segment. But where it’s placed in a segment can often depend on when a winery was built, how it’s financed, equipped, and operated.” Thus, it is important to include indicators of a firm’s production costs in our examination of price formation.
Any focus on costs must begin with a consideration of the costs of the grape. Grapes typically represent a winery’s single largest production cost. For some super-premium vintages, grapes may account for up to 60 percent of a winery’s operating costs, though for most small wineries, grapes average about 40 percent of operating expenditures (Stuller and Martin, 1989). We would expect wineries to charge higher prices for wines produced from more expensive grape varietals and for wines produced in years when the cost of grapes was relatively higher. Grape prices were obtained from the Final Grape Crush Report, 1981-1991, a publication produced by the California Department of Food and Agriculture (Sacramento, CA, various years) in compliance with Section 55601.5 of the Food and Agriculture Code. We included in our analysis the average price per ton of the primary grape used in a given wine, as indicated by the varietal, appellation, and year listed on its label, for all years and all varietals for which data were available. Drawing on these data, we were able to provide cost information for approximately one-third of the bottles in the sample. For the rest of the observations, we set cost equal to 0 and constructed an indicator variable coded 1 if information was missing on cost and 0 otherwise. Use of such an indicator variable allows for unbiased estimates of the effect of cost and does not require excluding numerous cases with missing cost data.
Economic models assert that a firm’s costs are not only a function of its primary inputs but also a function of economies that the firm is able to realize. The economies most relevant to the wine industry are learning economies and economies of scale. Learning economies refer to the reduced production costs that accrue from learning about the production process over time. We used the log of the winery’s age as a proxy for these learning economies. Newer wineries will have higher costs than older wineries if only because they do not have as good an understanding of the production and distribution of wine. Because the log of age should have a negative effect on costs, the log of age should also have a negative effect on price.
Economies of scale refer to the reduction in per-unit cost that accompanies increased production volume. Moulton (1984) found that production costs per case of wine generally go down as winery size expands. His estimates suggest that costs fall by approximately 27 percent as production increases from 5,000 to 100,000 cases and by an additional 35 percent as production increases to 3 million cases. We included a number of proxies for scale economies in our analysis: the number of acres owned by the winery, the storage capacity of the winery, and the number of brands produced by the winery. The number of acres owned by the winery and its storage capacity are both direct indicators of the winery’s size and volume. To the extent that an increase in scale leads to a decrease in cost, these measures should have a negative effect on price. The number of brands used by the winery is also an indicator of size and potential scale economies, but brands may have something to do with a winery’s quality and pricing strategy as well. Partitioning products of varying quality across numerous brands can help a winery protect the high-quality, high-status image of its superior products from the lower status of lesser-quality products. Of course, to the extent that all products are of sufficiently high quality, a winery will have little incentive to create multiple brands – it will simply use a single high-status, high-quality brand and command a premium price. To the extent that quality varies across products, however, a winery will have an incentive to partition its products across multiple brands. Thus, consistent with the predictions suggested by the economies of scale argument, a winery with more brands should command lower prices for its products, on average, than a winery with fewer brands.
Analysis. Since our observations on the price of individual bottles are pooled across a number of vineyards, errors are likely to be correlated across observations. To correct for the potential effects of nonrandom error variance in the analyses predicting price, we estimated the proposed relationships among status, quality, and price using a random effects generalized least squares model (Johnston, 1984). The model interprets each regression intercept realization (for each winery at a given point in time) as a random draw from a population with a fixed mean and randomly distributed error term that varies by organization. This estimation technique allows an unbiased assessment of parameter estimates by factoring out nonrandom error variance across vineyards.
To examine the effects of status on investment in quality, we used a standard logistic regression model in which the dependent variable is the probability that a given quality is chosen for a particular bottle of wine. If our third hypothesis is correct – that higher status actors will be more likely to invest in and produce goods of a given quality – the coefficient for our measure of organizational status should be positive.
Status and Returns to Quality
Table 2 presents the means, standard deviations, and correlation matrix for all variables of interest. While the correlations between the variables denoting quality and status are all positive and statistically significant, the strongest correlation between the two most robust variables – three-year status and three-year quality – is only about .6. The absence of a stronger correlation among the status and quality variables reinforces the basic claim that status and quality are not interchangeable.
Table 3 depicts the effects of status and quality on price. Model 1 is a baseline model including only the effects of quality and the appellation listed on the bottle. Results show that the quality of the bottle has a significant positive effect [TABULAR DATA FOR TABLE 2 OMITTED] [TABULAR DATA FOR TABLE 3 OMITTED] on price. Since price is expressed in terms of dollars, interpretation of the effects is rather straightforward: a one-unit increase in bottle quality produces a $1.27 increase in the price of a bottle of wine. As a point of comparison, the average bottle of wine in our sample costs $10.65.
Model 2 adds control variables for the cost of the grapes, learning economies, and economies of scale. As anticipated, the cost of the grape has a significant, positive effect on the price of the wine. Our other expectations for the control variables receive more mixed support. We used three separate variables to control for economies of scale: storage capacity of the winery, vineyard acreage owned, and the number of brands owned by the winery. Although the coefficients associated with all three variables are negative and thus consistent with the existence of scale economies, vineyard acreage is not statistically significant. Because wineries can purchase grapes in addition to or in lieu of growing their own, acreage may not always be directly indicative of a winery’s actual economies. Finally, while a negative coefficient for log-age is consistent with the notion of learning economies, the effect is also not significant.
Model 3 reports the main effects for winery status and winery quality. The quality of the winery, as measured by the average quality of all wine it has produced over the three prior years, has a positive and significant effect on price – one that is distinct from the effect of bottle quality. In our view, it is reasonable to think of this variable as a reputation effect, since it represents the winery’s past quality performance and is thus consistent with the economic definition of reputation. Consistent with hypothesis 1, the main effect for winery status is positive. Although this effect is not significantly different from 0 in model 4, the inclusion of the interaction variables in model 4 means that one cannot consider the main effect independent of the level of the variables with which vineyard status is interacted. Hence, the results of model 3 provide a more straightforward test of hypothesis 1, since this model essentially reports the average effect of winery status across all the observations. Moreover, even if one uses the results in model 4, one can evaluate the effect of vineyard status at the mean value of vineyard quality and the mean value of bottle status. The effect is as follows: -.195 + (.898(*).61) + (.862(*).55) = .827, which is very close to the average value reported for model 3. Thus, consistent with prior research examining the impact of status on organizational outcomes, the results support hypothesis 1. As predicted, organizational status has a main, positive effect on the prices an organization commands for its products.
Results of model 4 also reveal positive and significant effects for the interaction of winery status with winery quality and for the interaction of winery status with bottle status.(4) Though not reported in the table, each interaction was positive and significant when entered into the analysis individually as well. The positive interaction between winery status and winery quality provides support for hypothesis 2, that high-status affiliations increase returns to past demonstrations of quality. The positive and significant interaction between winery status and the status of the appellation on the bottle supports hypothesis 4. High-status wineries appear to derive greater returns from a given high-status affiliation than do lower status wineries. Wineries with a history of the highest status affiliations (i.e., vineyard status = 1) can charge 86.2 cents more per bottle for a given affiliation than wineries with a history of the lowest status affiliations (i.e., vineyard status = 0), holding all else constant.(5)
There is no significant main effect for the status of the bottle’s appellation, however, when the interaction term between winery status and bottle status is excluded from the analysis and bottle status is entered alone. Affiliating with a high-status appellation for the production of a single wine produces no positive benefit if a winery has not invested in high-status relations in the past. This result supports our basic contention that the value of an affiliation cannot be assessed in isolation but can be understood only in the context of other affiliations formed by that winery.
If one takes the point estimates literally, then only the highest-status wineries can command higher prices for higher-status appellation affiliations. Specifically, the effect of the status of the appellation listed on the bottle is as follows: -.770 + .862 * Winery Status. This point estimate suggests that only those wineries with status greater than .770/.862 = .89 should benefit from higher-status affiliations. Given that there is a broad confidence interval around this interaction, however, it would be a mistake to interpret the point estimate too literally. For example, if one uses estimates for the main and interaction effects that are only one standard deviation above the point estimate, then the joint effect is (-.770 + .287) + (.862 + .413)(*)Winery Status or .483 + 1.275(*)Winery Status. In this case, all vineyards with an average status above .483/1.275 = .378 would derive a positive benefit from higher-status affiliations. In either case, the results remain sufficient to reject the null hypothesis that high-status and low-status wineries receive the same net benefit from affiliating with appellations of a given status. In rejecting this null hypothesis, the results provide support for hypothesis 4. High-status vineyards are able to command higher prices for an affiliation of a given status, thereby enabling high-status wineries to outbid their low-status competitors for grapes from high-status regions.
Similar results obtain when we rely on the expert assessments of status. Model 5 portrays the full model using the expert assessments as the basis for the status measure. The measure of bottle status based on expert assessments has a correlation of .73 with the measure of bottle status derived from the network measure and a correlation of .82 with the measure of three-year winery status based on the network measure. In interpreting the results of model 5, it is important to bear in mind two features of the analysis. First, the scale for expert assessment, which can range between 1 and 7, is different from the scale for the network measure of status, which can range between 0 and 1. Bottle status based on expert assessment has a mean of 4.88 and a standard deviation of 2.01. Winery status based on expert assessment has a mean of 4.78 and a standard deviation of 1.96. Second, though there are some differences in the main effects, the main effects cannot be interpreted independent of the interaction terms. When considered with the interaction terms, the results using the expert assessments are largely consistent with the results based on the network measure. The interaction between three-year status and three-year quality is positive and statistically significant. In the results reported for model 5, the interaction between three-year status and bottle status is positive and just beyond conventional significance levels (p = .11). If nonsignificant control variables are removed from the model, however, the interaction between three-year status and bottle status becomes significant at the .10 level, indicating that the higher-status wineries derive a greater return from a high-status affiliation. Given that the status ordering experienced some changes over the 10-year period and that the expert ratings were administered at the conclusion of that period, we believe these results can be interpreted as consistent with our network-based status results: namely, a history of high-status affiliations provides a positional advantage for high-status wineries, which, at an aggregate level, facilitates the reproduction of the overall status ordering.(6)
Effect of Status on Quality Choice
Hypothesis 3 posited that if high-status relations increase returns to a given quality of product, actors with higher-status affiliations would subsequently be more likely to make higher-quality products than would actors with lower-status affiliations. To test this hypothesis, we conducted two analyses. First, we examined the effect of past status on the cost of the grape used in the wine in an analysis including the same [TABULAR DATA FOR TABLE 4 OMITTED] control variables as in the analysis of bottle price, excluding the cost of the grape. Table 4 reports the regression results. Again, the unit of analysis is the bottle. The number of observations is smaller because data on grape cost are available for only approximately one-third of the cases. Results of model 1 show that three-year status has a positive effect on the cost of grape used in the product. Superior returns afforded by status appear to lead higher-status producers to purchase higher-quality grapes as inputs. Conversely, because lower-status producers cannot obtain as great a return, they do not purchase more expensive, higher-quality grapes.
Because there was a positive interaction effect of three-year status and three-year quality on bottle price, we examined whether such an interaction also influences subsequent quality through the purchase of higher-quality grapes. Since higher-status affiliations increase the returns to past demonstrations of quality, these affiliations should also increase the extent to which that quality is maintained and enhanced through the purchase of superior inputs. The results of model 2 show that the effect is positive but not significant (p = .20). Hence, while the direction of effects is as anticipated, we cannot rule out the null hypothesis that there is no joint effect of quality and status on grape price.
We next examined the extent to which past affiliations influence the quality of the final product. As in the other analyses, quality was defined as the California Connoisseur’s Guide’s rating for a particular bottle of wine. Because the Guide rates the quality of each bottle on a four-point rating scale, our dependent measure, quality, is categorical and ordinal. Thus, we used an ordered logistic regression model to assess the effect of past status on subsequent quality. The ordered logistic regression is simply a generalization of the dichotomous logit, in which there are N-1 intercepts for N categories. The advantage of the ordered logit over an ordinary least squares model is that it does not require the assumption of constant intervals between categories and cannot [TABULAR DATA FOR TABLE 5 OMITTED] give rise to predicted values that are outside the range of possible values for the dependent variable.
Table 5 reports the results. The analysis includes controls for past quality, grape price, relevant vineyard characteristics, and year and varietal effects. Model 1 shows the effect of past affiliations without the inclusion of the price of the grape and bottle status as controls. Model 2 includes the cost of the grape and the status of the bottle as explanatory variables. The effect of three-year status is positive and significant in model 1 but loses significance in model 2. Our interpretation of this result is that the effect of past affiliations on subsequent quality typically plays out through the selection of the quality of the raw material input. Because lower-status vineyards cannot obtain the same returns from a high-quality grape, they do not purchase expensive, high-status grapes. More generally, these results suggest that, on average, status of affiliation affects subsequent selection of inputs but not necessarily the quality of the process of converting inputs to outputs.
After obtaining this result, we wondered whether it applied at all quality levels of input. based on our knowledge of the industry, we suspected that wineries would devote more resources and effort to converting higher-quality grapes to high-quality products than they would to converting lower-quality grapes to high-quality products. Superior returns from past affiliations might enable owners of high-status wineries to hire superior winemakers or to buy superior equipment to realize the potential of their high-quality inputs, while low-status wineries might not be able to do so. To test these assertions, we examined the interaction between three-year status and grape cost. Model 3 shows the result. As expected, the interaction is positive, indicating that even when the quality of the grape and status of the bottle affiliation are included in the model, three-year status has a positive and significant effect on final quality when the raw material is a high-price and implicitly high-quality grape. Taken together, the results in tables 4 and 5 provide evidence that an actor’s position may significantly influence and constrain important actor attributes, namely, the quality of its inputs, production, and final products.
Sociologists have frequently argued that economic models of reputation portray an undersocialized conception of firm identity (e.g., Granovetter, 1985), but such critiques have offered little guidance on how reputation considerations might be incorporated into a richer vision of the social structure of the market. In this paper, we have tried to lay a foundation for integrating the fundamental dynamic emphasized by economic models of reputation with a sociological conception of status.
Reputation models in economics draw attention to how investments in quality at one point in time affect market opportunities over some subsequent time. Our central claim has been that the consequences of quality choices are not independent of the tangible status ordering in which market participants are situated. A firm’s position in the status ordering influences the attention that others pay to quality, their assessment of quality, and their regard for the product more generally. Relative to lower-status firms, higher-status firms therefore derive greater benefit from producing a given quality product. As a consequence, the status ordering helps to determine which firms will develop reputations for quality and which will not. In many cases, reputation differences may not be ascribable solely to differences in the underlying capabilities of producers but, rather, may be ascribable to differences in the pattern of affiliations. That is, where a firm is located in the social structure of a market and who the firm affiliates with may strongly influence the perceived quality of the firm within the market.
We do not wish to assert that differences in structural position are completely exogenous, nor that differences in quality have no effect on structural position. There is undoubtedly a reciprocal relationship between the level of quality that a firm achieves and the structural position that a firm obtains. We simply assert that the existence of the status ordering constrains how firms can develop reputations for quality. Specifically, firms receive lower returns to their quality investment to the extent that they fail to establish affiliations that reflect that investment. More important, highly discrepant shifts in quality, affiliation, and status do not appear to achieve returns capable of sustaining a firm’s investment. A firm’s history or pattern of affiliations over time constrains the returns available to subsequent affiliations and quality decisions. More generally, the existence of a status ordering creates disincentives for individual firms to improve quality in ways that do not contribute to the reproduction of the status ordering.
One interesting consequence of the alignment of quality with status pertains to innovative behavior on the part of firms. In the context of our study, the existence of the status ordering means that wineries have little incentive to develop high-quality wines that are not simultaneously high status. Suppose, for example, that a winery could develop a more well-balanced and thus higher-quality wine by blending equal proportions of grapes from two distinct regions, say Mendocino County and Sonoma County.(7) In this case, the winery by law would not be allowed to use either region’s appellation, since neither region contributes 85 percent of the grapes that go into the wine. Recognizing that consumers look to the appellation listed on the bottle and its status relative to other appellations in making inferences about a wine’s quality, wineries have a disincentive to develop high-quality blends that draw from diverse regions. In essence, the status ordering inhibits high quality, low-status innovations, thereby restricting the range of potential innovations that may be considered in a given market.
Obviously, because the wine industry is only one industry and has some unique features, it is important to be cautious in generalizing the results of this study to all industries, although we believe that similar status dynamics exist in other industries. For example, one would likely see similar dynamics in the film industry. Films associated with a high-status director, actor, or writer might derive greater revenue, holding constant the evaluations of quality provided by film critics. Similarly, a systems integrator or software developer might receive greater returns for a given quality of product or service to the extent that the provider is formally affiliated with a high-status producer such as Microsoft or Sun Microsystems. The wine industry, however, has some unique features that make it an especially compelling context for examining the role of status as a market signal. One of its most compelling features is that wine is often considered a conspicuous consumption good (Veblen, 1994), thereby allowing an explicit comparison of the role of status as market signal versus the more traditional understanding of status as a basis for conspicuous consumption. Wine may be purchased from a high-status producer not only because the producer’s status signals high quality but also (or more importantly) because the act of consuming such a wine strengthens the individual’s ability to claim higher social status. So, for example, an individual may choose to buy a bottle of Opus One rather than another lower-price bottle he or she believes to be of equal quality because he or she would like to be perceived as the type of individual who drinks Opus One.
In our view, there is no underlying incompatibility between these two understandings of status. Status may serve both as a signal of quality and as a tool of conspicuous consumption. If an individual’s social standing is enhanced by drinking wine from a high-status winery, then the individual should be willing to pay a higher price for that wine. Moreover, if the individual perceives that one’s social standing is enhanced by drinking wine affiliated with a particular region, or by drinking wine from a winery that has affiliated with that region for a substantial period of time, it is quite likely that the individual will be willing to pay substantially more for a product that offers both the current and past associations. Despite their consistencies, the conspicuous consumption rationale and the status-based model remain distinct in some important and defining ways. The most notable distinction is the role that quality plays in determining price and purchase. While we predict that status serves as a signal of quality and thus increases returns to past investments in quality and incentives to invest in future quality, according to the conspicuous consumption view, quality is relatively insignificant. The act of conspicuous consumption is precisely the demonstration of wealth through the throwing away of money on more expensive goods that provide no greater utility but cost significantly more. Thus, the conspicuous consumption rationale would not explain why a vineyard’s status would vary with the underlying level of the vineyard’s quality. In short, while we do not deny that conspicuous consumption may be an important motive underlying the purchase of wine, such a view alone does not lend itself to the hypotheses we tested in this paper.
In addition to providing an opportunity for explicitly discussing the interpretations of status as a signal and as a basis for conspicuous consumption, there are at least three other reasons why the wine industry represents a compelling case for the broader significance of status processes. First, there is an abundant supply of information about quality in the industry. Prior research has demonstrated that firms attend more to the status of their exchange partners when there is greater uncertainty about qualitative differences among competing producers (Podolny, 1994). This suggests that status should be less relevant in contexts in which quality can be more easily ascertained. In the wine industry, reliable quality estimates are provided by numerous experts through a variety of sources. The plethora of wine columns, newsletters, and books dedicated to rating wines suggests that the major obstacle to obtaining reliable information on the quality of a particular wine is the search cost involved in locating an appropriate information source.
A second reason is that wine is a consumer product that is distributed to a market of mass consumers. The case thus differs substantially from contexts in which status effects have been previously studied. Previous work has focused almost exclusively on producers of specialized services such as investment banks (Podolny, 1994) or accounting firms (Han, 1994) and their negotiated relations with potential buyers. These buyers tend to be, on average, other firms that are themselves linked by broad and expansive networks. In contrast, the wine industry is a context in which buyers are dispersed and, in general, isolated from all but a small number of other buyers. Our analysis indicates that even when the buyer side of the market more closely approximates a mass market structure, status processes are still relevant to its operation.
A third and final reason that this analysis supports the significance of status processes more generally is that the status ordering is reproduced in what is arguably a very price-sensitive industry. In three of the four primary segments of the wine industry – the jug market (less than $3 per 750 mi. bottle, retail), the premium market ($3 to $7 per bottle), and the super-premium market ($7 to $14 per bottle), consumers have been found to react sharply to even slight price increases (Valette, 1992). This price sensitivity suggests that status should have less of an effect on price in the wine industry than in other industries in which consumers are less sensitive to price. Price is also a strong determinant of affiliations on the production side. This might be expected to reduce the stability of the status ordering. In earlier analyses of status, such as Podolny’s (1994) analysis of the investment banking industry, the status ordering is preserved to the extent that high-status actors wish to avoid associating with low-status actors. In the work presented here, grape growers do not consider a winery’s status when deciding to whom they should sell their grapes. The price that a winery offers for the grapes is the primary purchase factor considered. Even if a grape grower from a prestigious region believed that the image of the region might suffer if a low-status winery used the appellation designation, there would still be a tremendous free-rider problem. The benefit that the grape grower would obtain from selling the grapes would likely more than outweigh the cost that might ensue from a potential decline in image that one winery might bring.
Yet, while growers from high-status regions do not appear to actively avoid associating with low-status wineries, the status order continues to be reproduced. The fact that low-status wineries cannot sell a bottle with a given appellation for the same price that high-status wineries can is sufficient to ensure a certain reproduction of position within the status ordering. Because higher-status wineries can command higher prices for their wines, they can afford to outbid their lower-status competitors and thereby preserve their relative position in the status hierarchy. Accordingly, while we wish to remain cautious about making claims for the general applicability of these findings, we also believe that the unique features of this industry help to establish a strong case for the importance and broader generality of the status processes we have described. If the effects we have shown here replicate in other industries, we may have a better understanding of why the status ordering within industries remains relatively stable and how producers may more effectively move within the ordering if they so choose.
The authors thank John Adams, Bart Balocki, Bill Barnett, Roberto Fernandez, Pam Haunschild, Mark Mizruchi, Don Palmer, Dan McAffrey, Christine Oliver, and several anonymous ASQ reviewers for their helpful comments. A substantial part of this research was completed while Beth Benjamin was an organizational behaviorist at the RAND Corporation.
1 Though some sources measure quality using finer gradations than a four-point scale, they lack the comprehensiveness and/or representativeness of the Connoisseur’s Guide.
2 We believe that this 21 percent figure is a conservative estimate and would likely be much lower in a fully random sample. Unlike the Connoisseur’s Guide, which rates a wide spectrum of California products, the Wine Spectator rates products that are submitted to it for review by wineries, or products that are sufficiently well known to be of interest to the well-informed eonophile. As a result, the sample of wine reviewed by the Spectator is likely to be skewed toward higher-end, higher-priced wines that are less representative of the overall market. This skew, in turn, likely leads to an underestimate of the agreement across the two rating guides.
3 Although it might be reasonable to include interactions for fluctuations in demand by varietal and by year, in case demand for some varietals increases while demand for others declines, examinations of year-varietal interactions for the more common varietals revealed no evidence of such interaction effects.
4 Given the inclusion of two interaction terms in model 4, we were concerned about possible multicollinearity. We screened for potential multicollinearity effects using a set of multicollinearity diagnostics proposed by Belsley, Kuh, and Welsch (1980) and available in SAS version 6.1. The approach computes a condition index and coefficient estimates for each effect. The diagnostics produced results that fell well below Belsley, Kuh, and Welsch’s recommended cutoffs, thereby suggesting that multicollinearity concerns were minimal. In addition, we also tested the model using mean deviated interaction terms. Once again, the results were significant and in the predicted direction.
5 Since the R-squared does not show a marked increase with each model, we questioned whether the inclusion of the additional variables was justified. We concluded that it was, for two reasons. First, a t-test is equivalent to an F-test with one additional variable; hence, a significant t-statistic denotes a significant improvement in R-squared even though the improvement may not be registered at the second decimal place. Second, we decided to enter the variables in the model in this particular order to establish a baseline model that would allow us to determine whether there was any net effect for the status variables after the controls were entered. Obviously, the amount of variance attributed to a particular variable is contingent on order of entry. If vineyard status and the interaction terms were entered into the regression first, then more of the increase in R-squared could be attributed to these variables, but such an approach would not allow us to assess a net effect for the variables of interest.
6 When the interaction terms are excluded from the analysis, the only difference from the results based on the network measure is that the main effect for bottle status is positive and significant, whereas the main effect for three-year status is positive and not significant. Since there are good theoretical reasons for believing that the effect of three-year status interacts with other variables and since model 4 provides some empirical support for this prediction, there seems little reason to focus on differences in what is clearly an incomplete model.
7 This example is drawn from an interview with a Mendocino winemaker who stated that such trade-offs between blending for higher quality or choosing not to blend so as to maintain the appellation affiliation are quite common. He reported that on many occasions both he and other winemakers choose to maintain the appellation affiliation rather than blend away imperfections in their products.
1984 “Reputation and product quality.” Rand Journal of Economics, 15:311-327.
Amerine, Maynard A.
1984 “Sensory evaluation or how to taste wine.” In D. Muscatine, M. A. Amerine, and B. Thompson (eds.), The Book of California Wine: 448-452. Berkeley, CA: University of California Press.
Asch, Solomon E.
1940 “Studies in the principles of judgments and attitudes II: Determination of judgments by group and by ego standards.” Journal of Social Psychology, 12: 433-465.
Baum, Joel A. C., and Christine Oliver
1991 “Institutional linkages and organizational mortality.” Administrative Science Quarterly, 36: 187-218.
Belsley, David A., Edwin Kuh, and Roy E. Welsch
1980 Regression Diagnostics. New York: Wiley.
1987 “Power and centrality: A family of measures.” American Journal of Sociology, 92: 1170-1182.
Bowles, Samuel, and Herbert Gintis
1976 Schooling in Capitalist America: Educational Reform and the Contradictions of Economic Life. New York: Basic Books.
1980 “Autonomy in a social topology.” American Journal of Sociology, 85: 892-925.
1992 Structural Holes: The Social Structure of Competition. Cambridge, MA: Harvard University Press.
1992 “Reputation and predecessor selection: Parsons and the institutionalists.” American Sociological Review, 57: 421445.
Cohen, Michael D., James G. March, and Johan P. Olsen
1972 “A garbage can model of organizational choice.” Administrative Science Quarterly, 17: 1-25.
1991 “Agents without principles? The spread of the poison pill through the intercorporate network.” Administrative Science Quarterly, 36:583-613.
Elias, Norbert, and J. L. Scotson
1965 The Established and the Outsiders: A Sociological Inquiry into Community Problems. London: Frank Cass.
Fernandez, Roberto, and Nancy Weinberg
1997 “Sifting and sorting: Personal contacts and hiring in a retail bank.” American Sociological Review, 62: 883-903.
1991 “Is there gout de terroir out there? Findings of the First Annual Grand Harvest Awards: Charting the sensory landscape of America’s viticultural areas.” Vineyard and Winery Management, March-April: 32-34.
Granovetter, Mark S.
1974 Getting a Job: A Study of Contacts and Careers. Cambridge, MA: Harvard University Press.
1985 “Economic action and social structure: The problem of embeddedness.” American Journal of Sociology, 91:481-510.
Granovetter, Mark S., and Richard Swedberg
1992 The Sociology of Economic Life. Boulder, CO: Westview Press.
1994 “Mimetic isomorphism and its effect on the audit services market.” Social Forces, 73: 637-664.
1994 “How much is that company worth? Interorganizational relationships, uncertainty, and acquisition premiums.” Administrative Science Quarterly, 39: 391-411.
Hovland, Carl I., Irving L. Janis, and Harold Kelley
1953 Communication and Persuasion. New Haven, CT: Yale University Press.
1984 Econometric Methods, 3rd ed. New York: McGraw-Hill.
1994 “The emergence of exchange structures: An experimental study of uncertainty, commitment, and trust.” American Journal of Sociology, 100: 313-345.
1992 Making Sense of California Wine. New York: Morrow.
1987 Science in Action. Cambridge, MA: Harvard University Press.
1935 Dynamic Theory of Personality. New York: McGraw-Hill.
Manfreda, John, and Richard Mendelson
1988 “U.S. wine law.” Unpublished manuscript, Sonoma County Wine Library, Healdsburg, CA.
1984 “The economics of wine in California.” In D. Muscatine, M. A. Amerine, and B. Thompson (eds.), The Book of California Wine: 380-405. Berkeley, CA: University of California Press.
1961 “Organizational prestige: Some functions and dysfunctions.” American Journal of Sociology, 66: 335-341.
Pfeffer, Jeffrey, and Gerald R. Salancik
1978 The External Control of Organizations. New York: Harper and Row.
Podolny, Joel M.
1993 “A status-based model of market competition.” American Journal of Sociology, 98: 829-872.
1994 “Market uncertainty and the social character of economic exchange.” Administrative Science Quarterly, 39: 458483.
Podolny, Joel M., and Damon Phillips
1996 “The dynamics of organizational status.” Journal of Industrial and Corporate Change, 5: 453-472.
Podolny, Joel M., and Toby E. Stuart
1995 “A role-based ecology of technological change.” American Journal of Sociology, 100: 1224-1260.
Podolny, Joel, Toby E. Stuart, and Michael T. Hannah
1996 “Networks, knowledge, and niches: A sociological examination of worldwide competition in the semiconductor industry.” American Journal of Sociology, 102: 659-689.
Raub, Werner, and Jeroen Weesie
1990 “Reputation and efficiency in social interactions: An example of network effects.” American Journal of Sociology, 96: 626-654.
Ross, Lee, and Richard E. Nisbett
1991 The Person and Situation: Perspectives of Social Psychology. New York: McGraw-Hill.
1983 “Premiums for high quality products as returns to reputations.” Quarterly Journal of Economics, 98: 659-680.
Stuart, Toby E., Ha Hoang, and Ralph C. Hybels
1999 “Interorganizational endorsements and the performance of entrepreneurial ventures.” Administrative Science Quarterly, 44:315-349.
Stuller, Jay, and Glen Martin
1989 Through the Grapevine: The Great Story Behind America’s Eight Billion Dollar Wine Industry. New York: Wynwood Press.
1984 “The critics look at California wines.” In D. Muscatine, M. A. Amerine, and B. Thompson (eds.), The Book of California Wine: 468-536. Berkeley, CA: University of California Press.
1992 Scanning the Trends in the Wine Industry. San Francisco: Hambrecht & Quist.
1994 The Theory of the Leisure Class. (Originally published in 1899.) New York: Dover.
Wiewel, Wim, and Albert Hunter
1985 “The interorganizational network as a resource: A comparative study of organizational genesis.” Administrative Science Quarterly, 30: 482496.
1985 “Reputations in games and markets.” In Alvin Roth (ed.), Game Theoretic Models of Bargaining: 27-62. Cambridge: Cambridge University Press.
Beth A. Benjamin [coauthor, “Status, Quality, and Social Order in the California Wine Industry”] is the director of the Stanford Consortium on Global Organization at the Graduate School of Business, Stanford University, Stanford, CA 94305-5015 (e-mail: email@example.com). Current research interests focus on the allocation of decision rights in global organizations and the development of global leadership skills. Recent publications include Building Leaders: How Successful Companies Develop the Next Generation, with Jay Conger (Jossey-Bass, 1999), and “Implementing Change: Lessons from the Civil Justice Reform Act” (Institute for Civil Justice, Facts and Trends, 1997), completed while she was an organizational behaviorist at the RAND Corporation. She received a Ph.D. in organizational behavior from the Graduate School of Business, Stanford University.
Joel M. Podolny [coauthor, “Status, Quality, and Social Order in the California Wine Industry”] is a professor of strategic management and organizational behavior at the Graduate School of Business, Stanford University, Stanford, CA 943055015 (e-mail: firstname.lastname@example.org). His research interests include the significance of organizational status for market competition, the role of social networks in the workplace, and the structural determinants of organizational learning. Recent publications include “Social Status, Entry, and Predation: The Case of British Shipping Cartels 1879-1929,” with Fiona M. Scott Morton (Journal of Industrial Economics, 47:41-67) and “Resources and Relationships: Social Networks and Mobility in the Workplace,” with James N. Baron (American Sociological Review, 62: 673-693). He received his Ph.D. in sociology from Harvard University.
COPYRIGHT 1999 Cornell University, Johnson Graduate School
COPYRIGHT 2000 Gale Group