OrganizationScienceVol. 22, No. 6, November–December 2011, pp. 1449–1463issn 1047-7039 �eissn 1526-5455 �11 �2206 �1449 http://dx.doi.org/10.1287/orsc.1100.0619

© 2011 INFORMS

Category Reinterpretation and Defection: Modernism andTradition in Italian Winemaking

Giacomo NegroGoizueta Business School, Emory University, Atlanta, Georgia 30322, giacomo_negro@bus.emory.edu

Michael T. Hannan, Hayagreeva RaoGraduate School of Business, Stanford University, Stanford, California 94305

{hannan@stanford.edu, rao_hayagreeva@gsb.stanford.edu}

When two groups of market actors differ in how to interpret a common label, each can make claims over the label. Onecategorical interpretation and the group that supports it risk disappearance if the rival interpretation gains ground. We

argue that when members of the endangered category become partial defectors that span categories, their history presentschallenges to the identity of nondefectors that will inhibit further change. Our empirical analysis of “traditionalism” and“modernism” in the making of Barolo and Barbaresco wines supports this argument.

Key words : organization and management theory; economic sociology; organizational ecologyHistory : Published online in Articles in Advance March 11, 2011.

IntroductionThis study examines the implications of widespreadchallenges to conventional classifications arising fromproducers’ participation in multiple categories. Thesechallenges are more than static anomalies for the under-standing of markets, which typically thrive on con-formity to institutionalized arrangements and practices.Memberships in categories also reflect collective iden-tities considered worth preserving, so producers whoreject the limits imposed by categorical conventionsstand as refutations that can affect change in the systemof categories.

Current research finds that multiple-category member-ship entails various kinds of disadvantages (Zuckerman1999, Hsu 2006, Hsu et al. 2009, Negro et al. 2010).This work assumes that audiences preserve agreementabout schemas—the sociocognitive representations usedto interpret reality. However, category straddling canweaken the strength of categorical beliefs (Hannanet al. 2007). Straddling undermines consensus becauseit introduces disagreement about which features are con-sidered typical of the category, and disagreement can ini-tiate disputes involving different groups of market actors,particularly producers.

We examine such disputes and consider two factors—the object and the subject of the claims made by thesedifferent groups—that we expect will influence how theyare resolved. We argue that situations of category forma-tion in which the emergence of a new consensual schemais associated with a new label (the category’s descrip-tive tag) differ from situations in which new schemas areapplied to an existing category label. We also claim that

the implications of these disputes differ by the group thatinitiates them—in particular, whether the reinterpretationclaim is initiated by insiders (producers who defectedfrom a category) or by outsiders.

We compare the cases of claims made over new andexisting labels and of claims made by insiders and out-siders, and we argue that schema disagreement is mostproblematic when insider defectors make opposing inter-pretations of an existing label. Associating an existinglabel with a new schema is what we refer to as categoryreinterpretation.

We introduce two broad arguments. First, categoryreinterpretations can be perceived as a threat by the pro-ducers who support the ex ante consensus about themeaning of the category label—the “loyalists.” Second,insider defectors assume particular relevance in marketsbecause they have a history; they were once typicalmembers of the original category. Hence their defec-tion makes them disloyal to their former identity andalso validates the competing interpretation as a legiti-mate claim to the label. Insider defections are thus morethreatening than the entries of producers without priorexperience that associate with the reinterpretation.

Spanning interpretations, taking actions that align witheach interpretation, has special significance. Spanningcan arise from the actions of defectors, what we term“partial defectors,” or new entrants (de novo categoryspanners). De novo spanners are more marginal thandefectors and, as such, are expected to be less typicalmembers of a category (Hsu et al. 2011). We think thatspanning by defectors matters more. Partial defectors’former category membership and bearing of a common

1449

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and Defection1450 Organization Science 22(6), pp. 1449–1463, © 2011 INFORMS

label pose a fundamental categorization problem, andloyalists risk losing their categorical specificity or evenbeing assimilated into the rival category. This circum-stance often sparks efforts by loyalists to defend theoriginal, traditional view and to insist on the distinctionbetween schemas.

Finally, the defense by loyalists not withstanding,category straddling can eventually cause audiences toupdate their views on the meaning of the category. Ifinsider defectors persist in category spanning, audienceswill gradually redefine their assumptions about what fea-tures are typical of the bearers of the label and cometo accept fuzziness of category boundaries as natural.By this kind of process, increased category spanningcan reduce the appeal of the offerings associated withthe label. If the label loses its power to shape expecta-tions, loyalists are less likely to mobilize to defend theirposition.

We study the trajectory of changes in the interpretationof a pair of classic Italian wines, Barolo and Barbaresco.The opposed meanings applied to these labels wereanchored in the practices of vinification. These wines,made in the Langhe, at the southeastern corner of Italy’sPiedmont—close to the French and Swiss borders—are generally regarded as among the world’s greatestwines. A reinterpretation of Barolo/Barbaresco emergedto challenge the prevailing tradition after some vintnerschose to use wine-aging practices that did not respectthe region’s established practices of winemaking. Thischallenge evolved into an opposition between categoricalmeanings. The “modernist” reinterpretation gained favoramong a new generation of Barolo/Barbaresco producersand received the praise of wine critics. In response, “tra-ditionalist” producers constructed a collective identity indefense of the original interpretation.

To lay out this account, we rely on material from inter-views with 45 winemakers in their cellars, with winejournalists and enologists in the Piedmont area and else-where in Italy during 2005–2007. We also conduct astatistical analysis of the response to the categorical dis-pute. We empirically examine the rate of defection fromthe use of the old-style aging practices that criticallydefined the distinction between tradition and modernismof Barolo/Barbaresco winemaking. In general, this studybrings attention to the active role played by insiders inregulating collective category dynamics and shows thatexpressions of identity can be seen as solutions to main-taining categorical distinctions.

Category Interpretations, Fuzziness,and DefectionsCategories are semantic objects; for purposes of soci-ological analysis, they can be considered to be socialagreements about the meanings of labels applied tothem. Meanings can be represented as schemas that tell

which feature values are consistent with membership inthe category and which are not. Familiar examples ofsuch schemas are the codes specifying genres in graph-ical art (Becker 1982, Fine 2004), literature (Griswold1987), films (Zuckerman and Kim 2003, Hsu 2006),music (Peterson 1997, Grazian 2003), cuisine (Rao et al.2003, Carroll and Wheaton 2010), and beer (Carrolland Swaminathan 2000). In each case, the prevailingschemas tell what features are relevant for judging mem-bership in a category.

Producers and their offerings often fit only partiallyto widely accepted schemas. For instance, Hsu (2006)reports that critics assign feature films to slightly morethan three genres. Because the genres impose differentconstraints, a film that gets assigned to multiple gen-res cannot fit perfectly in any one of them. (Hannan2010 reviews a rapidly growing body of research thatdocuments the generality of such partial assignments ofmembership.)

Hannan et al. (2007) argue that the issue of partial-ity of membership ought to be at the center of analysisof the emergence and persistence of categories. Theseauthors follow a branch of cognitive science in definingcategories as fuzzy semantic objects whose boundariesare not necessarily sharply delineated. In this view, aproducer’s membership in a category (for an audience)reflects the degree to which its feature values fit audi-ence members’ schemas for the category.

Fuzzy-set theory allows partial memberships in sets.A fuzzy set is defined by a grade-of-membership (GoM)function, which maps objects in some universe of dis-course to the 60117 interval. An object’s GoM in a cate-gory (or degree of typicality as a member of a category)from the perspective of an audience member tells thedegree to which its perceived feature values fit herschema of the category.

A social category emerges when the members of anaudience come to substantial agreement about what alabel means, and a category persists so long as theaudience retains a high level of such intensional con-sensus. Actions by category members and assessmentsby the audience affect the emergence and persistenceof consensus, a central component of market processesbased on categorization and valuation (White 1981,Zuckerman 1999, Cattani et al. 2008). Consensus morelikely emerges when the objects being labeled and clas-sified are highly similar. Likewise, increasing diversity(and violations of category schemas) after categorizationthreatens the durability of a consensus.

The clarity of a category’s boundary can be under-stood in terms of fuzziness. A category has sharpboundaries if audience members seldom assign low ormoderate GoM to bearers of the category label; bound-aries are weaker if such partial assignments of member-ship are common. The concept of category contrast (and

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and DefectionOrganization Science 22(6), pp. 1449–1463, © 2011 INFORMS 1451

its mirror concept fuzziness) captures this idea. The con-trast of a category is the average GoM for those withsome positive degree of membership. In other words,in a high-contrast category, producers are generally per-ceived to be either nearly full-fledged members or virtu-ally not members at all. And the higher the contrast ofa category, the lower is its fuzziness.

When fuzziness increases, the producers to whichaudience members apply a label tend to differ on valuesof schema-relevant features. Such dissimilarity sparksdisagreement about the meaning of the label and aboutwhat producers deserve the label. In making this argu-ment, we build on the notion that consensus about themeaning of a category decreases as its fuzziness—as per-ceived by the audience—increases (Hannan et al. 2007).Fuzziness increases when producers straddle categoryboundaries, that is, when they adopt practices and pro-duce offerings that (partially) fit more than one category(Hsu et al. 2009). Below, we argue that the presenceof producers with positive GoM in a pair of categoriesmakes the boundary problematic and arguably increasesits salience, especially to the full members, if categoricalidentities get constructed as oppositions.

We think it is important to distinguish cases in whichthe emergence of a new consensual schema is associ-ated with a new label (category emergence) from thosein which it is applied to an existing category label (cat-egory reinterpretation). We claimed that it also matterswhether the new collective schema comes from insid-ers (defector initiated) or outsiders (de novo initiated).Consideration of these distinctions suggests four maintrajectories, depicted in Figure 1.

Consider first the situation with different labels formultiple categories, where increased fuzziness is drivenby de novo spanning (quadrant D). In art worlds, cat-egory structures often evolve in a dialectical form, butart movements typically involve new cohorts of artistswho propose new styles and use different labels torefer to them. In a study of American avant garde art,Crane (1987) argues that leading members of a styleexhibit most or all characteristics of a style, whereasmarginal members exhibit only one or two—so cate-gory spanning is limited and carried on by new artists.The categorical tension that might erupt when multi-ple categories engage the audience tends to be resolvedvia artists’ replacement and use of new labels. For

Figure 1 Four Trajectories to Change of Categories inMarkets

Labels forreinterpretation

Common Distinct

InsidersOriginators of (defection)reinterpretation New producers

(de novo)

A B

C D

example, Crane notes that when “minimalism” replaced“abstract expressionism,” artists who continued to iden-tify as abstract expressionists became isolated and didnot even see each other’s work.

Quadrant B in Figure 1 represents situations whereinsiders are the source of the categorical disagreementbut different labels are applied to the new and oldschemas. In the context of French gastronomy, chefsdefected from “classical” cuisine to “Nouvelle” cuisine,a category that was distinctly labelled and codified, andthe label extolled experimentation and combination (Raoet al. 2003, 2005). The two categories did not competefor a single, true interpretation of French cuisine. Theflow of defections from classical cuisine sustained thegrowth of Nouvelle cuisine and seemed to slow downonly when rampant category spanning incurred disfavorfrom the confused critics.

A collective reinterpretation of an existing categorycan yield a pair of opposed categories (Quadrant C).Consider the American beer industry. The microbrew-ery movement supported artisanal production and devel-oped into antagonism to industrial, mass-produced beer.Carroll and Swaminathan (2000, p. 725) quote onemicrobrewer as saying, “Today’s craft brewing move-ment is a reaction against the mongrelization of beer,”whereas another adds, “I would hesitate—dare I say—tocall some of that mass produced stuff ‘beer.”’ A relatedexample of a category reinterpretation is the British pre-cursor to the American microbrewery movement, the“Campaign for Real Ale,” whose original name was the“Campaign for the Revitalisation of Ale.” In each case,the rival categories claim the original label, “beer” and“ale,” respectively.1

Carroll and Swaminathan (2000) pay particular atten-tion to so-called contract brewers. These are firms thatmake claims of being authentic craft brewers but do notown brewing facilities or make their own beer. Industrialbrewers also make beers that claim to be craft brews ortake equity positions in craft brewers, engaging in par-tial defection and resulting in fuzzier boundaries. Thecontention about what constitutes authentic craft beerwas intense. The insiders—microbrewers and the enthu-siast audience—accused contract firms of being fakesand “virtual” brewers (Carroll and Swaminathan 2000,pp. 727–728). They considered themselves true to theoriginal style.

The growth of contract brewers was challenging formicrobreweries to address. However, contract brew-ers were mostly outside businesses hiring breweries tomake their product and could be discredited for pass-ing (Goffman 1963) as microbrewers. Rao et al. (2000,p. 264) note that “craft-brewing enthusiasts policed pre-tenders to their identity by quickly ridiculing them asimpostors. Arguably, the policing of inauthentic incur-sions by enthusiasts sustained the identity of craft-brewing and spurred the growth of the movement.” And

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and Defection1452 Organization Science 22(6), pp. 1449–1463, © 2011 INFORMS

we think the effective policing of the category boundarycalmed down the activation of microbrewers.

Quadrant A directs attention to a situation with rivalinterpretations of a common label and fuzziness stem-ming from insiders’ partial defections. This is thesituation we theorize below, drawing on the contextof Italian winemaking that we studied. In the empiri-cal case we study, dissatisfied producers reinterpretedthe conventional practices for making of “Barolo” and“Barbaresco” but claimed these labels for their wines.Written legal codes, called disciplinare di produzione,mandate the properties required to apply the label Baroloor Barbaresco to a wine (Caldano and Rossi 2004).The codes also specify the maximum allowable yield,minimum alcohol content, and a variety of chemicalproperties. Crucially for our analysis, they allowed dis-cretion in choice of aging technology: vintners candecide whether the barrels are made of oak or chestnut,and they have free choice on the sizes of the barrels. Formore than a century, Barolo/Barbaresco makers reliedon very long maceration, uncontrolled fermentation, andaging in Slovenian oak casks (botti grandi) that can beas large as 120 hectoliters. These practices producedaustere wines, which are very tannic when young andrealize their full potential only after considerable aging.Quality was uneven.

The choice of what kind of wood containers to usefor aging became a fault line. A group of rebels beganto challenge the prevalent technology of winemaking byusing barriques, small (usually 225-liter) barrels madeof aromatic French oak, and changing other vinificationpractices. Aldo Vacca, managing director of the Coop-erativa Produttori del Barbaresco, summarized matterssuccinctly:2

In the late 1970s and early 1980s, this area was exposedto new huge markets, first the U.S.A. and then Japan. Ourwines have strong tannins, strong acidity, not too muchfruit, and are not so balanced. This type of style wasgood for collectors. 0 0 0 To enter those [new] markets thewine should be a little less tannic, a little “easier.” Lotsof producers started to produce easier-to-drink and ready-to-drink wine (a maximum of six years to be ready) andnot super tannic.

Gianni Fabrizio, an editor of the Gambero Rosso wineguide, explained,

Barriques fix the antocyans, so the color is deeper. Sec-ond, they induce higher exchange of oxygen, resulting insofter tannins. According to the modernists, the biggestproblems of Barolo were the lack of color and the pres-ence of too much tannin. These characteristics were per-ceived similarly by journalists, who thought that thepublic wouldn’t love this (old) kind of wine.

The use of the small French barrels became a sym-bol of modernity in making Barolo and Barbaresco.Rebellious winemakers strongly criticized the old-style

Barolo/Barbaresco wines as defective or stylisticallystale. But they continued to claim the original label fortheir wines. Elio Altare, a leader in a collective breakwith tradition, recounted the story as follows:

I was really ambitious, and I still am. If you are ambi-tious, you seek to compete only with the best. I asked tomyself, “Who can forbid a young producer to compete, tomake investments?” Who can tell you, “You don’t changebecause I am Barolo! There are rules!” For me a goodwine must be good always, not just after 20 years. If thewine is good, I want to enjoy it now.

Altare undertook a notorious radical act that symbol-ized the cultural break and put the focus squarely on thechoice of barrel. After spending time with winemakersin Burgundy, he tried without success to convince hisfather to use barriques in the family cellar. In 1982, hisfrustration with the constraints of tradition led him todestroy the family’s botti with a chainsaw. His daugh-ter Silvia told us, “When Elio started to make all thesechanges, my granddad didn’t talk to his son for almost10 years. He was so disappointed he even stopped goingto church because the people in the village thought thatmy father was crazy.” Indeed, his father disinherited him.

Far from being outcast, he instead became the leaderof a faction. Daniel Thomases, an evaluator for both theI Vini di Veronelli and Robert Parker’s Wine Advocate,characterized Altare’s significance:

The revolution introduced by Altare consisted principallyof an extremely short fermentation of the wine—threeor four days—and the use of small oak barrels insteadof larger casks. Altare had a very important influencebecause he was from La Morra, a town with many youngproducers who were just beginning to market their wine.He basically created a school in his own area.

The evidence we have collected suggests the bar-rique/botti distinction serves as what Hannan et al.(2007) call a minimal test code for assessing member-ship in these categories. Knowing only that a produceruses the French barrels leads outside audience membersto treat as default that they also use most, if not all,of the harder-to-observe “modernist” practices of vini-fication. Incorporating an attribute that signals a styleincreases the degree of membership in the category asso-ciated with that style (Crane 1987).

The members of the audience clearly interpreted theuse of barriques as a practice that determines categorymembership over the period that we study. The choice ofbarrel/cask is durable, because the wine ages for three tofive years in the cellar; thus this choice is easily observ-able to anyone who visits. Other aspects of vinificationcan be observed only during brief periods in the wine-making process.

Using barriques for aging Barolos and Barbarescoswas adopted as an emblem by producers seeking toenhance quality. Informal groups of experimenters began

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and DefectionOrganization Science 22(6), pp. 1449–1463, © 2011 INFORMS 1453

to meet, taste each others’ wines, and exchange ideas.According to Claudio Conterno of the Conterno-Fantinowinery, “This big group of 20 to 30 producers changedthe history of this region in the sense that they madepeople talk again about Barolo and, most important, theymade people open bottles of Barolo again.”

These informal networks aided the construction of anew common schema for what the journalists and critics(but not the protagonists) began to call modern Barolosand Barbarescos. Altare (along with other producers likeClerico, Sandrone, and Scavino) sought to frame thebarrique as an essential tool for reliability in craftinghigh-quality wines that appealed to customers. Manyinsurgents claimed that old-style Barolo/Barbaresco wasbad wine—too tannic, unbalanced, oxidized, and con-taining off-putting aromas. They did not simply arguethat winemaking practices should be reformed by adopt-ing technical innovations. They challenged the validityof those practices and implicitly the basic values under-lying the cultural model in which the practices wereimmersed.

The new-style wines found favor with the critics andthe market. One wine journalist explained that, when 50or more wines are tasted in a day, those with soft tan-nins and a fruit-forward taste stand out more than austerewines. Critical praise translates into success with audi-ences in the wine industry. An analysis of retail pricedata for Barolo and Barbaresco reveals that (i) wineswith higher ratings receive substantially higher prices: asuper three-star wine sells for $32 more on average thana one-star wine; and (ii) over and above this, wines madein the traditional style command lower prices net of thecritical ratings—roughly $9 less on average (Negro et al.2007). Thus a difference in critical reception translatesto a substantial difference in marketplace reception.

The use of a single label (either Barolo or Barbaresco,depending on the location) for two styles (traditional ormodern) poses a problem of interpretation. If segmentsof the audience associate different schemas with a label,then communication about the producers, their practices,and their products becomes awkward. Participation inthe activities associated with the label loses appeal forthe members of the audience under such conditions. Thissituation of low consensus on the meaning of a categorylabel puts the category at risk of dissolution and providesfertile ground for contention.

In response, the producers who began to see them-selves as “traditionalists” insisted on a sharp categoricaldistinction. In 1997, Bartolo Mascarello wrote an openletter to the Consorzio (the industry association respon-sible for monitoring the use of the disciplinare code)criticizing “French-Californian models” and issued theclarion call, “There is one and only one Barolo; defendit!” He started using the label on his wines “Il ne fautpas faire des barrique mais des barricades” (Make bar-ricades, not barriques).

“Traditionalists” argued publicly that the “modernist”wines were no different than “international” wines, didnot reflect the terroir and traditions of Barolo andBarbaresco, and were therefore inauthentic. GiacomoConterno (of the Poderi Aldo Conterno winery) arguesthat Barolo producers should seek to “intensify the per-sonality of Nebbiolo and its site identity, to make theopposite of a Coca-Cola wine.” For him, this meansrejecting barriques.

Others offered a more fundamental critique. Theyargued that defecting from the old ways represents acapitulation to the market and constitutes a kind of moraltransgression. This critique drew an association betweenrampant market orientation and a lack of integrity. MariaTeresa Mascarello, who took over Cantine MascarelloBartolo after her father Bartolo’s death, put it this way:“My father always said that he wasn’t embarrassed ofour land. 0 0 0 He said we had to let the wine taste likewhere it came from. You can’t violate your environment;you can’t make it something it’s not” (Esposito 2008,p. 177). Likewise, Giuseppe Rinaldi, a highly regardedtraditionalist, proclaimed,

When you have a wine that is really tannic like Barolo—and this feature is very typical, you have to accept Baroloas it is. You don’t have to make a wine for everybody. Ifyou don’t like it, don’t drink it. The main problem is theinterference with the nature of the wine just to make itmore attractive for the market. It is a damned bad thing,because a land like this, with a history, a tradition, and anold identity, erased everything it had in order to becomemore attractive on the market.

Fuzziness and the Rate ofCategory DefectionThe success of the new-style wines induced someold-style producers to straddle by making wines in bothstyles (and often by combining practices from bothstyles). Defection from tradition has risen. Figure 2depicts variation over time in the proportion of “tradi-tionalists” who have started using barriques in the focalyear. Note that the (yearly) probability of defection rosein the 1990s and peaked at over 0.30 in 1993 beforedeclining to roughly 0.04 in the last four years of obser-vation. The wines made during the period of greatest rateof defection from the traditionalist camp were reviewedby the guide and released on the market during the years1996 and 1997.

Straddling categories signals a lack of commitment toany category and highlights issues of identity associatedwith category membership. The behavior of straddlersconfuses the audience, because many bearers of a cat-egory label engage both interpretations of its meaning.Experts argued that answering the question “traditionalor modern?” is not simple anymore and that the middleground of producers who combine botti and barriqueshas exploded, which makes classification challenging(Esposito 2008).

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and Defection1454 Organization Science 22(6), pp. 1449–1463, © 2011 INFORMS

Figure 2 Proportion of “Traditionalists” Defecting to“Modernism” By Vintage

0.4

0.3

0.2

0.1

0

Prop

ortio

n of

“tr

aditi

onal

ist”

def

ecto

r pr

oduc

ers

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

Vintage year

It is not surprising that new producers also beganusing barriques in their first vintages, and this trendincreased over time. For instance, for vintages coveredby I Vini di Veronelli before 1990, only 25% of theBarolo/Barbaresco producers reviewed for the first timeused at least some barriques for their wines. During1990–2001, however, 40% of the wineries entering theguide used them. If producers from the initial categoryswitched to the emerging one by adding feature valuesthat fit the rival interpretation and dropping those thatfit the original interpretation, then the boundary betweenthe categories would remain sharp.

“Traditionalists” challenged modernism in general butsaw distinctions between de novo straddlers and partialdefectors. Giuseppe Rinaldi argued that

[n]ew producers were not proud to be winemakers, theywere not proud to be part of the tradition, as if they wereashamed; until few years ago being a farmer was verydifficult 0 0 0many new producers started to follow [earlyinnovator, Angelo] Gaja 0 0 0 this was positive, becauseyoung people, instead of leaving the country, started tocultivate the land.

Straddling by new entrants added to the confusion.Nonetheless, partial defections are more challenging asa form of straddling. What makes partial defectors prob-lematic is the sequence of mobility. Partial defectorsused to be fully affiliated with the original category,but now they display some features inconsistent withthat category and others that are consistent with it.Hsu et al. (2011) argue that earlier affiliations typecastsocial actors into a category and reduce their chances offuture affiliations to other categories (if the categoriesare highly taken for granted). In this sense, partial defec-tors displayed some features typical of the original cate-gory schema, and audience members are likely to regardthem as members of that interpretation of the category.

Partial defection exposes a contradiction, because thedefectors can be associated with both interpretations.The loyalists, the remaining members of the original cat-egory, who are still perceived as similar to the defectors,see a risk of losing control of their collective identity.To them, this condition is thornier than being defeatedby rivals; it is more like losing a connection to their ownhistory.

Maria Teresa Mascarello made the connectionexplicit:

However, there is another type of traditionalist, who sim-ply are opportunists. They canceled the experiences accu-mulated in this world just to take other stories and othertechniques as models. 0 0 0 This is a matter of knowledgeof one’s own history, and a matter of dignity. 0 0 0 Wine-making for me is not improvisation. For me, it is thework that my father had transferred to me. So it is part ofmy identity. I don’t want to erase my roots, my history,because it identifies me.

Resharpening boundaries through commitment to “tra-dition” becomes a way for the remaining category mem-bers to clarify the situation and to prevent the initialschema from being submerged. And if the two inter-pretations are to coexist, something must prevent thewholesale defection of the loyalists (as seems to havehappened in French cuisine). Therefore we focus on thedefection rate among full-fledged “traditionalists.”

Our argument focuses on variations in the strengthof the “traditional” collective identity. A reinterpretationof a category label (especially if well received in somerelevant audience) threatens this identity. The loyalistsreact to this threat by more clearly articulating the col-lective identity as oppositional. As is the case generallywith oppositions, this strengthened interpretation of theoriginal categorical identity seeks to rule out the possi-bility of spanning. In the case of successful oppositions,a producer must sit in one camp or the other. Gainingagreement within the “traditional” camp on this matterstrengthens its collective identity. If the strength of theidentity regarded as oppositional gains widespread sup-port, then this changes the meaning of partial defection,making it a full defection from one’s former identity.This change in meaning makes partial defection an actof enduring disloyalty. Therefore, a strong oppositionalidentity ought to slow the flow of defections and stabi-lize the coexistence of the two interpretations.

According to this argument, variations in the intensityof the threat are crucial. Stronger threats are more likelyto heighten the salience of the identity and an effortto redefine it as oppositional. We argued that spanningcreates such a threat but that spanning by new produc-ers (with no prior categorical association) is much lessthreatening than defection by insiders.

We also think that the presence of insider defectorswho affiliate partly with the original interpretation isespecially problematic. These are the cases that most

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and DefectionOrganization Science 22(6), pp. 1449–1463, © 2011 INFORMS 1455

confuse matters and blur the boundary. Insider spannerswho remain highly similar in most respects to the loyal-ists pose the greatest identity threat.

As we explain, we calculated grades of membership inthe two styles in terms of the proportion of a producer’sofferings in a vintage that fits each style. So we translatethe argument above in terms of the GoM in the traditionof the insider defectors. Accordingly, we formulate theimplications of this argument in terms of the followinghypothesis.

Hypothesis 1. Higher average grade of membershipof insider partial defectors in the original category willlower the rate of defection from that category.

Adjusting expectations about membership requireseffort and time. But if partial defectors continue tostraddle categories, then the audience will make fewerassumptions (take less for granted) about whether pro-ducers’ features conform to a category schema. InBarolo/Barbaresco winemaking, more and more produc-ers displayed increasing diversity in methods of produc-tion and associated styles.

We asked intermediaries in wine markets about themore current relevance of the category distinction toconsumers and about the tie of the barrel/cask distinctionto the “modernist” interpretation of the category. Ourinterviews with wine merchants and sommeliers in Albaand Milan revealed that these agents perceived culturalfault lines in Barolo/Barbaresco that coincide with the“traditionalist/modernist” divide. However, a sommelierfrom a restaurant whose wine menu listed on separatepages the Piedmont wines aged in botti and those agedin barriques told us that his restaurant has “a wine chartwhere this kind of difference is highlighted. If there areintermediate situations, we put producers in one of thetwo categories according to our judgement.”

If partial defectors manifest the inconsistency betweeninterpretations, audience members will generally updatetheir beliefs and revise default assumptions (Hsu et al.2011). Shifting away from reliance on defaults aboutconformity increases fuzziness and lowers the contrastsand the standing of the two categories (pairings of thesame label with different schemas). As straddling con-tinued in the Langhe, the contrast of the “modernists”declined. Recently, several Barolo and Barbaresco pro-ducers admitted that there is a “range of possibilities” asto how wine can be made, somehow revising the starkdistinction that ruled in the past. This shift ought toweaken the contention about categorical membership.

Unlike the reasoning behind the first hypothesis,which concentrated on the views of producers, a secondargument brings the audience more squarely into the pic-ture. If the two interpretations of the label confuse theaudience and cause them to regard the whole set of pro-ducers as fuzzy, then there is less to defend. The identitystakes are lower once the audience has come to regard

the situation as inherently fuzzy. Critics then began tospeak of a middle ground in winemaking and of a blur-ring of stylistic differences between the camps of theBarolo and Barbaresco producers. These developmentsought to defuse the threat from modernism and weakenthe link between spanning and the strength of activationof a “traditionalist” identity.

We think that enduring spanning will lead the enthu-siast audience to come to regard the categorical divideas having ended. For us, this means that long durationsof spanning by once full-fledged traditional winemakerswill weaken the process claimed in the first hypothesis.Therefore, we hypothesize an interaction effect.

Hypothesis 2. Increasing average time elapsed sincepartial defection will attenuate the negative effect of par-tial defectors’ grade of membership in the original cat-egory on the hazard of defection.

Empirical Analysis of Defection from“Tradition”This part of the analysis uses data on vintners’ use ofaging techniques culled from the leading wine guide IVini di Veronelli (hereafter Veronelli guide) and fromtelephone interviews conducted in 2006 with all makersof Barolo/Barbaresco who were listed in the guide forat least one vintage. The crucial information comes fromindications for each wine listed whether it was aged inbotti only, barriques only, or a combination of botti andbarriques.

Dependent VariableUsing event-history methods we analyze the hazard ofdefection from traditionalism of the 192 producers whohad at least one Barolo or Barbaresco reviewed in theVeronelli guide in its first 15 editions and for whom dataon the practices used for wine aging are available. Wedefine a defection from “tradition” as a producer’s shift-ing from a history of exclusive use of botti to some useof barriques. Given the timing of the modernist reinter-pretation, we begin our analysis in 1980. We end theperiod of analysis with the 2001 vintage, because 2002and 2003 were poor vintages, and many producers didnot release Barolos and Barbarescos. The 2004 vintagehad not yet been reviewed when we built our data file.

The number of producer-years at risk of (first) defec-tion over 1980–2001 is 1,140; the number of observeddefections is 62. Thus the simple hazard of defection(in yearly time units) is 0.054. The implied probabilityof remaining traditional over the roughly 20-year his-tory since the rise of modernism would be approximately0.32 if this hazard were constant over time.

We adopt the piecewise exponential specification oftenure dependence in the hazard to allow the base rateof defection to vary flexibly with organizational tenure.This approach splits duration into pieces. The base

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and Defection1456 Organization Science 22(6), pp. 1449–1463, © 2011 INFORMS

failure rate is specified as constant within each time-piece, although these base rates can vary across pieces.As a result, the piecewise model does not require anystrong assumption about the exact form of durationdependence. After exploratory analyses, we decided thata reasonable specification uses break points at 5, 10,and 15 years, corresponding to four intervals, with thelast open on the right. The first segment includes spellswithin the first 5 years of tenure, the second segmentincludes spells between years 5 and 10, the third seg-ment includes spells between years 10 and 15, and thefourth segment includes spells after 15 years of tenure.

Independent VariablesWe conceptualize the two interpretations as styles (Hsuet al. 2007, 2009; Negro et al. 2010). We follow theidea that distinct styles are associated with each of thetwo methods of wine aging, old style (using botti) ornew style (using barriques). Accordingly, we definedthe function �4s1 x1 t5 that denotes the GoM of the pro-ducer x as an exponent of the style s at time t. Wemeasure a producer’s GoM in the old style in each vin-tage as the proportion of its Barolo/Barbaresco labelsthat received only botti aging. Then the GoM in the newstyle is the complement of this proportion: 1 minus theGoM in the old style.3 A producer’s style profile is apair of GoMs: 8�4trad51�4new59.

At one extreme, we have producers such as CantineBartolo Mascarello, Giuseppe Rinaldi, and Coopera-tiva Produttori del Barbaresco, whose GoM functionin the styles is 81109 because they age all theirBarolos/Barbarescos in botti. At the other extreme, wehave producers such as La Spinetta and Rocche deiManzoni, which use only barriques (GoM function80119). In between, we encounter wineries that combinethe styles in all wines, such as Einaudi and Marchesi diBarolo, whose GoM in the styles is 800510059.

To test our first hypothesis, we must measure thedegree to which defectors fit the schema for the tradi-tional style. From the producer-level GoMs we calcu-late two variables, the average GoM in the “tradition”of partial defectors (average membership in “tradition-alist” style of partial defectors) and the average GoMin the “tradition” of the producers who entered as cate-gory straddlers (average membership in “traditionalist”style of de novo spanners). We include the defector GoMin “tradition” to test the first hypothesis and add thede novo GoM in “tradition” to account for remainingeffects of category spanning. (The GoMs in “tradition”are lagged one vintage.)

To test our second hypothesis, we calculated the aver-age number of years spent since producers who were“traditionalist” started using barriques. We add the aver-age tenure of de novo category spanners to separatethe effects of schema updates by producer type. Ourhypothesis is tested with an interaction term between the

Figure 3 Grade of Membership in the “Tradition” of PartialDefectors and Their Tenure by Vintage

0

1

2

3

4

5

6

Ten

ure

as p

artia

l def

ecto

r

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

GoM

in “

trad

ition

” of

par

tial d

efec

tors

1985 1990 1995 2000Vintage year

GoM in “tradition” of partial defectorsTenure as partial defector

average GoM in “traditionalist” style of partial defec-tors and the average tenure of partial producers sincedefection. Figure 3 shows how the average GoM in “tra-dition” of partial defectors and the average tenure aspartial defector vary over the study period.

ControlsControl variables include the producer’s maximum rat-ing in the Veronelli guide in the prior vintage (to addresswhether capability and/or perceived quality affects theadoption of unconventional practices). The number ofdifferent wines produced is added as a control for organi-zational size—larger wineries might be more inertial andtherefore more resistant to trying the new approaches.The ratings and size variables are lagged one vintage.We add two dichotomous variables to address differ-ences in defection across locations and local dynam-ics of organizational mimesis. The first variable equals1 for producers located in La Morra (Altare’s town, ahotbed of the new modernism) and 0 otherwise. The sec-ond variable is equal to 1 for producers located in thetown of Barbaresco (where modernist precursor AngeloGaja’s winery is located) and 0 otherwise. These con-trols account for local influences on the adoption ofunconventional practices. Next, we include controls forthe density of traditionalist producers to account for thecapacity of producers in the “traditional” category tosustain collective action, and the number of de novospanners and partial defectors to account for populationdensity-based processes of categorical defection. Also,we control for the overall quality level of each vintageto account for weather-related fluctuations.4

Finally, we include a dichotomous variable equal to 1for vintage years from 1980 to 1985 (pre-1986 period)and 0 otherwise. As the “modernist” producers wereorganizing, a scandal broke out in March 1986 when 26deaths were linked to the consumption of Italian winesthat had been adulterated with methanol to increasealcohol content. Angelo Gaja recalled, “We startedto make money with this business only after 1986,

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and DefectionOrganization Science 22(6), pp. 1449–1463, © 2011 INFORMS 1457

after the methanol scandal. Before then, throughout the1980s 0 0 0wineries did not make money; merchants did.”In this sense, the methanol scandal facilitated the shifttoward modernization. The scandal taught consumersthat high-quality wines also meant higher prices and thatbuying from established winemakers was an importantguarantee of quality. Adding period effects allows us torule out a simple alternative story, that the pattern ofdefections simply tracks consumer tastes that vary overthe period; e.g., there was little demand for the modernwines before 1986, but it increased sharply and remainedhigh thereafter. The descriptive statistics and correlationsfor the variables used in the analysis of defections canbe found in Table 1.

ResultsTable 2 reports the estimates of hazard models to assessthe conditions under which producers defected (beginusing barriques). Model 1 includes effects of controlvariables only. The number of different wines made by aproducer, our proxy for organizational size, does not sig-nificantly affect defection rates. Information reported bythe Chamber of Commerce of Cuneo in 2000 reveals thatthere is not much size variation among these wineries;they are very small by international, and even national,standards. The median number of bottles produced fora vintage is 15,000 (Associazione Vignaioli Piemontesi2000). We do not find any significant effect of the num-ber of defectors or de novo spanners, but we do findsome effect of local imitation, as shown by the positiveeffect of the La Morra variable.

Perhaps more surprisingly, tenure does not affect thehazard of defecting. One could expect the strengthof the traditional identity to vary—and particularly toincrease—with tenure. In hindsight, we think the lack oftenure dependence makes sense in the context. Accord-ing to our reading of the situation, the “traditionalist”identity is construed in the debate over the insurgencyand its methods. Moreover, this category constructionbecomes salient to the relevant producers at roughly thesame time; in other words, the impact of the rise of theinsurgency on the identity of those who have not yetchanged their practices does not depend on their tenure.

The nonsignificance of the effect of tenure as a non-defector on the hazard also runs opposite to a poten-tial rival explanation to our argument: a mover–stayerprocess, in which tenure is spuriously correlated withmobility.5 Suppose that the set of winemakers using tra-ditional practices consists of two subtypes: a hard-coregroup committed to tradition and a set of pragmatistswho happen to have begun with the old practices butare not committed. As the gain from using small Frenchbarrels grows, the pragmatists will tend to defect, but thehard core (producers like Cappellano et al.) will neverdefect. Eventually, the set of “traditionalists” will con-sist entirely of the hard core, and the hazard of defection Table

1SummaryStatisticsan

dCorrelationMatrixAmongReg

ress

ors

forAnalys

isofHaz

ardofDefec

tionfrom

Trad

ition,19

80–2

001

Mea

nSt

d.de

v.(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)(1

2)(1

3)

(1)Max

imum

ratin

gv−

110

920

1002

0—

(2)Num

berof

winelabe

lsv−

120

140

1066

000

383

—(3

)Lo

catio

n:La

Morra

0014

200

349

−00

079

−00

092

—(4

)Lo

catio

n:Barba

resc

o00

093

0029

100

072

0009

1−

0013

0—

(5)Vintag

equ

ality

9002

603

−00

003

−00

050

0002

600

017

—(6

)Pre-19

86pe

riod

0089

0028

6−

0001

2−

0002

600

101

0000

6−

0009

2—

(7)Den

sity

of“traditio

nalist’’

catego

ryv−

139

0826

0700

152

0006

3−

0012

4−

0002

800

022

−00

419

—(8

)Den

sity

ofde

novo

span

ners

v−

150

1760

2100

144

−00

025

−00

134

−00

029

0026

4−

0025

700

487

—(9

)Den

sity

ofpa

rtialde

fectors v

−1

1403

1900

132

−00

053

−00

152

−00

037

0020

7−

0023

600

513

0094

6—

(10)

Ave

rage

mem

bershipin

“traditio

nalist”styleof

deno

vosp

anne

rsv−

1

0005

700

075

0008

600

083

−00

066

−00

011

0016

8−

0023

700

693

0043

300

344

—

(11)

Ave

rage

mem

bershipin

“traditio

nalist”

styleof

partiald

efec

tors

v−

1

0062

500

349

−00

123

0002

300

118

0002

9−

0001

300

336

−00

743

−00

517

−00

614

−00

428

—

(12)

Ave

rage

tenu

rein

catego

rysp

anning

ofde

novo

span

ners

v−

1

2095

1024

0013

900

031

−00

114

−00

031

−00

197

−00

340

0065

800

395

0039

800

447

−00

461

—

(13)

Ave

rage

tenu

rein

catego

rysp

anning

ofpa

rtial

defectors v

−1

1075

1079

0016

1−

0003

3−

0014

1−

0002

900

293

−00

307

0060

200

959

0094

800

482

−00

650

0046

4—

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and Defection1458 Organization Science 22(6), pp. 1449–1463, © 2011 INFORMS

Table 2 Determinants of the Hazard of Defection from Tradition, 1980–2001

Model 1 Model 2 Model 3

Tenure 0<u≤ 5 −4034 (2.72) −4087 (3.90) −3054 (6.17)Tenure 5<u≤ 10 −4054 (2.74) −5015 (3.91) −3084 (6.18)Tenure 10<u≤ 15 −4013 (2.76) −4077 (3.92) −3044 (6.21)Tenure >15 −5041 (2.81) −6002 (3.97) −4065 (6.16)Maximum ratingv−1 −00075 (0.154) −00121 (0.164) −00151 (0.172)Number of wine labelsv−1 −00002 (0.094) 00029 (0.092) 00022 (0.090)

Producer location:La Morra 1028∗∗∗ (0.291) 1022∗∗∗ (0.290) 1030∗∗∗ (0.285)Barbaresco 00259 (0.462) 00210 (0.449) 00274 (0.455)

Vintage quality −00002 (0.029) 00019 (0.041) 00029 (0.058)Pre-1986 period −1502∗∗∗ (0.285) −1402∗∗∗ (0.435) −1402∗∗∗ (0.445)Density of “traditionalist” category v−1 00019∗∗ (0.006) 00012 (0.011) 00002 (0.013)Density of de novo spannersv−1 −00042 (0.052) 00011 (0.073) −00013 (0.086)Density of partial defectorsv−1 00033 (0.018) 00010 (0.027) 00036 (0.033)Average membership in “traditionalist” category of de novo spannersv−1 00957 (3.00) −00624 (3.64)Average membership in “traditionalist” category of partial defectorsv−1 −2020∗∗∗ (0.586) −5079∗∗∗ (1.05)Average tenure in category spanning of de novo spannersv−1 00210 (0.176)Average tenure in category spanning of partial defectorsv−1 −1047∗ (0.615)Average membership in “traditionalist” category of partial defectorsv−1 2090∗ (1.21)

×Average tenure in category spanning of partial defectorsPseudo log-likelihood −115.9 −108.9 −104.2Model X2 16,119.7∗∗∗ 16,251∗∗∗ 17,482.5∗∗∗

No. of producers 192 192 192No. of events 62 62 62

Notes. Maximum-likelihood estimates of constant-hazard specifications. Standard errors (adjusted for clustering on producers) are inparentheses. v denotes a (rated) vintage year.

∗p < 0005; ∗∗p < 0001; ∗∗∗p < 00001.

will drop to zero. In such a case, no collective-identityactivation would be needed to cause the hazard to fall.

Model 2 tests the first hypothesis, that the success ofthe modernist insurgency in attracting partial defectionseventually discouraged further defection. It shows thatthe average GoM in the “traditionalist” category amongthe partial defectors has a negative and statistically sig-nificant effect on the hazard of defection. This effect isstrong in substantive terms: at its mean level, the GoM inthe “tradition” for partial defectors reduces the multiplierof the rate of defection by 74.7%. The inclusion of themembership variables improves model fit over the firstspecification (X2 = 13097, Pr < 0001, with two degreesof freedom). This evidence supports the first hypothesis.Moreover, the effect of the grade of membership in the“tradition” of the de novo producers who straddle “tra-dition” and “modernism” is small and not significant,which is also in line with our argument.

Model 3 tests the second hypothesis. The key inter-action effect between GoM in the “tradition” of partialdefectors with the average tenure of partial defectorssince defection is positive and significant, and the inclu-sion of the tenure variables improves model fit overthe previous specifications (compared with Model 2:X2 = 9041, Pr < 0005, with three degrees of freedom;compared with Model 1: X2 = 23037, Pr < 0001, withfive degrees of freedom). The negative effect of schema

inconsistency, therefore, attenuates with the time spent inspanning categories, which is in line with our argument.

Insight into the substantive implications of theseeffects can be gained by examining Figure 4, which plotsa three-dimensional surface whose vertical dimension isthe predicted multiplier of the hazard of defection. Thehistorical scenario we study begins at the front left edgeof the graph, where defectors from “tradition” had lowaverage tenure as defectors and a high average gradeof membership in “tradition.” As the “modernist” insur-gency gained traction, tenure in “tradition” increased,and GoM in the “tradition” of the partial defectors firstdecreased and then decreased. This process traces a pathtoward the middle region of the space, where the pre-dicted hazard is sharply higher. Eventually, tenure asa defector continued to increase, but average grade ofmembership in “tradition” stabilized. This causes thehazard of defection to also stabilize (at a level above theone at the origin). By our argument, the increasing fuzzi-ness caused by defectors moving more to the middle ofthe continuum between the pure types defuses the mobi-lization of the “traditionalists” to defend their collectiveidentity.

One might wonder why the contention between rivalcategories induced “traditionalists” but not “modernists”to close ranks. A symmetric activation could be expectedif the same form of category straddling occurred in thetwo camps. Our data indicate, instead, a skewed pattern.

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and DefectionOrganization Science 22(6), pp. 1449–1463, © 2011 INFORMS 1459

Figure 4 Effects of Average GoM in “Tradition” and Average Tenure of Partial Defectors on the Multiplier ofthe Rate of Defection

Early defections

Late defections

12

34

5

0

0.25

0.50

0.75

1.001.00.80.60.40.20543210

1.00

0.75

0.50

0.25

00

Avg. GoM in “tradition”

of partial defectors

Avg. GoM in “tradition”of partial defectors

Mul

tiplie

r of

rat

e de

fect

ion

from

“tr

aditi

on”

Mul

tiplie

r of

rat

e de

fect

ion

from

“tr

aditi

on”

Avg. tenure in “tradition” of partial defectors

Avg. tenure in “tradition”

of partial defectors

0.20.4

0.6

0.8

1.0 0

The number of producers moving away from traditionwas large and consisted mostly of partial defectors (N= 53) who also maintained a high GoM in the “tradi-tionalist” category (mean = 0.58). In contrast, we onlyfound five partial defections from modernism and onlyone defector maintaining a higher GoM in the “mod-ernist” category.

Alternative MechanismsWhat factors other than activation of collective “tra-ditionalist” identity could inhibit defection? One alter-native explanation is that some producers might havelacked the necessary capabilities or resources to adoptmodern vinification practices. The estimates in Table 2do not support this view. It was not the low-performingproducers that accounted for the bulk of the defections.The maximum rating from the prior vintage has aninsignificant negative effect on defection. If the mostcapable defected, then the capabilities involved did notshow up in the wines as judged by the critics. Also, inunreported analyses, we do not find significant evidencethat the rate of defection is affected by other organiza-tional variables such as ownership form.

Another possibility is that the prospective defectorsmight anticipate penalties and, therefore, decide notto defect. If those who defected would tend to getpenalized—they would not receive upgrades and would,instead, be downgraded by the critics. Negro et al.(2007) estimated the effect of defection on subsequentratings and found defection increases the probability that

a maximum rating rises and reduces the probability thatit falls. That is, defectors are rewarded with upgradesand are less likely to suffer downgrades.

Finally, perhaps broader shifts in the market accountfor the persistence of the rival interpretations. The infor-mation on export activities during our survey is clearlyincomplete, but it indicates that “traditionalist” produc-ers are not avoiding international competition. Of the 60producers for which data are available, the difference inreliance on exports by traditionalists (71.6% of sales)and modernists (71.0%) is not significant. The differenceis also not significant when “traditionalists” that did notdefect are compared with defectors (65.5%).6

One could still argue that “traditionalist” producersare responding only to global processes that partitionthe wine industry into a mass/modern segment and acraft/traditional segment. As noted above, all the Baroloand Barbaresco producers are very small in scale byworld standards. If the world wine industry is partition-ing, its epicenter ought to lie outside the Langhe regionand ought not to involve the opposition between cate-gories inside the Langhe that our results demonstrate.

DiscussionAs we see it, the modernist challenge and the recon-struction of the traditionalist identity as oppositionalconstitute a category reinterpretation. The modernistwinemakers critiqued practices in the vineyard and inthe cellar that had long been accepted and taken for

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and Defection1460 Organization Science 22(6), pp. 1449–1463, © 2011 INFORMS

granted. This reinterpretation manifested a disagree-ment that caused audience members to attend morecarefully to the characteristics of the category schema.The traditionalists’ reaction offered a reinterpretation indefense of the old practices but also in opposition tothe modernists’ claims, where the oppositional part ofthe category schema was new—there could not be acountercritique before the critique. What we observedis indeed a pair of oppositions that culminates in thetraditionalists’ attempt to make the boundary crisp, sim-ilar to the case of real ale and microbrews (Carroll andSwaminathan 2000).

Research on categories in markets (see Hannan 2010for a review) has largely focused on external audiences’categorization as an enabling and constraining factorfor valuation of producers and/or their offers. Previ-ous research has shown that association with multiplecategories causes producers to be perceived as poorerfits with category schemas than specialized counterpartsand that poorer fit reduces appeal to relevant audiences(Zuckerman and Kim 2003, Hsu 2006, Hsu et al. 2009,Kuilman and Li 2009, Ruef and Patterson 2009, Negroet al. 2010). Our study brings attention to the role ofinsiders by suggesting that collective identity can beorganized to maintain categorical distinctions. At thesame time, the findings highlight the role of identity het-erogeneity within categories. When market agents aimat differentiating at times, this can encourage efforts todistance their identity from those of their peers. How-ever, differentiation can threaten the consensus aroundschemas when distinct identities share the same cate-gory label.

Our study has potential implications for identity poli-tics (Bernstein 2005), which examines collective actionon the basis of identities, where identities form a partof the basis for grievances. Identity politics is politi-cal in the sense that interactions between groups, accessto state actors, and opponents determine the types ofidentities expressed; thus identity is one type of practicein which groups engage to transform dominant culturalpatterns (Armstrong 2002). Identity politics concernsidentity in the sense that constructing and reconstruct-ing collective category schemas is a goal more thanan instrument of collective action that offers fulfillment(Bernstein 1997). In our analysis, traditionalist produc-ers, initially neglected by the audience, seek to maintainschematic divides to express a collective identity. Theiraction intensifies when the fuzziness of their categoryschema and the risk of being submerged by an alterna-tive schema are greater but weakens when the schemaassumptions are revised.

In another area of study that applies to industriesand markets, the notion of institutional logics—the setof assumptions, values, beliefs, rules, and schematathat provide meaning to material practices in society(Friedland and Alford 1991)—researchers have explored

that competing meaning systems replace, clash with,or coexist with one another (Stark 1996, Haveman andRao 1997, Thornton and Ocasio 1999, Rao et al. 2003,Rao and Giorgi 2006, Schneiberg et al. 2008). In thisperspective, groups of agents identify with and usethese logics purposively. Studying the merger activi-ties and career mobility that transformed Lloyds Bankfrom a regional organization into a national bank, Stoveland Savage (2006) argue that bank directors favoredmobility of newly hired employees rather than lateraltransfers that violated the principle of local banking sup-ported by workers of merged banks. Only after a mergerwave did the new logic become accepted and lateraltransfers spread. Bank directors seemed mindful thatconfusing boundaries by changing the identities of insid-ers could generate organizational tensions and challengethe integration process. The initial defense of a collectiveidentity was as crucial for local workers as it was forour traditionalist producers, and disputes could be partlysettled by durable spanning of category boundaries.

The dispute over label interpretations also highlightsissues of authenticity associated with category member-ship. In the context studied here, it is tempting to por-tray modernists as inauthentic and driven by money andtraditionalists as driven by love. In this view, profit-maximizing modernists seek high prices, and utility-maximizing traditionalists consume tradition and sochoose lower prices (Scott Morton and Podolny 2002).Such a portrait treats individual producers as atomisticdecision makers and overlooks the collective sources ofidentity and motivation. Modernists themselves had tomobilize collectively to construct the “modernist” cat-egory and to gain supporters from new entrants. Tra-ditionalists argue that barrique-aged Italian wines areno different from Californian wines and are, therefore,inauthentic reflections of the terroir. The activation oftheir identity involved technical and ideological claimsand resulted in adherence to conventions even in thepresence of evident gains from defection. Some pro-ducers persisted with traditional practices because theydefined themselves in terms of the land and its ter-roir, and they believed that the only authentic way tomake Barolo wine was to use botti. In this sense, weakpragmatic justifications make traditions more meaning-ful (Hobsbawm 1983).

Instead of a struggle between authentic and inauthen-tic styles, the case we analyze presents a compellingexample of competing views of authenticity based ondiffering interpretations of categorical schemas. On onehand, the modernist interpretation of winemaking consti-tutes a different claim of authenticity, one that embracescreativity. Breaking with traditional practice and adopt-ing barriques, among other technologies, was motivatedby a desire to make original (and technically excel-lent) wines. On the other hand, “traditional” producers

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and DefectionOrganization Science 22(6), pp. 1449–1463, © 2011 INFORMS 1461

espouse a notion of authenticity as conformity to con-ventions and “natural” processes (Peterson 1997, Rozin2005). The latter group reacted collectively when itfaced the risk of a loss of distinctiveness of categoryboundaries because of straddling, particularly by thecontradictory position of producers that defected to mod-ernism but maintained a stronger perceived resemblanceto tradition.

Thus, our study makes a broader contribution toresearch on authenticity and illustrates that authentic-ity is intrinsically a polemical concept (Trilling 1972,Spooner 1986). We suggest that one interpretation ofthe category (the “modernist”) embodies authenticityby intervention that leads to originality in the maininterest of gaining good outcomes in the market. Theother (the “traditionalist”) seeks authenticity throughrule-based abstention that leads to genuineness. Theseopposed meanings evoke the distinction between typeauthenticity (conformity to a type) and moral authentic-ity (sincerity of commitment to a set of values) proposedby Carroll and Wheaton (2009). Our study suggests thatthese types are intertwined. On one hand, the concernsof “traditionalist” producers were not only premised onstylistic conventions but were also permeated by a moralview of what it means to be a category member. Onthe other, “modernist” vintners sought to find their owncreative voice by challenging the style of establishedconventions.

AcknowledgmentsThe authors are grateful to Christophe Boone, Glenn Carroll,Greta Hsu, Özgecan Koçak, Ming Leung, Susan Olzak,Fabrizio Perretti, László Pólos, Peter Roberts, and semi-nar participants at Stanford University, Durham University,the Massachusetts Institute of Technology, University ofWisconsin–Madison, Sabancı University, University of Cali-fornia at Davis, and the members of the Nagymaros Group forcomments. The Stanford Graduate School of Business, EntERBocconi, and the Durham Business School provided finan-cial support. Agnese Orlandi and Elena Perondi provided ableresearch assistance. Finally, the authors thank their informantsfor their willingness to spend time with them and for allowingthem to quote informants by name.

Appendix. Data on Methods of ProductionWe relied on the Veronelli guide because it provides the mostcomprehensive data on Italian wines and covers more wineriesand a much longer period of time than alternative data sourceslike Gambero Rosso’s Italian Wines, Parker’s Wine Advocate,or the Wine Spectator. The Veronelli guide has reviewed morethan 10,000 “good Italian wines” by 2,000 “good winemakers”every year since 1991, and it is considered an authoritativesource on wine in Italy. Its founding editor, Luigi Veronelli,was internationally acknowledged as Italy’s most celebratedwine and food critic. The Veronelli guide enjoys wide cir-culation among expert and nonexpert audiences, includingwine merchants, restaurant sommeliers, and consumers seek-ing direction in their purchase decisions. The annual list of

outstanding wines has achieved high prominence inside andoutside the industry. For example, the release of a new editionof ratings receives broad coverage in the media and is featuredin every major Italian national newspaper.

The guide provides information organized by producer andgeographic location. Three key product details are displayed:the type of barrel used to age the various vintages of eachwine, the retail price category, and a critical evaluation. Theevaluation follows a tasting done by Luigi Veronelli and histhree coeditors; all of them are wine experts. (After Veronelli’sdeath in 2004, the coeditors continued the operation.) Theguide rates each wine employing a single-blind tasting method.In fact, this data source is regarded as reliable and impartial.

The observation period for our data corresponds to the vin-tages covered by the guide during its first 15 editions, begin-ning in 1991 and ending in 2007. The Italian law regulatingwine appellation systems requires that the products age for aperiod variable from at least two (Barbaresco) to at least three(Barolo) years prior to their release on the market and, con-sequently, to their review by the critics. Such a rule imposesa systematic time lag on the structure of our data. For exam-ple, the 2007 edition reviews wines from the vintages of 2001,2002, and 2003 as the latest available for Barbaresco and 2001and 2002 for Barolo. Yet the coverage of 2002 is incompletebecause reserve labels are released after five years of aging. Toallow for full comparability across the producers, we end ourobservation window in 2001. All our interviews revealed thatexperimentation with cellar techniques was negligible beforethe 1980s, and we start our analysis with the vintage of 1980(restricting the initial date of observation avoids inflation inthe estimated effects of the use of traditional techniques). Ourmost inclusive sample consists of all producers of Barbarescosand Barolos and their respective labels, which have receiveda rating in the Veronelli guide. The risk set included in theanalysis is the subsample of traditional producers who age alltheir wines in large casks (botti). The subsample is composedof 192 different producers and 1,140 producer–vintage spells.

Endnotes1Throughout, we distinguish what linguists call the object lan-guage (in this case, the language of the producers and audi-ences we analyze) from the so-called metalanguage or analyticlanguage of the authors by placing terms in the object lan-guage in quotes.2This quote and those that follow come from our interviewswith winemakers and wine critics.3Negro et al. (2010) employ a similar approach to conceptual-ize and measure GoM and category fuzziness. They consider“traditionalist” and “modernist” as the candidate interpreta-tions of Barolo and Barbaresco winemaking. However, theynote that some audience members, notably the critics, iden-tify three possible styles in which each wine label can beproduced—“traditionalist,” “modernist,” and a hybrid labeled“international” that mixes botti and barriques—instead of two.Our study focuses on modeling defection rates from the “tra-ditional” category and is concerned with the categorization asseen by the producers. For “traditional” producers, the crucialmarker is the sole use of botti, so they tend to have a moresimplified conception of the field in which only two categoriesoperate. Consequently, we follow this account in our analysis.

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and Defection1462 Organization Science 22(6), pp. 1449–1463, © 2011 INFORMS

4We use the quality ratings assigned by the Wine Advocatemagazine to wines produced in the Piedmont in each vin-tage (http://www.erobertparker.com). The ratings are expressedusing a 100-point scale ranging from 50 (appalling vintage) to100 (extraordinary vintage).5We examined the issue of heterogeneity within the “tradition-alist” category with available data in another (indirect) way. Inour telephone survey, we asked producers if they used prac-tices that can be viewed broadly as nontraditional, such as theuse of rotofermentor machines (spinning, horizontal tanks thathelp mixing the juice and skins when the layer of grape skins(cap) rises to the surface of red wine during fermentation).We calculated a dummy for producers that had ever adoptedrotofermentor machines as 1 and 0 otherwise. In unreportedanalyses we added this variable to the specification of Model 2and found that (i) it has a positive (but statistically nonsignif-icant) effect on the rate of defection from tradition, (ii) ourother findings are unaffected, and (iii) the model fit does notimprove significantly.6Of the 60 producers, 43 began as “traditionalists,” 32 ofwhich never defected; 11 did. Seventeen producers started as“modernists.”

ReferencesArmstrong, E. A. 2002. Forging Gay Identities: Organizing Sexu-

ality in San Francisco, 1950–1994. Chicago University Press,Chicago.

Associazione Vignaioli Piemontesi. 2000. Atlante delle Etichette delBarolo. Atlante delle Etichette del Barbaresco. AssociazioneVignaioli Piemontesi, Mondoví, Italy.

Becker, H. S. 1982. Art Worlds. University of California Press,Berkeley.

Bernstein, M. 1997. Celebration and suppression: The strategic usesof identity by the lesbian and gay movement. Amer. J. Sociol.103(3) 531–565.

Bernstein, M. 2005. Identity politics. Annual Rev. Sociol. 31 47–74.

Caldano, G., A. Rossi. 2004. Codice Denominazione di Origine deiVini. Unione Italiana Vini, Milano, Italy.

Carroll, G. R., A. Swaminathan. 2000. Why the microbrewery move-ment? Organizational dynamics of resource partitioning in theU.S. brewing industry. Amer. J. Sociol. 106(3) 715–762.

Carroll, G. R., D. R. Wheaton. 2009. The organizational construc-tion of authenticity: An examination of contemporary food anddining in the U.S. Res. Organ. Behav. 29 255–282.

Cattani, G., S. Ferriani, G. Negro, F. Perretti. 2008. The structureof consensus: Network ties, legitimation, and exit rates of U.S.feature film producer organizations. Admin. Sci. Quart. 53(1)145–182.

Crane, D. 1987. The Transformation of the Avant-Garde. TheNew York Art World, 1940–1985. University of Chicago Press,Chicago.

Esposito, S. 2008. Passion on the Vine: A Memoir of Food, Wine andFamily in the Heart of Italy. Broadway Books, New York.

Fine, G. A. 2004. Everyday Genius: Self-Taught Art and the Cultureof Authenticity. University of Chicago Press, Chicago.

Friedland, R., R. Alford, 1991. Bringing society back in: Sym-bols, practices, and institutional contradictions. W. W. Powell,P. J. DiMaggio, eds. The New Institutionalism in OrganizationalAnalysis. Chicago University Press, Chicago, 232–266.

Goffman, E. 1963. Stigma: Notes on the Management of Spoiled Iden-tity. Prentice Hall, Englewood Cliffs, NJ.

Grazian, D. B. 2003. Blue Chicago: The Search for Authenticity inUrban Blues Clubs. University of Chicago Press, Chicago.

Griswold, W. 1987. The fabrication of meaning: Literary interpreta-tion in the United States, Great Britain, and the West Indies.Amer. J. Sociol. 92(5) 1077–1117.

Hannan, M. T. 2010. Partiality of memberships in categories and audi-ences. Annual Rev. Sociol. 36 159–181.

Hannan, M. T., L. Pólos, G. R. Carroll. 2007. Logics of OrganizationTheory: Audiences, Codes, and Ecologies. Princeton UniversityPress, Princeton, NJ.

Haveman, H. A., H. Rao. 1997. Structuring a theory of moral senti-ments: Institutional and organizational coevolution in the earlythrift industry. Amer. J. Sociol. 102(6) 1606–1651.

Hobsbawm, E. 1983. Introduction: Inventing traditions. E. Hobs-bawm, T. Ranger, eds. The Invention of Tradition. CambridgeUniversity Press, Cambridge, UK, 1–15.

Hsu, G. 2006. Jacks of all trades and masters of none: Audiences’reactions to spanning genres in feature film production. Admin.Sci. Quart. 51(3) 420–450.

Hsu, G., M. T. Hannan, Ö. Koçak. 2009. Multiple category member-ships in markets: An integrated theory and two empirical tests.Amer. Sociol. Rev. 74(1) 150–169.

Hsu, G., M. T. Hannan, L. Pólos. 2011. Typecasting, legitimation, andform emergence: A formal theory. Sociol. Theory 29(2) 97–123.

Hsu, G., G. Negro, F. Perretti. 2007. Same, yet different: Similarity,distinctiveness and the evolution of film genres. Acad. Manage-ment Best Papers Proc., Academy of Management, BriarcliffManor, NY.

I Vini di Veronelli. 1991–2005. Veronelli Editore, Bergamo.

Kuilman, J. G., J. Li. 2009. Grades of membership and legitimacyspillovers: Foreign banks in Shanghai, 1847–1935. Acad. Man-agement J. 52(2) 229–245.

Negro, G., M. T. Hannan, H. Rao. 2010. Categorical contrast andaudience appeal: Niche width and critical success in winemak-ing. Indust. Corporate Change 19(5) 1394–1425.

Negro, G., M. T. Hannan, H. Rao, M. D. Leung. 2007. No barrique,no Berlusconi: Collective identity, contention, and authenticityin the making of Barolo and Barbaresco wines. Research Paper1972, Graduate School of Business, Stanford University, Stan-ford, CA.

Peterson, R. A. 1997. Creating Country Music: Fabricating Authen-ticity. University of Chicago Press, Chicago.

Rao, H., S. Giorgi. 2006. Code breaking: How entrepreneurs exploitcultural logics to generate institutional change. Res. Organ.Behav. 27 269–304.

Rao, H., P. Monin, R. Durand. 2003. Institutional change in ToqueVille: Nouvelle cuisine as an identity movement in French gas-tronomy. Amer. J. Sociol. 108(4) 795–843.

Rao, H., P. Monin, R. Durand. 2005. Border crossing: Bricolageand the erosion of categorical boundaries in French gastronomy.Amer. Sociol. Rev. 70(6) 968–991.

Rao, H., C. Morrill, M. N. Zald. 2000. Power plays: Social move-ments, collective action, and new organizational forms. Res.Organ. Behav. 22 239–281.

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.

Negro, Hannan, and Rao: Category Reinterpretation and DefectionOrganization Science 22(6), pp. 1449–1463, © 2011 INFORMS 1463

Rozin, P. 2005. The meaning of “natural”: Process more importantthan content. Psych. Sci. 16(8) 652–658.

Ruef, M., K. Patterson. 2009. Credit and classification: The impact ofindustry boundaries in 19th-century America. Admin. Sci. Quart.54(3) 486–520.

Schneiberg, M., M. King, T. Smith. 2008. Social movements and orga-nizational form: Cooperative alternatives to corporations in theAmerican insurance, dairy, and grain industries. Amer. Sociol.Rev. 73(4) 635–667.

Scott Morton, F. M., J. M. Podolny. 2002. Love or money? The effectsof owner motivation in the California wine industry. J. Indust.Econom. 50(4) 431–456.

Spooner, B. 1986. Weavers and dealers: The authenticity of an orien-tal carpet. A. Appadurai, ed. The Social Life of Things: Com-modities in Cultural Perspective. Cambridge University Press,Cambridge, UK, 195–235.

Stark, D. 1996. Recombinant property in East European capitalism.Amer. J. Sociol. 101(4) 993–1027.

Stovel, K., M. Savage. 2006. Mergers and mobility: Organizationalgrowth and the origins of career migration at Lloyds Bank.Amer. J. Sociol. 111(4) 1080–1121.

Thornton, P. H., W. Ocasio. 1999. Institutional logics and the histori-cal contingency of power in organizations: Executive successionin the higher education publishing industry, 1958–1990. Amer.J. Sociol. 105(3) 801–843.

Trilling, L. 1972. Sincerity and Authenticity. Harvard UniversityPress, Cambridge, MA.

White, H. C. 1981. Where do markets come from? Amer. J. Sociol.87(3) 517–547.

Zuckerman, E. W. 1999. The categorical imperative: Securities ana-lysts and the legitimacy discount. Amer. J. Sociol. 104(5)1398–1438.

Zuckerman, E. W., T.-Y. Kim. 2003. The critical trade-off: Identityassignment and box-office success in the feature film industry.Indust. Corporate Change 12(1) 27–67.

Giacomo Negro is an associate professor of organizationand management at the Goizueta Business School, Emory Uni-versity. His research interests include the ecology of organiza-tions, categories, and identities in markets.

Michael T. Hannan is the Stratacom Professor of Man-agement at the Stanford Graduate School of Business andProfessor of Sociology at Stanford University. His researchconcentrates on the ecology of organizations and on the devel-opment of formal theories of sociological processes.

Hayagreeva Rao is the Atholl McBean Professor of Orga-nizational Behavior at the Graduate School of Business, Stan-ford University. He studies the social and cultural causes oforganizational change.

INFORMS

holds

copyrightto

this

article

and

distrib

uted

this

copy

asa

courtesy

tothe

author(s).

Add

ition

alinform

ation,

includ

ingrig

htsan

dpe

rmission

policies,

isav

ailableat

http://journa

ls.in

form

s.org/.