Research Policy 35 (2006) 1273–1290

Convergent designs in fine fashion: An evolutionary

model for stylistic innovation
Rossella Cappetta a,∗ , Paola Cillo b,1 , Anna Ponti a,2
a Institute of Organization and Information Systems, Bocconi University, SDA Bocconi School of Management,
Viale Isonzo, 23, 20135 Milan, Italy
b Institute of Management and Business Economics, Bocconi University, SDA Bocconi School of Management,

Via Bocconi, 8, 20136 Milan, Italy

Received 1 April 2005; received in revised form 1 November 2005; accepted 1 February 2006
Available online 20 September 2006

Abstract
In this paper we propose and test an evolutionary model of stylistic innovation, that is the change in the aesthetic and symbolic
elements of products and services. Style is an increasingly relevant phenomenon in many industries and our goal is to develop a
model of how styles get introduced and adopted. We build on existent literature on technological change to frame the dynamics
of how style changes across time. In the model we propose, companies face periods of ferment with relevant stylistic variations
and periods of convergence on a single trend—that is the stylistic convergent design versus the technologic dominant design. We
identify the drivers of convergence in stylistic innovation. We test the model with a longitudinal empirical study (1984–2002) in the
fine fashion industry, explaining the main differences between a stylistic and a technological evolutionary model.
© 2006 Elsevier B.V. All rights reserved.

Keywords: Evolutionary change; Innovation; Stylistic innovation; Symbolic industries

1. Introduction over the others (Abrahamson and Rosenkopf, 1997;

Adner and Levinthal, 2001; Anderson and Tushman,
An abundance of innovations are constantly strug- 1990; Christensen and Bower, 1996; Tegarden et al.,
gling for market acceptance in different industrial con- 1999; Tripsas, 1997a,b). Frequently, though, the focus
texts. As a consequence, a large amount of researchers’ of these studies is on building and testing new models
efforts has been devoted to the understanding of on the emergence and affirmation of a new design in a
the sequencing of different designs (Abernathy and technology-driven context. Indeed, many of these studies
Utterback, 1978; Tushman and Anderson, 1986) and of have been inspired by the seminal works by Utterback
the factors affecting the dominance of one specific design and Abernathy (1975) and by Tushman and Anderson
(1986). The researches following these models have been
devoted to different industries and have been investi-
∗ Corresponding author. Tel.: +39 02 5836 2630/2632; gating different variables related to this issue—e.g., the
fax: +39 02 5836 2634. performance of the companies introducing the innova-
E-mail addresses: [email protected] (R. Cappetta), tion (Tegarden et al., 1999), the organizational impact
[email protected] (P. Cillo), [email protected]
(A. Ponti).
of the change in the dominant design (Anderson and
1 Tel.: +39 02 5836 6844; fax: +39 02 5836 6888. Tushman, 1990; Henderson and Clark, 1990), the tech-
2 Tel.: +39 02 5836 2663/2633; fax: +39 02 5836 2634. nical superiority of the new standard with respect to the

0048-7333/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.respol.2006.02.009
1274 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290

previous one (Christensen and Bower, 1996), the drivers ical change and how the convergent stylistic design is
of the affirmation of innovations and of dominant designs different from the technological dominant design.
(Adner and Levinthal, 2001). The common thread of We test the conceptual framework on affirmation and
these studies is represented by the fact that they consider decline of stylistic innovation through a longitudinal
technology as a major driver for change, both at an indus- empirical study in the fine fashion industry. Overall, we
try and at an organizational level. Yet, in many contexts – treat the fine fashion industry as a revealing context for
from the automotive to the hotel one, not to mention cre- understanding some innovation processes usually hidden
ative industries, such as fashion and design – a growing in less style-focused industries. The relevance of stylistic
portion of innovation is linked to the aesthetic and sym- innovation is so transparently obvious in the fine fashion
bolic elements of the products or services (Caves, 1999), industry that its analysis may lead one to draw insights
i.e. to their style. As Postrel (2003, p. 5) points out ‘the that go beyond this industry and to develop concepts
issue is not what style is used but rather that style is used, that might be applied to other industries where the role
consciously or conscientiously, even in areas where func- of style is likely to be somewhat more latent.
tion used to stand alone’. The dynamics of affirmation of The reminder of the article is organized as follows.
stylistic innovation have not so far been systematically The next section presents the research framework and the
addressed in the literature. Yet, this field of investiga- hypotheses. Then the method used for the empirical anal-
tion appears as extremely relevant. Indeed, the aesthetic ysis is presented, with a focus also on the characteristics
and symbolic elements are becoming more and more of the industrial setting chosen for the study, that is the
important in different contexts, even in those settings tra- fine fashion context. Finally, findings and implications
ditionally driven by technological standards. The mobile are discussed with specific regard to how companies
phone industry or the computer one are just two examples in different settings might cope with the complexity of
of settings where the affirmation of a new standard might stylistic innovation.
be explained also by considering the stylistic elements
of the products: the iMac produced by Apple in the com- 2. Conceptual development and research
puter industry and the innovation efforts by Nokia in the hypotheses
mobile industry highlight for instance the importance of
style in the innovation process and in the design affirma- Consumers increasingly make brand choices on the
tion. The implication of this is that the factors explaining basis of the aesthetic and symbolic value of products and
the emergence of a technological standard may not tell services (Bloch et al., 2003; Dumaine, 1991; Schmitt
the whole story. and Simonson, 1997). Companies, on the other side,
The purpose of the paper is to draw and test a con- invest large amount of resources in developing new
ceptual framework for the affirmation and evolution of designs for the products and these designs have not as a
stylistic innovation in industries where the main driver primary goal to make the products more functional but
of these innovative processes is not so much technology more fashionable (Pesendorfer, 1995). Postrel (2003,
development as social changes. To achieve this goal we 2001) shows that the “look and feel” of people, places,
design a research framework based on the cyclical model and things, are more important than we think. This
of technological change developed in high-tech contexts means that the aesthetic and symbolic dimensions of a
(Anderson and Tushman, 1990). We formulate specific product – that is its style – is increasingly relevant in
hypotheses on the factors affecting the affirmation and many differentiated industries in order for a company
adoption of stylistic innovation and we propose an evolu- to be successful, but “style is qualitative . . . it is hard to
tionary model of stylistic innovation in which companies be assessed” (Postrel, 2001, p. 1).
face periods of ferment with relevant variations in the
styles they propose and periods of convergence on a 2.1. The concept of stylistic innovation
fashion trend. Notwithstanding the relevance of a sin-
gle company’s ability in the definition of the style and In order to introduce the concept of stylistic inno-
the needs of differentiation on the customers’ side, we vation, we start from defining the concept of style, by
argue that for style – like for technology – convergence adopting the metaphor introduced by leading sociolo-
on a single standard matters, and that different styles are gists according to whom style is a code composed of
cyclically affected by a reference trend, that we define elements that join together like letters of the alphabet
convergent stylistic design. Furthermore, we intend to and are mixed up and thus give life to umpteen combi-
show how the stylistic dimension affects some relevant nations (Barthes, 1983; Eco, 1976; Simon-Miller, 1985).
aspects of the traditional cyclical model of technolog- As Davis (1992) pointed out: ‘style is a code, however,
 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290 1275

radically dissimilar from those used in cryptography; a product generating both a new product – from a phys-
(. . .) it is of ‘low semanticity’. Perhaps it can best be ical point of view – and a new meaning.
viewed as an incipient or quasi-code, which, although it In contrast to technological innovation driven exclu-
must necessarily draw on the conventional visual and tac- sively by the addition or the alteration of tangible fea-
tile symbols of a culture, does so allusively, ambiguously, tures (Abernathy and Clark, 1985), stylistic innovation
and inchoately, so that meanings evoked by the combi- is mainly a change in the social meaning that is assigned.
nation and the permutations of the code’s key terms are In the case of stylistic innovation, the physical form of
forever shifting or in process” (Davis, 1992, p. 5). That the product can change – and it generally changes in
is, the code’s key terms regard the aesthetics following its aesthetic characteristics – but this is not the relevant
conventional categories (colors, forms); the final com- point: what is important, indeed, is the creation of dif-
bination of these key terms has an aesthetic value, but ferent intangible meanings.
it can assume different and not completely predictable Given the main focus on meanings, stylistic innova-
symbolic meanings. The aesthetic of a product or ser- tion is exposed to a phenomenon of ‘revival’: old styles
vice is the sensory experience it elicits; it is related can come back (e.g. the hippie style diffused in the 1970s
to those characteristics that create a product/service’s came back in 2000) when the meaning associated to it
appearance, such as materials, proportion, color orna- is again ‘in’ and relevant. Indeed, differently from the
mentation, shape, and size (Lawson, 1983; Rafaeli and technical functionality, heavily affected by technologi-
Vilnai-Yavetz, 2005). The symbolism regards the mean- cal change, meanings can maintain their strength over
ings or associations the product/service elicits (Strati, time (Davis, 1992).
1992). Product/service aesthetics frequently carry sym- Furthermore, technological innovation, by affecting
bolic meanings: elegance, ease of use, youthfulness, the physical and tangible features of the product and
and durability, may stem from aesthetic choices (Forty, its function, exists independently from a process of
1986). For these reasons, we define style as both an aes- communication; on the contrary, stylistic innovation
thetic and a symbolic choice a company makes regarding exists just if it is perceived and used by a specific social
the products/services, their main features, and how they community. Indeed, the new meaning assigned to an
are combined. existing or to a partially modified product exists just if a
Style is a ‘temporal’ phenomenon that has a short certain amount of people adopts this language to signal
life span; it is an “obligatory transformation of taste” its belongingness to a specific reference group—that
(Simmel, 1985). As a phenomenon, it is characterized is a secondary diffusion process in Hirschman’s terms.
by a provisory life span: every style is temporary and The explanatory power of this language relies on the
inevitably destined to be abandoned by reason of its intensity of the communication activity enacted by the
own popularity and diffusion (Robinson, 1961). Simmel companies inside the industry.
(1985) sustained that style was what he called bewe- For this vital need of social perception and social
gungsform, that is, something created from a movement awareness, one of the most notable role in the indus-
that later intensifies it. For this reason, innovation in tries driven by stylistic innovation is played by the press,
style is an imperative for a company that decides to care able to sustain this process of reassignment of a new
of the aesthetic and symbolic dimensions of its products meaning to an existing product or of attribution of a
and services. fascinating meaning to a product with a new aesthetic
Given the aesthetic and symbolic value of style, the form (Breward, 2003). When the product has a stylis-
stylistic innovation can be assimilated to the concept tic component, the press serves to emphasize the sym-
of symbolic innovation developed by Hirschman (1982, bolic value and to de-emphasize the materiality of the
1986; Hounshell, 1984), that is an innovation not rele- products. When companies compete on products with
vantly linked to technological changes, but to symbolic a relevant stylistic component, the press has a crucial
changes. Following Hirschman, symbolic innovations role as arbiter and/or witness of the success: they act as
are those that “result from the reassignment of social “gatekeepers” of these industries (Solomon and Rabolt,
meaning to an existing product, generating a secondary 2004)1 .
diffusion for it among those identifying with the rele-
vant reference group” (1982, p. 537). Our concept of
stylistic innovation is broader than Hirschman’s sym- 1 The status of magazines has been continuously strengthened by the
bolic innovation one: a stylistic innovation results from growing importance of stylish photography and by the emphasis on the
the reassignment of social meaning to an existing product idea that producing new aesthetic forms or meanings is very close to
and/or from the change of the aesthetic characteristics of art (Jobling and Crowley, 1996).
1276 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290

2.2. Convergence in stylistic innovation and/or aesthetic elements that mostly regard intangible
features of the product. It is the property of intangibility
Despite the growing relevance of style in different to make not relevant the technical compatibility issue. In
industrial settings, up to now no specific managerial this paper we argue that the irrelevance of technical com-
models have been developed for stylistic innovation. On patibility does not enact differentiating behaviors—both
the contrary, the literature on innovation in high-tech from customers and producers. Indeed, we sustain that
industries has developed conceptual models useful in the technical compatibility explanation is substituted by
supporting organizational decisions on the edge of inno- both a social and aesthetic compatibility and that con-
vativeness (e.g., Eisenhardt and Tabrizi, 1995). Many vergence emerges for stylistic innovation as in the case
research studies have attempted to develop and corrob- of technological innovation.
orate models, which can predict the success and decline Social compatibility is the first driver of convergence
of design and technological standards. One of the mod- in stylistic innovation. Customers look for differentiation
els presented received consistent empirical support: the and distinction but also for recognizability as élite. Being
cyclical model of technological change developed by an élite means following the élite’s habits, behaviors,
Anderson and Tushman (1990). In this model, tech- ways of dressing, that in order to be recognizable have to
nological innovation goes through different phases. A be similar and share the same characteristics (Bourdieu,
technological breakthrough initiates an era of techni- 1979; Barthes, 1982). In this perspective, style becomes
cal variation and selection—the era of ferment. In this a distinctive expression of a privileged and exclusive
period, before one technological configuration prevails social position: “The logic behind the production of high
over the others, there are a number of different designs on society goods and the strategy of distinction that is at
the market, which are all very different in terms of tech- the origin of their dynamics dictate that their use (i.e.
nical characteristics and functional performance. The fashion or literature, opera, etc.) renders the consumer
era of ferment is characterized by strong competition more exclusive by serving as instruments of distinction”
between different design processes and ends when one (Bourdieu, 1979, pp. 233–234). From a consumer’s point
design prevails over the others. This phase is followed of view, the desire for stylistic innovation is explained
by another one defined as a period of incremental change by the desire to own the ‘right’ products, but also by the
in which the reference design is gradually modified and desire “to interact with the ‘right’ people” (Pesendorfer,
comes to a close, thus giving rise to a new period of fer- 1995, p. 772). For this reason, when a style is created
ment when the dominant design is modified to such an it addresses a ‘cool’ and elitist community.2 Yet, partic-
extent that it no longer represents a reference point that ularly when a style succeeds, it creates messages that
most manufactures interpret. acquire meanings for a larger part of the community and
In the case of technological innovation, the major rea- that communicate who is in the right group to the mem-
son behind convergence is represented by the technical bers of this group but also the others. As Pesendorfer
compatibility issue that is particularly relevant whether (1995) remarks style is used as “a signalling device in a
there is a quite complex system of products (David and dating game (. . .). Customers want to signal their own
Greenstein, 1990). As Garud and Kumaraswamy (1995, quality to other customers because they are involved in
p. 94) note: “compatibility is a relational attribute that a matching game in which each person would like to
defines rules of fit and interaction across boundaries match up with a high-quality person rather than a low-
called interfaces”. In particular, technical compatibility quality person” (Pesendorfer, 1995, pp. 771–772). That
is the attribute of a system of components that have to is, a recognizable stylistic trend is necessary; each com-
work together: they are compatible from a technical point pany’s style has to have some convergent characteristics
of view if their tangible interfaces fit together. In order in order to be used as a language to communicate in a
to work properly, the system needs a perfect integration specific community and to provide people of the commu-
of standard components, responding to the same techno- nity with a sense of belongingness. If customers require
logical pattern (Suarez and Utterback, 1995). The adop- a common language to show their belongingness to an
tion of the same technological standard produces posi-
tive network externalities for those companies adopting
2 The elitist aspiration is common also to those styles that are inspired
the same standard (Tushman and O’Reilly, 1997) and
enhances the switching costs – both for customers and by people in the street (e.g. grunge); when they become a codified
style adopted by a company their characteristics are sophisticated, their
producers – to a different design. original meaning is transformed and an elitist target is identified (see,
The technical compatibility issue does not occur for for example, the story of the Versace’s sophistication of the hip-hop
stylistic innovation, because it works on social meaning style in Solomon and Rabolt, 2004).
 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290 1277

élite (but also to perceive the exclusion from an élite and In sum, given these considerations and using the
the desire of belongingness), companies need to reduce Garud and Kumaraswamy’s definition of compatibility
the potential variety of their styles. That is, styles con- (1995), we define social compatibility as the attribute
verge in the meanings they carry out in order to answer expressing the coherence of a system of social meanings
to a social need of convergence on a common language. used in the same social context. In this case, the different
This convergence is supported by the specific diffu- components are compatible from a social point of view if
sion of the style on the demand side. As we remarked, a their meanings fit together. We state that companies, on
style is created to address an élite, but the elitist aspira- a side, and consumers, on the other side, look for a com-
tion does not mean that the stylistic product is exclusively mon and coherent set of meanings and this phenomenon
sold to the élite during its entire life cycle. Indeed, fol- enacts a process of stylistic convergence.
lowing Pesendorfer (1995), we sustain that the demand The second driver of convergence in stylistic innova-
for style over time displays both the snob and the band- tion is linked to aesthetic compatibility. The root of this
wagon effects.3 If the ‘right’ people purchase a style concept is in the idea of visual fit of different products
(intentionally dedicated exclusively to them), it will be influencing the consumers’ aesthetic response, devel-
more valuable to all consumers, while if the mass of oped in the consumer behavior’s literature. Hutchinson
people purchases the design, it will be less valuable to (1998, p. 375), working on a system of products, intro-
all consumers. That is, “Since high types purchase the duce the concept of “visual display that connects its parts
design commodity first, there is a bandwagon effect if in a meaningful way” and that “involves some form of
few consumers buy the design, while there is a snob matching with respect to a salient visual characteristic”.
effect if many consumers buy the design” (Pesendorfer, To give an example, in the case of the choice of differ-
1995, p. 773). ent objects in a kitchen the aesthetic coherence means
These two different effects on the demand side are that a person will prefer elements (microwave, toaster,
strictly linked to the convergent/divergent sequence on scale, etc.) curve-shape, reflecting, and aluminum made,
the supply side (Wade, 1995). Specifically, the conver- in order to have a ‘high-tech’ style kitchen. In an envi-
gence of the companies during the periods of incremental ronment like this one, a translucent, red, plastic and
change is facilitated by the presence of an increasing squared toaster is not aesthetically coherent, it is not ‘in
number of ‘right’ buyers because of the bandwagon style’. More in general, there is a search for an aesthetic
effect. The strength of the bandwagon effect supports the compatibility behind the fact that the most part of the
strength of the convergent style and affects the duration household-electrics of last generation are curve-shape,
of the periods of incremental change. As far as the new light color and made in reflecting materials. Following
style becomes popular, it will be imitated by mass mar- these considerations, we define aesthetic compatibility
ket companies and sold on the mass market. Whenever as the attribute expressing a visual coherence in the aes-
the convergent style spreads too much, the snob effect thetic appearance of the elements located or used in the
will prevail and this effect will drive the decline of the same context. We state that companies, on a side, and
convergent style. The strength of the snob effect affects consumers, on the other side, look for a common and
the length of the period of incremental change and the coherent set of aesthetic characteristics and this phe-
beginning of the period of ferment.4 nomenon is the other force behind the process of stylistic
convergence.
Our framework builds on the two concepts introduced
3 The bandwagon effect represents the phenomenon of increase in so far – the symbolic and the aesthetic compatibility
the demand of a product if others are consuming it. This phenomenon – and emphasizes the relevance of convergence on a
takes account of “the desire of people to wear, buy, do, consume, and
behave like their fellows; they desire to join the crowd, be ‘one of
the boys’, etc.” (Leibenstein, 1950, p. 184). The snob effect represents
the opposite phenomenon where the demand of a product is decreased with cheaper materials. The perfect example of those stylistic imita-
because others are consuming it. This phenomenon takes into account tive behaviors is Zara, the Spanish fashion retail. Imitative behaviors
“the search for exclusiveness by individuals through the purchase of are facilitated by the absence of any kind of property rights on style.
distinctive clothing, foods, automobiles, houses, or anything else that However, complexity of the meaning associated to the style can play
individuals may believe will in some way set them off from the mass a crucial role. A style carrying out an ‘easy’ meaning is more likely
of mankind” (Leibenstein, 1950, p. 184). to face a snob effect (e.g. in our data this could be the case for the
4 The speed of the diffusion is affected by the presence of mass ‘animalier’ style with its message of aggressiveness); on the contrary a
market competitors. Those competitors imitate the style with a short style carrying out a complex meaning will spread less rapidly out of the
temporal lag and sell a simplified version of the style—that carries elitist group (e.g. in our data this could be the case for the ‘minimalist’
out easier meanings and is less expensive in price, also because done style).
1278 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290

specific stylistic innovation. In sum, we argue that, like With respect to the convergent style new entrants do
for technological innovations (Anderson and Tushman, not share the same resource endowment and the same
1990), it is possible to identify an evolutionary model of knowledge of incumbents and the introduction of inno-
stylistic innovation in which companies face periods of vations on the convergent style would be much more
differentiation with relevant variations in the styles they costly and risky for them than for incumbents.
propose (periods of ferment) and periods of convergence In sum, we state that incumbents are facilitated in
on a prevalent style (periods of incremental change). innovations in presence of a convergent style.

Hypothesis 1. Stylistic innovation follows an evolu- Hypothesis 2a. During the period of incremental
tionary model characterized by the sequence of periods change incumbents are more innovative than new
of ferment and periods of incremental change. entrants.

On the contrary, if a reference style has not affirmed

In contrast with technology-based industries (Bijker
yet new entrants can deploy at best their ‘newness’ by
et al., 1987; David, 1987; Hounshell, 1984; Hughes,
creating a completely new style. In comparison with
1983), we expect a more consistent number of styles
the incumbents, they are more able to invent something
and a less exclusive and mandatory stylistic trend. Con-
completely new, because of the lack of inertial frames,
sidering that for each company style is also the device
cognitive ‘sunk costs’, and existing routines (Henderson
used to be recognized as a brand, and given customers’
and Clark, 1990). New entrants are often driven by dra-
need for distinction and the emergence of a snob effect,
matic creativity that may blow without any stickiness to
we expect a certain degree of differentiation among com-
the existent style or identity. In addition, these compa-
panies also during the era of incremental change and a
nies do not have any “cannibalization” issue to take into
certain degree of stylistic differentiation for each com-
account. This issue, on the other hand, represents a major
pany in the same period of time. For these reasons, we
point for those companies that have been investing on a
define the stylistic trend as a convergent design, that is
specific style over the years and are therefore less keen
the style that most companies use as a reference point
towards abandoning it (Chandy and Tellis, 1998). In sum,
over a particular period of time, rather than a dominant
we state that new entrants are facilitated in innovations
design (Tegarden et al., 1999). In sum, we do not expect
when a convergent style has not emerged.
a clear phenomenon of dominance, but a certain degree
of convergence. Hypothesis 2b. During the period of ferment new
entrants are more innovative than incumbents.
2.3. Evolution of stylistic innovations over time and
role of the companies Differently from technological innovation, stylistic
innovation exists just if it is perceived and its meaning
Similarly with technology-based industries is used by a specific social community. For this reason,
(Anderson and Tushman, 1990; Henderson and companies’ stylistic success cannot be analyzed without
Clark, 1990; Wade, 1995), we expect companies to considering their results in gaining secondary diffusion
play different roles and to have different opportunities for their styles. That is, when stylistic innovation is the
during the periods of convergence and the periods of variable under investigation, the consideration of the tra-
divergence. ditional concept of company success might drive to mis-
In particular, incumbents tend to have a consolidated leading conclusions. Indeed, high revenues might come
style and they aim at preserving their own style and from a product that is very standard in terms of stylis-
identity built over years, and at remaining coherent to tic content, but exactly for this reason is sold to a large
this style to increase their reputation and recognizability and less sophisticated population. The results gained in
whatever are the styles “in” in the period. For this reason, the secondary diffusion process (Hirschman, 1982) are,
they tend to develop more the ability to innovate incre- instead, the issue to take into account: in the case of low
mentally on a given style than the ability to generate new visibility the process of communication of the innova-
styles radically. tion will not be enacted and the meaning associated to
Furthermore, incumbents rely on a specific knowl- the style will not be appropriated by a social community.
edge asset related to the existing convergent style that This is the evidence of a low stylistic success.
may support them in incrementally innovating on the Given these considerations, incumbents are supported
existing style itself (Chandy and Tellis, 1998; Garcia and in achieving success in stylistic innovation by their legit-
Calantone, 2002; Leonard Barton, 1995). imacy and reputation towards consumers’ community.
 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290 1279

Indeed, with respect to the social community, they have step was a content analysis of a fashion magazine pub-
an established position and they know which language lished between 1984 and 2002. This part of the study has
works at best. In addition, with respect to the other com- been undertaken to test the hypotheses. Explanations of
panies in the market, they have the power to enact a the context of the study, of the variables used, and the
process of cooperation and to create the organizational method are given in the following sections.
support necessary to reach convergence on a specific
style (Wade, 1995). These incumbents’ abilities are par- 3.1. The fine fashion context
ticularly useful in stable context conditions, when a con-
vergent style has emerged, and there is the opportunity Despite the traditional view of fashion as a some-
to exploit the benefits of the bandwagon effect. Indeed, what ephemeral phenomenon, the fashion industry has
the convergence process allows incumbents to reduce been the subject of longstanding research in sociological
the risk associated with their activity, while lowering the literature (Barthes, 1983; Descamps, 1979; Eco, 1976;
likelihood for a new entrant to become visible and to Simmel, 1957; Veblen, 1899). Recently, however, fash-
communicate new and socially significant meanings. ion has also become an interesting field of investigation
in organization and management literature (Cappetta
Hypothesis 3a. During the period of incremental and Gioia, 2005; Djelic and Ainamo, 1999; Richardson,
change incumbents are stylistically more successful than 1996).
new entrants. There is a notable pecking order among fashion indus-
try companies. At the highest level there is “haute cou-
On the contrary, new entrants have not to stick to con-
ture” (“high fashion”), which produces custom-made,
solidated styles and suffer less for the snob effect. For
one-of designs. Next, there is prêt-à-porter (“ready to
these reasons they will be keen to creating very distinc-
wear”), which produces fashion products that the aver-
tive and disruptive styles, that consumers will appreciate
age person would still think of as being quite expensive
for their ability to signal distinction. New entrants’ stylis-
but which nonetheless can be bought off the shelf (Davis,
tic success is related to their ‘out of schemata’ character
1992). Even within the fine fashion segment, there is a
and to their ability in opposing to the stylistic status quo.
pecking order: high (e.g., Armani, Dior, Gucci, Louis
For these reasons, they are more likely to be visible when
Vuitton, which stage the “runway” fashion shows in
a snob effect is enacted and a strict stylistic reference
Milan, Paris, London and New York and are frequently
has disappeared. In these ambiguous and shifting condi-
quoted in the top trade magazines), middle (e.g., Calvin
tions, new entrants represent newness and freshness and
Klein, MaxMara, Banana Republic), and low (e.g., Zara,
incumbents will be more likely associated with the ‘old
Gap, Benetton). Our focus in this article is on the high
regime’. Furthermore, because of their size, incumbents
end of prêt-à-porter or what is usually referred to as the
tend to suffer more for the lack of the bandwagon effect.
“fine fashion” industry. Style is the main product of the
Hypothesis 3b. During the period of ferment new fine fashion industry. It can be defined as an aesthetic
entrants are stylistically more successful than incum- choice regarding the individual pieces of its entire col-
bents. lection (including clothing and accessories), their main
features, and how they are combined. More specifically,
3. Method the fashion style is made up of various types of tex-
tiles and fabrics, weavings, color combinations, mixing
In order to test the cyclical model for stylistic innova- and matching garments, volumes and shapes, silhouettes
tion, we have run an empirical study in the fine fashion (Davis, 1992; Volli, 1990).
industry. We structured our study in two steps. The first In the fine fashion each company wants to be charac-
step was qualitative and aimed at defining the period of terized by a distinctive and recognizable style over time,
time more suitable for our analysis, and the definition of but at the same time it has to consider the trends that
style to be used to classify the events of the study. For affect the whole industry for specific periods. For exam-
these purposes, we run semi-structured interviews to 12 ple, the trend of the early 1990s was “minimalism”, a
experts of the fine fashion industry, coming from dif- style (initially associated with the company Prada) char-
ferent countries – e.g., fashion consultants, and fashion acterized by an extremely feminine and soft cut (skirts
editors – and selected according to their reputation in the are more common than pants for women), knee (or
fashion world. An additional evaluation of our results by calf)-length unstructured clothing in black, a few other
these experts was, then, useful in validating the classi- pale colors, no printed patterns and natural and unusual
fication of styles we adopted in our study. The second fabrics.
1280 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290

For a fine fashion company stylistic innovation is not The period of time chosen for the analysis is between
only mandatory – to be renewed coherently with com- 1984 and 2002. Indeed, according to the experts we inter-
pany style and with the fashion trend of the moment – viewed, 1984 marks an important date in the history of
but it also has to happen following specific and recur- fine fashion (starting from 1984, the fashion industry
rent dates: its clothing, accessories, store windows, etc. witnessed a deep renewal in styles in what experts call
have to change at least for the two seasonal shows—the the birth of the prêt-à-porter), and 2002 marks a date that
winter collection shows in September and the summer allows us to analyze a period long enough to comprise
collection shows in February (Robinson, 1961; Simmel, an interesting succession of styles, but not so recent to
1957). register ongoing and still unstable changes. The articles
Given these considerations, we argue that the fine published in 228 issues of Vogue Italia were analyzed in
fashion industry is a revelatory context of investigation order to create a list of styles and companies.6 For each
and, compared to industries such as home furniture and magazine issue, we collected data about the documenta-
consumer electronics, it represents the context with the tion source (month, year, chronological order of the mag-
longest life span where stylistic innovation is the core of azine issues and page reference), the event (that is our
the offering. content unit of analysis—the style precisely mentioned
in the magazine article), the classification of the event
3.2. Sample (based on cut, color, fabric, length and pattern elements),
and the name of the company that the magazine indicates
The population we took into consideration to test our as being the most active for quoted style (taken from the
hypotheses is composed by all the companies cited at photograph captions) (an example is given in Appendix
least once in Vogue Italia magazine between 1984 and A). The magazine pieces extracted from the 228 issues
2002. Vogue Italia was considered the most authoritative were analyzed by the authors who worked independently
magazine for the fashion industry by all the experts con- and were controlled by the experts involved in the qual-
tacted during the preliminary research phase. The use of itative part of the study. Each researcher made a classifi-
Vogue Italia as primary source of data is crucial in order cation of the articles and these classifications were later
to monitor stylistic innovation because it has been con- compared. Agreement on classifications exceeded 85%.
sidered for years as the key player in the reassignment Classifications that were not the same in the three stud-
of the meaning through the secondary diffusion process, ies were subjected to additional analyses. The reference
that is the core characteristic of the stylistic innovation population is composed by 1046 companies mentioned
(Breward, 2003). in Vogue Italia (Appendices B–D).7
The use of this specific magazine brought several
advantages. First, we were able to run a longitudinal
style Vogue would loose its primacy on the market for information.
analysis: Vogue Italia maintained the same structure
The powerful and recognizable players are more likely to stick to a
over the time frame of our study. The comparability of specific style and to maintain uniformity over time; on the contrary,
data is particularly important for researches, like this new designers are easily associated with newness. Not by chance, a
one, whose object is the study of a trend. Second, we recognized capability of the Vogue editors is the one of ‘discovering’
could collect a large sample that let us obtain reliable and launching new and unknown talents able to invent radically new
styles (Breward, 2003; Jobling and Crowley, 1996).
results (see Appendix A). Third, the collected data are 6 In order to build our database, we did not consider any form of
high-quality data, because articles have been written by advertising paid by companies and we included exclusively the quota-
industry experts and addressed to qualified and com- tions mentioned in the magazine articles.
7 As from Appendix C, the population of the analysis has changed
petent readers. Finally, the longstanding reputation of
Vogue and need for news in the market for information over time. Indeed, the years 1985–1990 are considered as years of
radical cultural developments that invested different aspects of social
is the guarantee of a sustainable degree of neutrality and
and economic life. In the fashion industry this expansion has boosted
independence in judging companies within the industry.5 the growth of existent fashion houses, but has also encouraged the
emergence of new designers in France, Italy, the US, and Japan. The
1990s are considered as a period of recession, with a strong reaction
5 The longstanding reputation of Vogue is the best guarantee against to the ostentatious consumption and production that had characterized
a systematical preference for the most powerful players. In addition, a the previous years. A new sobriety is introduced, especially during the
real judgmental independence is maintained also because of the contex- second half of the 90s, with a strong attention from many designers
tual action of two relevant forces, one favouring incumbents (for their to wearability, quality and value. These tendencies, together with a
recognizability, their economic power, etc.) and the other favouring contraction in the consumption, have acted as a deterrent to companies’
new entrants (for the newness they represent). In fact, press commu- proliferation (Mendes and De La Haye, 1999). Finally, the last two
nication is completely based on news: in the absence of news in the years of our data (2000–2002) are those characterized again from a
 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290 1281

The study sample seems to be consistent with previ- 3.3.2. New entrants and incumbents
ous researches conducted on technological convergence In order to test Hypotheses 2a, 2b, 3a and 3b, we split
and relative to a specific industry (e.g., Tegarden et al., our sample in new entrants and incumbents. Entrances
1999; Wade, 1995). and exits are measured not as entrances and exits from
the industry (i.e. economic failures), but as entrances and
3.3. Variables exits in and from the magazine in terms of citations. A
company is a new entrant in a given year if it is cited in
3.3.1. Style, convergent style, divergence and the magazine for the first time that year, or if it is cited
convergence again after a 2-year period of absence in the magazine.
Consistently with the definition previously given, a A company is an incumbent in a given year if it is cited
style is the combination of the following elements: at least once during the previous year, or if it has a max
1-year period of absence in the magazine.8
(i) The selection of the items of clothing to use (skirts, In order to control our definition of “new entrants”
trousers, jackets, sweaters, dresses and suits) and and “incumbents”, we have run two additional analy-
how they are combined. ses: the first one using two different fashion magazines
(ii) The selection of the cut (close-fitting, constructed, (Elle and MarieClaire), and the second on a different
soft, etc.), color (muted or bright colors), length source, where we considered runway shows that took
(especially skirts and trousers: mini-skirts, knee- place in London, Milan, and Paris during the whole
length or calf-length skirts, etc.), fabric (synthetic period of our study. In both analyses, we obtained high
or natural, processed or unprocessed raw material), inter-rate agreements between the results of the control
and patterns (printed, geometric or flower designs, analyses and the results of our sample (see Appendix
etc.) and how they are combined. E for details). Yet, we decided to keep the citations in
Vogue Italia as our main source for a conceptual rea-
We identify periods of convergence and divergence of son: for stylistic innovation magazines play a key role in
companies on styles. In particular, one period is defined the reattribution of social meaning. Fashion shows are
as a period of “incremental change” if it is characterized relevant media of communication for a restricted group
by a convergent style adopted by the majority of com- of experts – i.e. fashion editors, journalists, and buy-
panies within the sample (over 50%); in the opposite ers – and the content of the shows becomes socially
case it is defined as a period of “ferment” (Tushman and perceived only when it is communicated through the
Anderson, 1986). Furthermore, a period of ferment is magazines.
characterized by a greater mean number of new designs
than a period of incremental change (Anderson and
Tushman, 1990). Given these definitions, in order to test 3.3.3. Innovation and stylistic success rates
Hypothesis 1, we identify the percentage of companies In order to test Hypotheses 2a and 2b, we measured
adopting each style. We classify as convergent a style each company’s innovation rate. Empirically, the inno-
adopted by at least 50% of the reference population in a vation rate for a company in a given year is defined as the
given year. Then, we measure the number of new styles. number of citations about new styles related to this com-
Each style is considered as new for a particular year if it pany divided by the total number of citations totalized by
is presented for the first time that year. This means that the same company for that year. Finally, we compute the
in a new style there could be elements (cut, color, length, mean innovation rate of new entrants and of incumbents.
fabric and pattern/graphics) already adopted in the past, In order to test Hypotheses 3a and 3b, we introduce
but their combination is new, never seen before. In each a new measure: the stylistic success rate. A company
year, we have a percentage of old styles and a percentage is defined as stylistically successful in a given year if
of new styles; we take into account the mean number of it totalized a number of citations higher than the mean
new styles per year in each period of ferment and the number of overall quotations per year (see Appendix D).
mean number of new styles in the period of incremental As we remarked, a stylistic innovation exists if it
change. is communicated and socially perceived: a player in a
stylistic arena is successful if it is socially recognized

sort of renaissance in fashion. After almost a decade of minimalism

the beginning of the new century has assisted to the emergence of 8 According to the experts, a 2-year period of absence from the most

redundancy in styles, a research for happiness, colors, abundance of quoted magazines is reasonably long (it includes four collections) to
details (Breward, 2003). determine the exclusion of a company from stylistic competition.
1282 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290

Fig. 1. Periods of ferment, periods of incremental change and convergent styles in fine fashion.

in a stronger way than its competitors, and if it is able companies’ positions (incumbents or new entrants) on
to enact a secondary diffusion process stronger than its stylistic innovation and stylistic success.
competitors. As a consequence, the success of stylis- To test Hypothesis 1, we first worked to identify the
tic innovation has to be assessed on the visibility this presence of periods of convergence of fine fashion com-
innovation gains on relevant communication tools. This panies on a specific style. Our data show that, notwith-
concept, we named explicitly ‘stylistic success’, is a standing the multiplicity of styles and the need for differ-
construct very close to the concept of media reputation entiation, it is possible to identify a sequence of periods
developed by Deephouse (2000), defined as “the overall of convergence and periods of divergence. More specif-
evaluation of a firm presented in the media” (p. 1097). ically, over the time frame of our study (1984–2002)
In this sense, it expresses the consideration a specific we identified three reference styles on which companies
company has from the most influential media compa- have reached convergence. Fig. 1 shows the development
nies should target. Therefore, stylistic success needs to of fine fashion styles and trends between 1984 and 2002
be represented by a measure of result (has the company and depicts the three convergent styles.
reached visibility?) and not by a measure of effort (how The first convergent style is the ‘masculine’ one: in
much effort has the company exerted to reach visibil- 1984, 60.6% of companies adopted this style. It can
ity?). In fact, what is relevant to judge in order to evaluate be traced back to the ‘career woman’s uniform’ first
the stylistic success is the results in the secondary dif- introduced by the Italian company Armani at the begin-
fusion process independently from the effort exerted by ning of the 1980s.10 The first period of ferment occurs
the company to reach that result.9 between 1985 and 1990, when the relative frequency
associated with the masculine style gradually decreases
4. Empirical results from 29.7% to 0.51%. In this period many diametri-
cally opposed styles emerge.11 In 1989 the minimal
The paper aims at evaluating the strength of the evolu- style appears on the scene (31.4%). In contrast with the
tionary model for the stylistic innovation. Consequently previous androgynous style, it is characterized by an
our empirical analyses test companies’ convergence over
time on specific styles and evaluate the impact of the
10 It is worth pointing out that in our study this period lasts just for one

year because we started collecting data from 1984. This period could
9 Gaining visibility implies companies to reach a critical mass and be longer but, according to the experts we contacted, the features of the
companies of different sizes and/or with different positions within the masculine style in the previous years were not significantly different.
industry might have to exert different efforts to reach the same critical 11 More specifically, the emergent styles go from the feminine one

mass. But, if the result to be achieved is the affirmation and the visibility with plunging necklines, extremely short, fitted lines in fluorescent
of a specific stylistic innovation, small companies and large companies, colors (42.5% in 1985) to the “chaste” style characterized by dull col-
new entrants or incumbents, have to reach a similar level of visibility to ors, calf-length, tapered dresses, large belts and low boots (42.3% in
be considered as stylistically successful. Consequently, companies that 1986) and to the romantic style of 1987 (36.4%), characterized by
are not able to reach this level of visibility do not succeed in stylistic loose-fitting suits made up of a jacket and skirt in pastel colors or
innovation. flower patterns and decorated with lace, embroidery and gloves.
 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290 1283

Table 1
New styles within stylistic cycles: ferment vs. incremental changes
Era Years Mean new styles per year T d.f.

Ferment 1985–1990, 1994–1998, 2002 6.08

Incremental change 1991–1993, 1999–2001 4.16 8.000*** 6
*** p < 0.001.

extremely feminine cut (skirts are more common than per year introduced during the eras of ferment is 6.08 (in
trousers), unstructured clothing in pale colors, neutral particular, 6.83 in the first period of ferment, 4.8 in the
tones and unusual fabrics. The ‘minimal’ style beats second, and 8 in the third), whereas the mean number of
the competition and becomes the second convergent new styles per year introduced during the following eras
style: from the end of 1990 and the starting of 1991 of incremental change is 4.16 (in particular, 3.66 styles
the relative frequency of this style increases from 31.6% in the first period of incremental change and 4.66 in the
to 68.1%. The following period of incremental change second).12
lasts from 1991 until 1993: during this period the min- Given the analyses of the percentage of adopters and
imal style decreases from 68.1% until about 50%. The of the mean number of new styles, we can conclude that
second period of ferment starts at the end of 1993 as Hypothesis 1 is supported.
the minimal style decreases below the 50%. During Given the evolutionary model for stylistic innovation,
this period, one new rival style emerges, namely, the to test Hypothesis 2 we worked to identify if companies
kitsch style (between 1995 and 1998 its relative fre- in different positions play different roles in the different
quency increases from 17.1% to 42.8%). From the very periods.
beginning, it presents itself as directly opposed to the First, our data show that, as for technology-based
minimal style in its use of artificial materials, strong industries, incumbents are more innovative than new
colors and combinations that clash. The ‘kitsch’ style entrants during the periods of incremental change; on the
becomes the convergent style between 1998 and 1999: contrary new entrants are more innovative than incum-
during these years this style increases from 42.8% to bents during the periods of ferment.
75.7% in designers’ preferences. It is characterized by More specifically, during the eras of incremental
combinations of traditional fabrics (silk, linen, satin) change, the average innovation rate per year for incum-
with new materials such as metal, PVC, plastic, in col- bents is 0.22 (in particular, 0.20 in the first period of
ors decorated with sequin, rhinestones, feathers, fringes incremental change and 0.23 in the second), whereas the
and tiny mirrors. The following period of incremental average in innovation rate per year for new entrants is
change lasts until 2001, when the kitsch style drops 0.11 (in particular, 0.09 both in the first period and 0.14
below the 50%. in the second). Hypothesis 2a is supported by results
The third period of ferment starts in 2002. As we (see Table 2): incumbents’ innovation rates per year dur-
stopped our research in 2002, we have no information ing the periods of incremental change (1991–1993 and
for determining its length. We define 2002 as a period 1999–2001) are significantly higher than new entrants’
of ferment, though, because the kitsch style percentage ones (p < 0.01).
drops down to 12.4%.
In order to complete the test of Hypothesis 1, we
measure the number of new styles introduced during 12 As we remarked previously, the social and economic landscape
the different periods (following Anderson and Tush- has evolved over the 20 years considered. In particular, the second
man’s definition of ferment period and incremental period of ferment falls in a moment of strong economic recession.
change period). Table 1 reports the mean number of As a consequence, the value of the mean number of new styles in
this period is lower than its respective one in the two other periods
new styles per year introduced during periods of fer- (6.83 in 1985–1990 and 8.0 in 2002). On the contrary, the second
ment (1985–1990, 1994–1998, and 2002) and incremen- period of incremental change falls exactly at the beginning of the eco-
tal change (1991–1993, and 1999–2001). In support of nomic renaissance (and indeed 2001 is a year with a high increase
Hypothesis 1, the mean number of new styles intro- in the number of designers). Of course, the influence exerted by gen-
duced every year during periods of ferment is signifi- eral environmental changes is important, but it does not change the
predictions of the evolutionary model. Indeed, the difference between
cantly greater than the mean number of new styles intro- the second period of ferment and the second period of incremental
duced each year during periods of incremental change change, although smaller than those registered in the other periods, is
(p < 0.001). In our study, the mean number of new styles still positive.
1284 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290

Table 2 Table 5
Incremental changes: new entrants vs. incumbents innovation rate Ferment: new entrants vs. incumbents success rate
Company Mean innovation T d.f. Company Mean success rate T d.f.
rate per year per year

New entrants 0.11 New entrants 0.23

Incumbents 0.21 −5.590** 5 Incumbents 0.36 −4.890*** 11
** p < 0.01. *** p < 0.001.

Table 3 rate per year for incumbents is 0.36 (in particular, 0.39 in
Ferment: new entrants vs. incumbents innovation rate the first period of ferment, 0.34 in the second, and 0.30 in
Company Mean innovation T d.f. the third), whereas the average success rate per year for
rate per year new entrants is 0.23 (in particular, 0.22 in the first period,
New entrants 0.29
0.27 in the second, and 0.23 in the third). This means that
Incumbents 0.20 7.214*** 11 the difference between incumbents and new entrants in
***
periods of ferment is significant but is again positive for
p < 0.001.
incumbents. So Hypothesis 3b did not receive empirical
support (see Table 5). One possible explanation is that,
During the periods of ferment, the average innovation although the best opportunities for new entrants to get
rate per year for new entrants is 0.29 (in particular, 0.33 successfully into the system are in the periods of fer-
in the first period of ferment, 0.26 in the second, and ment, the critical mass necessary to enact the secondary
0.25 in the third), whereas the average innovation rate diffusion process hampers new entrants more than the
per year for incumbents is 0.20 (in particular, 0.24 in the sunk costs in the technology-based industries.
first period, 0.16 in the second, and 0.17 in the third).
Hypothesis 2b is supported by results (see Table 3): new 5. Discussion and conclusion
entrants’ innovation rates per year during the periods of
ferment (1985–1990, 1994–1998, and 2002) are signifi- Our results provide insightful contributions to the the-
cantly higher than incumbents’ (p < 0.001). ory of innovation, exploring the new concept of stylistic
With regard to Hypothesis 3 our data show that incum- innovation and its evolutionary dynamics. First, while
bents’ position represents a strength even greater than in previous studies provided detailed results on the dynam-
technology-based industries (Wade, 1995). More specifi- ics of technological change, we analyze innovation in
cally, we found that during periods of incremental change the aesthetic and symbolic elements of products and
the average success rate per year for incumbents is 0.39 services. Furthermore, we elaborate on the concept of
(in particular, 0.36 in the first period of incremental stylistic success. When studying stylistic innovation, the
change and 0.43 in the second), whereas the average consideration of traditional indicators of performance
success rate per year for new entrants is 0.24 (in par- might drive to misleading conclusions and the results
ticular, 0.21 in the first period and 0.29 in the second). gained in the secondary diffusion process represent a
Hypothesis 3a was supported (see Table 4): incumbents better indicator (Hirschman, 1982). Another major con-
are more successful than new entrants in periods of incre- tribution of our study is the test of a counterintuitive
mental change. convergent evolutionary model within a symbol inten-
Yet, in contrast to what we hypothesized, incumbents sive context. Indeed, so far research on symbolic and
are significantly more successful than new entrants also cultural industries has emphasized the importance of dif-
during the eras of ferment. In fact, the average success ferentiation in competitive dynamics: companies tend to
leverage and emphasize the uniqueness of their inter-
Table 4 nal resources and image. On the contrary, our empirical
Incremental changes: new entrants vs. incumbents success rate results show that companies develop convergent behav-
Company Mean success rate T d.f.
iors also as far as symbolic and aesthetic changes are
per year concerned. As for technological innovations it is possible
to distinguish between periods of convergence—where
New entrants 0.24
Incumbents 0.39 −8.589*** 5
the major part of the companies adopt the same style and
periods of divergence or ferment—where different styles
*** p < 0.001. live together.
 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290 1285

Yet, the different drivers behind the convergence innovation – in particularly if compared to technologi-
dynamics produce relevant differences between tech- cal one – and, as a consequence, this phenomenon offers
nological and stylistic innovation. Unlike technological a shorter time span of analysis for research. Third, our
innovation, there is never a dominance phenomenon study relies on the use of quotations from a leading mag-
but a convergence one. Considering the emergence of azine within the context under analysis. This choice, as
snob effects, consumers’ need for distinction, and the discussed earlier, is due to the nature of stylistic innova-
signalling power of style for companies, we found the tion and stylistic success and we have to trust the long-
presence of convergent designs rather than exclusively standing reputation of the magazine used as guarantee
dominant designs. Indeed, in the case of stylistic inno- of independent judgment. This is a common limitation
vation the new standard does not render the previous for longitudinal studies, especially for those industries
one obsolete and does not completely replace it. Indeed, under explored in literature. Furthermore, when sym-
the characteristics of stylistic innovation drive the cre- bolic and aesthetic changes are under investigation the
ation of market niches where those companies that do objective track of product modifications does not allow
not adopt the convergent style or do not exclusively any conclusion and secondary sources play the funda-
focus on that trend, might also come out to be stylis- mental role.
tically successful. Just as an example, Dolce & Gabbana
from the time it appeared on the scene (1986) suc- 5.2. Implications and further research
cessfully differentiated itself by proposing a rich and
baroque style inspired to the 1950s Sicilian context while Notwithstanding the discussed caveats, we may con-
competitors were converging on the minimalist style. clude that our logic has served its purpose and offers
In addition, each company adopts more than one style implications and insights for further explorations.
per year, even if one of these can prevail over the oth- First, we focused on stylistic innovation with a spe-
ers. An example is provided by Prada, considered as cific reference to symbolic industries, but we expect to
the creator of minimalism in fine fashion. During the find similar dynamics in other industries. Whenever a
minimalist era, Prada adopted two other styles (the mas- “stylistic” content is present even marginally, companies
culine and the dandy style) as well, even if with lower need to look for some coherence in the aesthetic and
emphasis. symbolic features of their offers. An example is given
Likewise technological innovation, our empirical by the mp3 products and, in particular, by the case of
study highlights the relevance of the companies’ posi- iPod introduced by Apple in 2001, that has gained in
tions for the innovation and the success rate. Regarding less than 1 year the 60% of the digital player market
the innovation rate, we found strong similarities between also leveraging the stylistic appeal of a timely design.
technological and stylistic innovation: incumbents are Furthermore, Apple has been able to revitalize the mar-
more innovative than new entrants in periods of incre- ket for these mature products by leveraging its symbolic
mental change because of their knowledge of the existent and aesthetic dimensions. Indeed, as soon as the techni-
styles; new entrants are more innovative than incum- cal standard has affirmed, companies may use stylistic
bents in periods of ferment because of their ‘out of the innovation to differentiate the product and gain a pre-
schemata’ character and the lack of inertial frames, cog- mium price (Eisenman, 2004).
nitive ‘sunk costs’, and existing routines. Regarding the This consideration drives another relevant insight of
success rate, we found some specificities: the nature of our study, related to complementarities between stylistic
stylistic success related to the secondary diffusion pro- and technological innovation. We believe that there are
cess seems to favor incumbents even more than in the conditions under which technological innovation and the
technology-based industries. stylistic innovation reinforce each other. In our empirical
research complementarities emerge when technological
5.1. Limitations innovations in the materials and into the production sys-
tem enlarge the potential variety companies can rely on
As with all research, ours is subject to limitations. to create new styles. An example is given by the introduc-
A first limitation is in the consideration of a single tion of new synthetic materials (nylon, PVC, synthetic
industrial setting, that is the fine fashion one. Although jersey) and by the treatment of natural fibers to cre-
we chose this industry because revelatory for the phe- ate rubber-similar effects that fostered the emergence
nomenon under investigation, we are aware that addi- of the ‘ultra-modern’ style. The complementarities are
tional researches could strengthen our evidences. Sec- even more evident in technology-based context, when
ond, stylistic innovation is a relatively young kind of the power of stylistic innovation is used to make a tech-
1286 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290

nological innovation be accepted more easily on the ipants at the 2004 Academy of Management Conference
market. This is the case of the mentioned example of and 2004 SMS Conference.
iPod, where the innovation in the message is amplified by
the newness of the technological features and technical Appendix A. Extract from the dataset with
functions are more appreciated because of the stylistic source, events, classifications and the companies’
innovation. names
The insights we presented open to further researches
on stylistic innovation in different industries and on the The Vogue articles have a standard format. They are
exploration of complementarities between stylistic and usually structured in two parts: one is an average-length
technological innovation, so far underdeveloped in the text (10–20 lines) and briefly expounds the style (it can
literature. be the presentation of a new trend or a service about
In conclusion, we believe our study significantly con- a designer); the second is the photograph reportage
tributes to shed light on a complex multifaceted phe- that describes, through images, what was previously
nomenon – stylistic innovation – and offers a new path described in words. We collected data about the year
for investigating innovation also in over explored con- (1 year means that the dataset contains the elaboration
texts. of data coming from the 12 magazine issues published
in that year); the style (namely, the categories used to
Acknowledgements classify the different styles); the style elements (in terms
of cut, color, length, fabric and pattern/graphics); and the
We gratefully acknowledge the valuable comments absolute and relative frequency of companies (obtained
of Raghu Garud, Dennis Gioia, Martin Hoegl, Vincenzo by dividing the absolute frequency by the yearly popula-
Perrone, Luigi Proserpio, Severino Salvemini, and Gian- tion of reference). It is worth pointing out that the method
mario Verona on earlier versions of the manuscript, adopted in this analysis does not allow us to encompass
and the constructive feedback from the two anonymous photographs as a source of information. In the following
reviewers. The paper benefited from comments of partic- table an example of the data collection and classification.

Year 1993: data classification

Elements of style Style classification Absolute frequency (%) Relative frequency (%)

Long dresses; simple cuts; light cloths Minimal 58 52.2

Simple cuts; black or striped clothes Masculine 35 31.5
White; soft style Colonial 32 28.8
White, gray, black Military 27 24.3
Jacket; laces Dandy 26 23.4
Delicate colors Lingerie 25 22.5
Recycled cloths; upside down dresses Grunge 23 20.7
Hippy style; patchwork 70s 19 17.1
Many drapes Asiatic 18 16.2
Moroccan shoes and hats; nomadic bag Ethnic 15 13.5
Strong colors; micropieces Hyperfeminine 12 10.8
Stretch cloths Sportive 11 9.9
Saint-gallen laces; chiffon; volants; flounces; Baby-doll 11 9.9
ribbons; dresses similar to antique dolls
Fur; pvc; enamel New materials 9 8.1
Feathers Glamorous jeans 5 4.5
 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290 1287

Appendix B. Sample and number of citations

Appendix C. Percentage of new companies and well-established companies in the sample

Year Total number of Total number of New entrants on total Total number of Incumbents on total
companies in the new companies number of companies incumbents number of companies
sample per year (%) per year (%)

1985 202 103 50.99 99 49.01

1986 186 74 39.78 112 60.22
1987 171 88 51.46 83 48.54
1988 183 56 30.60 127 69.40
1989 161 59 36.65 102 63.35
1990 192 82 42.71 110 57.29
1991 113 20 17.70 93 82.30
1992 120 33 27.50 87 72.50
1993 111 39 35.14 72 64.86
1994 102 28 27.45 74 72.55
1995 99 22 22.22 77 77.78
1996 125 45 36.00 80 64.00
1997 119 23 19.33 96 80.67
1998 120 24 20.00 96 80.00
1999 106 25 23.58 81 76.42
2000 108 23 21.29 85 78.7
2001 265 138 52.07 127 47.92
2002 272 114 41.91 158 58.08
1288 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290

Appendix D. Successful new entrants and incumbents

Year Total number Number of Successful new Successful new Total Number of Successful Successful
of new successful entrants on total entrants on total number of successful incumbents on incumbents on
entrants new number of new number of incumbents incumbents total number of total number of
entrants entrants (%) companies in the incumbents (%) companies in the
sample (%) sample (%)

1985 103 28 27.18 13.86 99 43 43.43 21.29

1986 74 11 14.86 5.91 112 38 33.93 20.43
1987 88 17 19.32 9.94 83 42 50.60 24.56
1988 56 17 30.36 9.29 127 53 41.73 28.96
1989 59 13 22.03 8.07 102 31 30.39 19.25
1990 82 14 17.07 7.29 110 37 33.64 19.27
1991 20 4 20.00 3.54 93 31 33.33 27.43
1992 33 8 24.24 6.67 87 30 34.48 25.00
1993 39 7 17.95 6.31 72 28 38.89 25.23
1994 28 8 28.57 7.84 74 28 37.84 27.45
1995 22 6 27.27 6.06 77 26 33.77 26.26
1996 45 18 40.00 14.40 80 29 36.25 23.20
1997 23 4 17.39 3.36 96 30 31.25 25.21
1998 24 5 20.83 4.17 96 30 31.25 25.00
1999 25 6 24.00 5.66 81 29 35.80 27.36
2000 23 10 43.47 9.25 85 48 56.47 44.44
2001 138 28 20.28 10.56 127 48 37.79 18.11
2002 114 26 22.80 9.55 158 47 29.74 17.27

Appendix E. Control analyses for data sources year, or if it has a max 1-year period of absence from
the shows. With the data we collected from the shows,
To control our definition of “new entrants” and we run the inter-rater agreement test. The test’s results
“incumbents”, we have carried out two additional anal- are positive: the inter-rater agreement between the data
yses: the first one using two different fashion magazines on fashion shows and the data from Vogue is 0.883. This
(Elle and MarieClaire instead of Vogue), and the sec- means that using the presence at the fashion shows and
ond on a different source, where we considered runway using the citations on Vogue for the whole period of 19
shows that took place in London, Milan, and Paris during years the results agreed on 88.3% of the classification of
the whole period of our study. For the first control anal- “incumbents” and “new entrants”.
ysis, the one on two additional fashion magazines, we
created a control sample for two periods (1988–1989 and References
1999–2001) for a total number of 120 issues analyzed.
We used the same methods for classifying the events, cal- Abernathy, W.J., Utterback, J.M., 1978. Patterns of industrial innova-
culating the number of citations per firm, and computing tion. Technology Review 40–47.
Abernathy, W.J., Clark, K., 1985. Mapping the winds of creative
the frequencies of adoption. Then, we compared these destruction. Research Policy 14, 3–22.
results with our findings, running the inter-rater agree- Abrahamson, E., Rosenkopf, L., 1997. Social network effects on the
ment test: for the control time frame under analysis the extent of innovation diffusion: a computer simulation. Organization
inter-rater agreement between Elle and Vogue is equal Science 8, 289–300.
to 0.788; the inter-rater agreement between MarieClaire Adner, R., Levinthal, D., 2001. Demand heterogeneity and technol-
ogy evolution: implications for product and process innovation.
and Vogue is equal to 0.823. For the second control, Management Science 47, 611–629.
based on the runway shows that took place in London, Anderson, P., Tushman, M., 1990. Technological discontinuities and
Milan and Paris during the whole period, a company is dominant designs: a cyclical model of technological change.
defined as a new entrant in a given year if it appears Administrative Science Quarterly 35, 604–633.
in at least one of these fashion shows for the first time Barthes, R., 1970/1982. Empire of Signs. Hill and Wang,
New York.
that year, or if it appears again after a 2-year period of Barthes, R., 1983. The Fashion System. Hill and Wang, New York.
absence in the shows. A company is an incumbent in Bijker, W., Hughes, T., Pinch, T., 1987. The Social Construction of
a given year if it is present at least once the previous Technological Systems. MIT Press, Cambridge, MA.
 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290 1289

Bloch, P.H., Brunel, F., Arnold, T.J., 2003. Individual differences in the Hounshell, D., 1984. From the American System to the Mass Produc-
centrality of visual product aesthetics: concept and measurement. tion. John Hopkins University Press, Baltimore.
Journal of Consumer Research 29, 551–565. Hughes, T.P., 1983. Networks of Power. John Hopkins University
Bourdieu, P., 1979/1984. Distinction: A Social Critique of the Press, Baltimore.
Judgement of Taste. Harvard University Press, Cambridge, Jobling, P., Crowley, D., 1996. Graphic Design: Reproduction and Rep-
MA. resentation Since 1800. Manchester University Press.
Breward, C., 2003. Fashion. Oxford University Press, Oxford. Lawson, B., 1983. How Designers Think. East-View Editions, West-
Cappetta, R., Gioia, D., 2005. Fine fashion: symbolic artifacts, sense- field.
giving, and sensemaking in the construction of organizational Leibenstein, H., 1950. Bandwagon, snob, and veblen effects in the
identity and image. In: Rafaeli, A., Pratt, M. (Eds.), Artifacts and theory of consumers’ demand. Quarterly Journal of Economics 64
Organizations. Lawrence Erlbaum Publishers. (2), 183–207.
Caves, R., 1999. Creative Industries. Harvard Business School Press, Leonard Barton, D., 1995. Wellsprings of Knowledge. Building and
Cambridge, MA. Sustaining the Sources of Innovation. Harvard Business School
Chandy, R.K., Tellis, G.J., 1998. Organizing for radical product inno- Press, Cambridge, MA.
vation: the overlooked role of willingness to cannibalize. Journal Mendes, V., De La Haye, A., 1999. Twentieth-Century Fashion.
of Marketing Research 34, 474–487. Thames and Hudson Ltd.
Christensen, C., Bower, J., 1996. Customer power, strategic invest- Pesendorfer, W., 1995. Design innovation and fashion cycles. The
ment, and the failure of leading firms. Strategic Management Jour- American Economic Review 85, 771–792.
nal 17, 197–219. Postrel, V., 2003. The Substance of Style. Harper Collins Publishers,
David, P., 1987. Some new standards for the economics of standardiza- New York.
tion in the information age. In: Dasgupta, P., Stoneman, P. (Eds.), Postrel, V., 2001. Aesthetics and instrumentality: can good looks
Economic Policy and Technological Performance. Cambridge Uni- guarantee a product’s success? Available at http://www.nytimes.
versity Press, Cambridge. com/2001/07/12.
David, P., Greenstein, S., 1990. The economics of compatibility stan- Rafaeli, A., Vilnai-Yavetz, I., 2005. Managing organizational artifacts
dards. An introduction to recent research. Economics of Innovation to avoid artifact myopia. In: Rafaeli, A., Pratt, M. (Eds.), Artifacts
and New Technology 1, 3–42. and Organizations. Lawrence Erlbaum Publishers.
Davis, F., 1992. Fashion Culture and Identity. University of Chicago Richardson, J., 1996. Vertical integration and rapid response in fashion
Press, Chicago. apparel. Organization Science 7, 400–412.
Deephouse, D.L., 2000. Media reputation as a strategic resource: an Robinson, D., 1961. The economics of fashion demand. The Quarterly
integration of mass communication and resource-based theories. Journal of Economics 75, 376–398.
Journal of Management 26 (6), 1091–1112. Schmitt, B., Simonson, A., 1997. Marketing Aesthetics: The Strategic
Descamps, M.-A., 1979. Psychosociologie de la mode. PUF, Paris. Management of Brands Identity and Image. Free Press, New York.
Djelic, M., Ainamo, A., 1999. The coevolution of new organization Simmel, G., 1985. Die Mode.
forms in the fashion industry: a historical and comparative study Simmel, G., 1957. Fashion. The American Journal of Sociology 62,
of France, Italy and the United States. Organization Science 10, 541–544.
622–637. Simon-Miller, F., 1985. Commentary: signs and cycles in the fashion
Dumaine, B., 1991. Design that sells and sells and . . .. Fortune 11, system. In: Solomon, R. (Ed.), The Psychology of Fashion. Lex-
86–94. ington Books/D.C. Heath, Lexington, MA.
Eco, U., 1976. A Theory of Semiotics. University of Indiana Press, Solomon, M., Rabolt, N., 2004. Consumer Behavior in Fashion. Pear-
Bloomington, IN. son Education Inc, New Jersey.
Eisenhardt, K.M., Tabrizi, B.N., 1995. Accelerating adaptive pro- Strati, A., 1992. Aesthetic understanding of organizational life.
cesses. Product innovation in the global computer industry. Admin- Academy of Management Review 17, 568–581.
istrative Science Quarterly 40, 84–110. Suarez, F., Utterback, J., 1995. Dominant design and the survival of
Eisenman, M., 2004. Mature Industries as Fashion Systems. Working firms. Strategic Management Journal 16, 415–430.
Paper. Columbia Business School. Tegarden, L.F., Hatfield, D., Echols, A., 1999. Doomed from the start:
Forty, A., 1986. Objects of Desire. Pantheon Books, New York. what is the value of selecting a future dominant design. Strategic
Garcia, R., Calantone, R., 2002. A critical look at technological Management Journal 20, 495–518.
innovation typology and innovativeness terminology: a litera- Tripsas, M., 1997a. Unravelling the process of creative destruction:
ture review. Journal of Product Innovation Management 19, 110– complementary assets and incumbent survival in typesetter indus-
132. try. Strategic Management Journal 18, 119–142.
Garud, R., Kumaraswamy, A., 1995. Technological and organizational Tripsas, M., 1997b. Surviving radical technological change through
design for realizing economies of substitution. Strategic Manage- breakthrough dynamic capability: evidence from the typesetter
ment Journal 16, 93–109. industry. Industrial and Corporate Change 6 (2), 341–377.
Henderson, R., Clark, K.B., 1990. Architectural innovation: the Tushman, M., Anderson, P., 1986. Technological discontinuities and
reconfiguration of existing product technologies and the fail- organizational environments. Administrative Science Quarterly 31,
ure of established firms. Administrative Science Quarterly 35, 439–466.
9–31. Tushman, M., O’Reilly III, T., 1997. Winning through Innovation.
Hirschman, E., 1982. Symbolism and technology as sources for the Harvard Business School Press, Cambridge, MA.
generation of innovations. Advances in Consumer Research 9, Utterback, J., Abernathy, J., 1975. A dynamic model of process and
537–541. product innovation. Omega 3, 639–656.
Hirschman, E., 1986. The creation of product symbolism. Advances Veblen, T., 1899/1992. The Theory of the Leisure Class. Transaction
in Consumer Research 13, 327–331. Publishers, London.
1290 R. Cappetta et al. / Research Policy 35 (2006) 1273–1290

Veryzer, R., Hutchinson, W., 1998. The influence of unity and proto- tion mechanisms, and organizational resources for competition in high
typicality on aesthetic responses to new product design. Journal of symbolic industries. She published on Organization Science.
Consumer Research 24 (4), 374–395.
Paola Cillo is Assistant Professor of the Institute of Management
Volli, U., 1990. Contro la moda. Feltrinelli, Milano.
and Business Economics at Bocconi University Milano (Italy), where
Wade, J., 1995. Organizational communities and technological band-
she teaches courses on innovation and new product development. She
wagons. Strategic Management Journal 16, 111–133.
earned her Ph.D. from Bocconi University in 2001. She has been Vis-
iting Scholar at Wharton School, 2000–2001. She teaches at the SDA
Rossella Cappetta is Assistant Professor of the Institute of Organi- Bocconi School of Management. Her research interests focus on mar-
zation and Information Systems at Bocconi University Milano (Italy), ket orientation and innovation in symbol intensive industries.
where she teaches courses on organization theory, organization design
and research methods. She earned her Ph.D. from Bocconi University Anna Ponti is Doctoral Student at the Bocconi University Ph.D. Pro-
in 2001. She has been Visiting Scholar at Wharton School, 2000–2001. gram. She has been Visiting Scholar at Queen’s University, Kingston,
She teaches at the SDA Bocconi School of Management. Her research Ontario, Canada. Her research interests focus on organizational
interests focus on knowledge transfer, intra and interfirm coordina- resources for competition in high symbolic industries.