How Fashion Designers Develop New Styles:

Document Sample
How Fashion Designers Develop New Styles: Powered By Docstoc

                     How Fashion Designers Develop New Styles:

                    Creative Epiphany Versus Market Feedback

                                      JOSEPH C. NUNES

                                       XAVIER DRÈZE

                                        PAOLA CILLO

                                  EMANUELA PRANDELLI

                                     IRENE SCOPELLITI

Joseph C. Nunes is Associate Professor of Marketing at the Marshall School of Business,
University of Southern California, Los Angeles (e-mail:

Xavier Drèze was Professor of Marketing, Anderson School of Management, University of
California, Los Angeles (e-mail:

Paola Cillo is Associate Professor of Management at the SDA Bocconi School of Management,
Bocconi University, Milan (e-mail:

Emanuella Prandelli is Associate Professor of Management at the SDA Bocconi School of
Management, Bocconi University, Milan (e-mail:

Irene Scopelliti is a post-doctoral research fellow in Marketing the Tepper School of Business,
Carnegie Mellon University, Pittsburgh (e-mail:

Questions should be directed to Joseph C. Nunes at

                     How Fashion Designers Develop New Styles:

                     Creative Epiphany Versus Market Feedback


Iconic fashion designers see themselves as artistic visionaries whose creations are impervious to
market critique. Using real-world data, this work refutes this notion, and is the first empirical
study to document a market response model of fashion. We show empirically how elite European
fashion houses introduce new styles based on how previous styles have been received by the
market. More specifically, we find designers introduce styles that are more similar to styles that
were successful in the past and less similar to styles that were less well-received. Further, the
styles a designer introduces are shown to depend on the market’s response to competing
designers’ styles. In showing how styles evolve systematically over time, this work illustrates the
systematic integration of market feedback on new product introductions, doing so in an industry
based on aesthetic innovation.

KEYWORDS: Fashion, Innovation, Creativity, Market Response, Learning, Feedback, Style

 “I am first and foremost a designer. (It is my job to take an intangible feeling and turn it into a
tangible article that a customer can buy or wear.) I read sales figures daily and then throw them
in the garbage. That is when I once again become a designer.”

                                                                                         Tom Ford
                                                                                  Fashion Designer

“All designers love to think that they are not commercial, that they are these creative, fragile
souls, just living in ivory towers. Well, you know, let them have that fantasy, but the bottom line
is that the bottom line is incredibly important.”

                                                                                   Lisa Armstrong
                                                                         Fashion Editor, The Times

Whether it was the invention of stiletto heels in the 1950s, miniskirts in the 1960s, or structure-

less jackets in the 1970s, the fashion world is characterized by change. Like other creative

industries including music, theater, and publishing, the public appetite for something novel in

fashion seems insatiable. Fashion thrives on change, and the success of the industry as a whole

depends on its ability to introduce new styles. World renowned design houses such as Chanel,

Prada, Gucci, and Balenciaga and their top designers Karl Lagerfeld, Miuccia Prada, Frida

Giannini, and Nicolas Ghesquière are devoted to introducing distinctive, original styles each

season. But how fashion designers derive their inspiration is often shrouded in mystery. Like

musicians, actors, and authors, designers are prone to see themselves as visionaries and creativity

as an epiphany, impervious to outside influence. For example, Karl Lagerfeld, creative director

at the helm of Chanel since 1983, claims his best work is effortless, coming to him in his sleep

(Chiles 2002).

       The role of designers as creative geniuses determining what new styles would be worn

began in earnest in Europe in the mid-nineteenth century. By stipulating who could wear what,

centuries of sumptuary laws made one’s attire a visible register of status differences. As a

byproduct, these mandates created an intense and enduring interest in apparel and secured its

symbolic role in communicating about the self. With increasing urbanization and the rise of the

bourgeoisie, by 1630 the notion of “fashion” in women’s apparel became an important factor in

European society (Freudenberger 1963). By this time clothes were no longer made at home but

were mass -produced, and a sense of aesthetics emerged as an increasing number of social

groups gained access to a wider range of attire (Belfanti and Giusberti 2000). Englishman

Charles Worth, who established his own design studio in Paris in 1858, was the first couturier to

sew his name into his clothing and thus brand his apparel. Worth designed one-offs for his elite

clients while selling collections of designs to a wider clientele. His popularity allowed Worth to

dictate what his customers should wear, and it made him the “supreme arbiter of taste” (De

Marly 1980, p. 23). His protégé and fashion impresario Paul Poiret defended the role of designer

as dictator when he famously proclaimed “Fashion needs a tyrant,” (Seeling 2010).

       While many of today’s designers have come to personify innovation in fashion

(Kawamura 2005), like Ford, iconic designers eschew the notion that they are concerned with

commercial success. Once asked by a reporter whether his spring haute couture show featured

pink to appeal to a younger market, Chanel creative director Karl Lagerfeld replied it would be

the “worst thing” he could do as it would result in “old work” (Giles 2002). Creativity and

originality reign over commercialism. Fashion critics chide designers who fail to display

sufficient originality as New York Times critic Cathy Horyn did when she described Phoebe

Philo’s 2012 collection shown in Paris for Céline as “…a club sandwich of cleverly reworked

ideas.” In creative industries, the mission is frequently seen as “an ephemeral, almost spiritual

quest” unaffected by the rational analysis and market prognoses utilized in most business

disciplines (Beverland 2005, p. 195). The perception of designers at prestigious European

fashion houses (e.g., Chanel, Gucci, Balenciaga) as avant-garde geniuses is propagated by

commemorative celebrations of their unique contributions that are showcased in museums and

galleries worldwide (e.g. the Gianni Versace retrospective at the Victoria and Albert Museum in

London, the 2000-1 show dedicated to the career of Giorgio Armani at the Solomon R.

Guggenheim Museum of Art).

       This research sets out to determine whether fashion designers are indeed immune to

caprices of the market. Specifically, we investigate whether fashion designers attend to market

feedback, and if so, how and to what extent. In doing so, we empirically test if the relative

success or failure of styles introduced in the past affects new styles that are subsequently

introduced. If designers are indeed immune to market feedback, we should observe no

relationship. If, however, designers are shrewd marketers who are attuned to what the market has

to say, the styles they introduce each season should be more similar to those styles that were

more successful in the past and less similar to less successful styles. In this case, we would

observe a relationship between what they do and what they, and their competitors, have done.

       Despite ongoing research in fashion within many disciplines, data driven research is

conspicuously absent in the literature. Perhaps the most recent use of real world data to explore

how styles evolve over time was anthropologist Alfred Kroeber’s 1919 research documenting

how the cut of women’s clothes reflected social and economic developments towards the turn of

the century. The vast majority of recent work has focused on evaluating different models

proposed to explain how new trends diffuse through a population (Yoganarasimhan 2012;

Pesendorfer 1995; Miller, McIntyre and Mantrala 1993; Sproles 1981). Our work attempts to

resolve whether designers are sensitive to how the market responds to what they do, but does not

address where their ideas come from or whether it is the upper class or various subcultures who

act as fashion leaders.

       Whether the new styles introduced each season depend on the market’s acceptance of

individual past styles is an important question, and a topic never before explored. By

documenting whether and how designers systematically integrate market feedback into new

product introduction decisions, we illustrate both the existence and the importance of a market

orientation in an industry based on aesthetic innovation.

       The remainder of this paper is organized as follows. First, we review the relevant

literature on fashion, particularly as it applies to innovation in the fashion industry and the

diffusion of fashion trends. In doing so, three fundamental hypotheses emerge which guided the

development of our model and analysis. We then present our model which tests for market

response in high fashion. Within this section, we describe the rich, real-world data we collected

which allow us to measure the extent to which a design house changes its styles from year to

year and how these styles depend on market feedback. We next present our empirical analyses

and the results before discussing our findings more generally. We acknowledge that much of this

work is exploratory in nature and we obtain a number of interesting results that were not

predicated on theory. Yet these results should help fashion theorists understand fashion trends

better and thus develop better models in the future of how styles evolve. We conclude by

discussing managerial implications, limitations of our findings and opportunities for future


                               Relevant Literature and Hypotheses

       Despite ongoing research in fashion within many disciplines, including consumer

behavior, marketing strategy, law, sociology, and management science, no work of which we are

aware has examined empirically how styles change from season to season, or how market

feedback directly influences these changes. To the best of our knowledge, there exist no

systematic empirical investigations regarding how styles change over time. As such, no research

examines whether and how innovation in fashion depends on influences outside of the firm, such

as market feedback and competitors’ behavior. While we rely on the existing literature to inform

our work, our work departs from existing fashion theories, which have focused on cultural and

social forces as antecedents of the fashion process (Crane 1999; Miller et al. 1993; McCracken

1985; Sproles 1981; Davis 1992). Instead, we focus on the effect of the market’s response to

innovation on innovation, with innovation defined as the novelty of the styles emanating from

fashion designers each season. Surprisingly little work has looked at fashion designers at all, the

competing forces of creativity and commercialism, and how this impacts innovation within the

fashion system (Craik 2009).

Research on Fashion Trends

       Given the dearth of research on fashion in the marketing literature, it is important to

describe the process and define our terms. The fashion process evolves in stages prompted first

by the introduction of new styles or variations of existing styles. A style of clothing is a

particular configuration of an apparel item based on several design dimensions such as

silhouette, line, hem, length, color, fabric, and so forth (Davis 1992; Cappetta et al. 2006).

Fashion designers create many of the new styles and the market either accepts them or rejects

them. For a particular style to be in fashion is to say that a particular combination of style

attributes is viewed positively by a particular reference group (Wallendorf 1980). To be “in

fashion” simply means that something is now more attractive than it was previously (Lieberson

2000). Sooner or later, styles that were in fashion are rejected and new styles are introduced to

take their place.

        In high fashion, designers introduce new styles each season on the catwalks during

fashion week to retail buyers and fashion leaders. It is often thought that customers may be

frightened by styles that are too different from what they currently wear, but may also reject

anything that looks old-fashioned or out of fashion. Consequently, designers are seen as

balancing novelty with themes that are reworked or reinterpreted each year. With continuity,

fashion trends emerge, and often a particular design house becomes connected with a “look” as

designers become associated with the aesthetic manifestations of their vision (e.g., Vivienne

Westwood’s rebellious punk and subculture style). The fashion process continues as new styles

are introduced every season to replace old styles.

        As mentioned earlier, a significant amount of work has concerned itself with fashion

flows, or where new trends come from and how they spread (King 1963; Sproles 1981; Davis

1992; Miller, McIntyre and Mantrala 1993). The classical model is exemplified by Simmel’s

top-down or “trickle down” theory ([1904] 1957). In the top-down model, the innovators are

fashion designers and new styles are first adopted by upper-class elites (the aristocracy, cultural

arbiters such as celebrities) who wear the latest styles that gradually percolate down to middle

and working classes (Crane 1999). Once the styles are popularized, there is an impetus for

designers to abandon that look and create a new look to allow higher-status groups to once again

differentiate themselves from their inferiors, and in this way the fashion cycle is perpetuated. The

social processes underlying this model are differentiation, social contagion and imitation

(McCracken 1985, p. 39).

       An alternative is the bottom-up model whereby styles emerge from subcultures, typically

lower-status groups who possess distinctive styles that attract attention and imitation among

different groups (Polhemus 1994; 2007). In the bottom-up model, new styles generally emerge

from urban areas that are hotbeds for other types of innovation such as music and art. For an

emerging style to become fashionable for the masses, it ultimately needs to be endorsed by a

designer by being introduced in a collection. A trickle-up phenomenon occurs when a designer

borrows an idea or motif and creates a high-fashion collection around it (e.g., designer Marc

Jacobs and other fashion houses organized a spread for West/East magazine paying homage to

the Skateboard look). Thus, designers accommodate these external intrusions into their modus

operandi and influences trickle both down and up (Crane 1999; Craik 2009). Most important for

our purposes, both models describe a succession of changes in styles over time that depend in

large part on the styles introduced to the market by fashion designers, whether initially

emanating from the design studio or the street.

Innovation Trajectories and Fashion Cycles

       There are two principal timeframes for fashion trends, long run cycles marking the

evolution of styles spanning decades and short run cycles marking the seasonal acceptance of a

single style that may last a period of months to years. Skirt lengths getting shorter typifies a short

run cycle while skirt lengths alternatively going from extremely short to extremely long and then

back to short typifies a long run cycle (Sproles 1981). Economists argue that cycles occur

because once a particular style has spread across a population, it is profitable for a new design to

emerge and render the old design obsolete (Pesendorfer 1995). While Yoganarasimhan (2012)

has documented fashion cycles for baby names that parents give their children, fashion cycles in

apparel, and in particular high fashion, have yet to be established empirically.

Previous work in fashion has argued that in the short run, styles tend to have attributes that differ

only incrementally from preceding styles (perhaps due to the need for continuity), although these

changes occur at varying rates across time, from very fast with fads to very slow with classics.

Short run changes are believed to progress in one direction until a point of excess or extreme is

reached and then cycle back (Robinson 1958). We look at the notion of incremental change

explicitly in the evolution of styles introduced by individual design houses. Because we set out to

look at how styles change in response to market feedback, we measure the extent to which styles

change across time. If designers are truly free spirits, the difference in style between any two

seasons should not be affected by the time distance between those seasons. But if designers care

about the market and don't want to scare consumers, they are more likely to introduce change

gradually. Given we expect designers to be sensitive to the market, we expect any similarity in

styles introduced across two separate seasons to decrease as the amount of time between these

two seasons increases. In other words, the longer the gap in time, the more different we’d expect

the styles to be. This leads to our first formal hypothesis.

       H1: Style evolves over time such that the difference in style between any two seasons
          increases as the amount of time between those two seasons increases

Note that when we say “season” we are really referring to a particular year (e.g., 2003), but

specifying whether we are comparing fall and winter designs or spring and summer designs.

These two are the major fashion seasons each year, with the far less common introductions in

between known as inter-seasonal collections (Resort/Cruise before spring-Summer, or Pre-Fall).

Innovation in Fashion and Learning

       In fashion-oriented products, appearance is the most strongly perceived contributor to

value. The fashion industry depends on aesthetic innovation, which occurs when novelty is

conferred on a product in terms of its visual attributes (Marzal and Esparzal 2007). A new style

of garment that replaces an existing style can be viewed as a particular kind of innovation.

Fashion designers create a new aesthetic by changing style elements that affect a product’s

appearance, and for apparel this includes elements such as materials, proportion, color,

ornamentation, shape, and size (Bloch et al. 2003). The innovation in goods and services that

primarily impacts aesthetics has been labeled “soft innovation” (Stoneman 2010). This is in

contrast to innovation based on new functions that are often considered to be conferred by

technological change. With aesthetic innovation, it can be easier for a product to be perceived as

radically different and thus displace earlier products. With few technical boundaries, it is also

easier for products to venture past the boundaries of what a market will accept. While in general

a creative product is seen as a good product (Horn and Salvendy 2009), customers tend to reject

styles that are too creative and differ too much from the styles they are currently wearing.

       In marketing, organizations that are customer oriented and generate, disseminate and

respond to market intelligence are considered market oriented. It is widely accepted that market-

oriented firms that respond to market information and apply what has been learned perform

better (Day 1994). As a market orientation “essentially involves doing something new or

different in response to market conditions, it may be viewed as a form of innovative behavior” in

its own right (Jaworski and Kohli 1993, p.56). In this sense, changes in style that integrate the

market’s response to previous styles are being innovative and this should benefit the firm. An

alternative view in the literature on innovation suggests that information on current customers’

behavior can stifle radical or discontinuous innovation (O’Connor 1998). In fashion, one might

view discontinuous innovation as the introduction of styles with relatively unprecedented style

elements. Discontinuous or radical innovation for fashion-oriented products should then hinge on

creativity and originality. In summary, fashion innovation that considers the market’s response

should tend to be more incremental, while fashion innovation that favors creativity while

disregarding the market should be more discontinuous.

       Ultimately, if fashion designers abide by their artistic instincts in pursuit of more radical,

discontinuous innovation, the relationship between market response to past styles and new styles

should be weak to nonexistent. If, however, fashion designers at top design houses are market-

oriented, new styles will be affected, at least in part, by market response to past styles. We focus

on determining whether designers follow the market in order to respond to how the market

responds to their designs. We test whether innovations in style are influenced by the relative

acceptance or rejection of past styles; what say does the market have in dictating how much

styles change and how much they change? The notion of market-oriented designers in fashion

motivates our second formal hypothesis.

       H2: Designers change styles to move further away from past styles that were reviewed
          more negatively

Hypothesis 2 states designers will distance themselves further from styles that the market likes

less. Of course, the inverse must be true as well; designers will stick more closely to designs

reviewed more positively. With our data, we cannot distinguish between the two: do designers

strive to move away from poorly received styles, stick close by styles that are well-received, or

both? Importantly, the result is the same. The more the elements of a season’s basic styles were

frowned upon in the past by the market, they more they will change.

        If designers do in fact pay attention to what the market wants, they may not only respond

to success and failure in terms of their own designs, but also in terms of how the market responds

to their peers. This is consistent with work by Narver and Slater (1990) who explore the link

between market orientation and profitability, and who include the notion of competitor

orientation in their conceptualization of market orientation. As opposed to focusing inward and

looking at how the market responds to the designer’s own styles, they may also focus more

broadly outward and look at the relative successes and failures of other designers. This is stated

more formally as hypothesis 3.

        H3: Designers change styles to move further away from other designers’ past styles that
            were reviewed more negatively

Hypothesis 3 is a strong test of designers’ market orientation or responsiveness to market

feedback. One might argue that a designer need not deliberately consider their own past

successes or failures in order for them to affect future work. But it is much more difficult to

argue a designer is aware of others’ successes and failures, but does not consciously attend to

them when considering what styles to develop in the future. Support for hypotheses 2 and 3

would suggest designers are more marketers than visionaries, and more fashion negotiators than

fashion dictators.

                              A Market Response Model of Fashion

        Our focal independent variable is market feedback to the styles presented by a design

house each season. As design houses are incredibly secretive with respect to their financials, and

there is a lack of reliable public reporting of sales or profitability, it was necessary for us to

utilize a different gauge of market response. To this end, we collected critical reviews of the

catwalk shows for each major season (spring/summer, fall/winter) for each year in the sample

(1999-2007). It is typical for fashion communication to flow from designers to fashion shows to

fashion critics to opinion leaders and subsequently on to early and then later adopters (Evans

1989). Critics descend on Milan and Paris during Fashion Week to observe what the world’s top

designers have created, and their takes on the collections inspires the clothes we wear. Evidence

from the arts, and more specifically film, suggests reviewers’ opinions do not influence how

consumers respond, but rather predict their response (Eliashberg and Shugan 1997). One of these

authors’ key findings was that critical reviews are uncorrelated with early box office results and

thus do not seem to be causal, but are correlated with cumulative box office and thus seen as

predictive. Hence, critics are less opinion leaders or influencers (Weiman 1991) and instead

serve as a fairly accurate barometer of what the market will ultimately accept. Accordingly, we

took reviews as an indicator of relative success or failure in the market and consider more

positive reviews as greater acceptance of a style by the marketplace.

                                       Empirical Analyses

       Our data collection occurred in multiple stages. First, we sought to gather information on

how designers changed their styles from year to year spanning the decade from 1999 to 2007. A

summary of the changes that occurred across seasons would become our dependent variable. We

chose to look at the behavior of fashion houses in Europe, specifically France and Italy, because

European designers are widely known for espousing creative intentions that shun

commercialism. While New York and London complete the “big four” fashion capitals of the

world, our data collection was also guided by practical concerns. We chose Paris and Milan for

their longstanding history as centers of art and fashion and because they are home to a sufficient

sample of highly prestigious and powerful fashion houses and the conglomerates that own them

(e.g., Louis Vuitton-Moët Hennessy, Pinault-Printemps-Redoute (PPR), Compagnie Financière

Richemont, the Aeffe Group, the Prada Group).

       Fashion Weeks are semiannual events where designers showcase their latest innovations

in style for the following season on the catwalk for the press and buyers to preview. This allows

time for retailers to arrange their purchases and incorporate the designers and their work into

their retail marketing. We obtained a listing of all brands that were included in the catwalk

calendars of the Camera Nazionale della Moda Italiana in Milan from 1999-2007 (fall/winter and

spring/summer). This nonprofit association coordinates the development of Italian fashion

through shows and events such as the Women’s Pret-a-porter Collections in March and October.

We did the same for the Fedération Française de la Couture et du Prêt-à-porter, the industry’s

governing body in Paris. Only those companies that put on runway shows with their seasonal

collections during Fashion Week for at least five of the nine focal years were included. This

resulted in 38 companies, with 22 from Milan and 16 from Paris (See Table 1).

       We created variables indicating whether the firm was French or Italian (Country), how

many years the firm reported being in existence (Age), whether a design house reported relying

on a single designer or a team of designers (Type_designer), and the number of employees each

company reported each year (Employees), which served as a proxy for the size of the firm. The

average number of employees working for a design house in our sample was 286, while the

median was 185. Summary statistics for these measures are reported in Table 2. We also

collected data on when designers or design teams were reported to have changed for each


Distance: The Degree to which Styles Change

       To develop a metric to gauge the extent to which each designer’s style changed from year

to year, we compared prototypical pieces that were offered commercially following each show.

How we did this warrants elaboration. Runway shows for spring/summer collections occur in

September and October. These are followed by advertisements for these collections the

following February and March (the next calendar year) with the actual products typically going

on sale to the public in April and May. Runway shows for fall/winter collections occur in

February and March, which are followed by advertisements in September and October with

products going on sale publicly simultaneously. According to experts in the fashion industry

which we consulted, advertisements during these key periods (February and March for the spring

collection, September and October for the fall collection) are indicative of the styles designers

created for that season. The photographs for these ads are usually shot immediately after the

catwalks before the collection can be influenced by market feedback. We collected a sample of

each design house’s advertisements for each year to assess the change in style for a specific

designer or design team. Sample ads that illustrate a relatively large versus small change in the

styles presented across two years are shown in Appendix 1. These ads also effectively illustrate

the continuity in “look” associated with each design house.

       The sources for our advertisements were Vogue Italia, Vogue France, Elle Italia, and

MarieClaire France (examples are included in Appendix 1). These four magazines were chosen

after consulting with industry experts because they are considered the most legitimate sources for

diffusing what designers present on the runways. Vogue Italia and Vogue France are targeted at

a more professional and sophisticated audience, whereas Elle and MarieClaire target the trendy

consumer market. All four magazines have monthly editions; experts recommended not using

Elle France because it is published weekly and its positioning is somewhat different. We

collected every ad published in these magazines for all of the design houses in our sample. Our

data therefore included information on styles introduced by 38 design houses across nine years

for two seasons derived from 5,343 advertisements.

        Like Cappetta, Cillo, and Ponti (2006), we focused on 11 primary types of garments (e.g.,

dresses, pants, etc.). Judges coded each garment on the eight or nine of 13 style elements that

were appropriate for the particular type of apparel (see Table 3). For example, tops were

evaluated on sleeve length while pants were not. Six elements were evaluated using continuous

measures (e.g., sleeve length, neckline), while seven elements were comprised of multiple

discrete measures (e.g., color, fabric). The discrete style elements (e.g., color) were coded using

dichotomous variables (1 if it was white or 0 if it was not, 1 if it was neutral or 0 if it was not).

Therefore, if three out of five ads included white, the score for white was .6 while the average

neckline was simply the average on that measure (deep to sculpture) for that designer that year.

For any single designer, the final design code (scores on these 61 measures) was the average

across all of the ads we collected for that season. Taken together, this created what we refer to as

the style genome for each of the 38 designers for each of the 18 seasons in our sample.

        We label the difference between any two style genomes (across time but either within or

across designers) as the style distance. A style distance measure reflects the relative change in

styles and was constructed by calculating the Euclidean distance in a 61-dimensional space based

on the 61 style measures using the 13 style elements for the 11 garment types. The style distance

is a single number that is indicative of the relative change in styles across time, but is always

calculated within season (fall/winter, spring/summer). Style distance serves as our dependent

measure given our focus is whether and how much styles might change in response to market

feedback. Our data allowed us to calculate 2,530 unique style distance measures; this falls short

of the expected number given that not all 38 designers placed ads every season in every year in

the magazines in our sample. The average style distance across all designers for any two years in

our sample was 2.87 with a range of 0.79 to 7.98.

Style Progression Across Time

       The average style distance across two consecutive years was 2.76 while across non-

consecutive years it was 2.9 and went as high as 3.84. These distances suggest that as the amount

of time increased, the magnitude of the difference in styles grew. The relationship between time

and a change in style is evident when looking at Table 4. These results are consistent with the

notion that styles evolve slowly over time (incremental innovation), progressing in one direction

and thus migrating away from where they have been, consistent with and in support of H1. We

test the significance of this trend more formally in our full model by including a variable

reflecting the number of years between two style genomes, or the years upon which a specific

style distance was calculated (Yearspan).

Critics’ Reviews

       We drew on published reviews in international magazines and newspapers after the

biannual fashion week shows (catwalks) organized in Milan and Paris. With the help of three

industry experts, we selected four different international news sources considered the most

respected among both buyers and consumers based on their opinions of styles introduced on the

catwalks each season. These sources were the International Herald Tribune, The New York

Times, WWD (Women’s Wear Daily), and,Vogue's official website.

       After gathering all of the articles that addressed the shows put on by our sample of design

houses, we amassed 1,814 individual reviews (µ = 2.65 per show). The reviews reflected

coverage of 92% of the shows (629 of 684); two independent judges rated the each review on a

5-point scale (1= negative to 5 = positive). In this way, we ascertained an average review score

(review_meanrating) for each of the 38 design houses for each season for the nine years of data

in our sample. As discussed earlier, the average review formed a proxy for the market’s response

to the styles introduced that year.

       Given we have data on reviews and not sales, whether and how reviews impact decisions

made by the fashion houses, as sales most certainly would, was not certain. In order to test

whether fashion houses paid attention to reviews and the weight of their impact, we ran a logistic

regression using our data on when designers or design teams were reported to have changed as

the dependent variable and the average review score as the focal independent variable. We

included a dummy variable indicating whether the fashion house relied on a single designer or

design team and the interaction. The results are shown in Table 5. The critics’ reviews

(review_meanrating) had a significant effect on whether the design house changed their designer

or design team; an increase in the reviewers’ mean rating led to a lower likelihood of change.

Neither the type of designer nor the interaction had a statistically significant effect. Thus, there

was preliminary evidence that design houses attended to critics’ reviews and that they had an

impact; the more successful a designer, the greater their job security. This also suggests

designers should be concerned with what the critics are saying about their work.

The Impact of Market Feedback on Style

        In order to test the effect of market feedback on the styles that designers introduce, we

ran a regression using GLM in SAS with the style distance measures between two seasons for

individual designers as the dependent measure and the reviewers’ critical assessment

(review_meanrating) for the earlier of the two years as the focal independent measure. We

included the number of years between two styles (Yearspan) to control for the amount of time

between when the two styles compared hit the market. We also included several other control

variables such as the number of employees (Employees), whether the season was fall/winter or

spring/summer (Season), whether the design house was based in France or Italy (Country), and

whether the fashion house was led by a single designer or a design team (Type_designer ). The

relevant interactions were also included. The results are summarized in Table 6.

        First, in further support of H1, the estimate for Yearspan was positive and significant

(0.044, p < .01) such that the more years between two seasons for a designer, the greater the

difference in styles. Empirically, we have evidence of styles progressing away from the past

across time. The trend that we observe away from what was done in the past means we do not

see strong signs of nostalgia, at least in the time period we observe (i.e., fewer than 10 years). If

some designers did revert back toward earlier designs, others must have moved further away

from the past for this result to hold.

        In support of H2, we observe a tendency for designers to move away from styles that

were reviewed less favorably in the past. The coefficient for critical reviews

(Review_meanrating) is negative and significant (-.094, p < .01). This result shows how

designers, intentionally or unintentionally, stick closer to styles that have been well-received in

the past and shy away from those not as well-liked. This is the first empirical demonstration

showing that, while designers distance themselves further from previous styles over time, exactly

how far they move is moderated by market forces (how the market responded to specific styles in

the past). This result dispels the long-held myth of the designer as dictator unfettered by how the

market responds to his or her creations.

       With respect to the other variables in our model, a number of other interesting findings

emerge. On average, designers from French firms changed their styles to a greater degree year-

over-year than designers from Italian firms. This suggests French designers were relatively more

innovative while Italian designers were more conservative. Given Paris remains the center of the

fashion system of modernity (Rocamora 2006), the willingness to try something new may be

partly responsible for, as well as partly a result of, the French hegemony of fashion.

       We also observe larger companies (Employees) changed their styles to a lesser extent

than smaller firms. It may be that smaller firms believe they need to introduce more radical style

changes to make a statement and garner attention away from their larger and better known

competitors. Or it may be that the larger, more iconic fashion houses possess a specific aesthetic

to which they need to adhere to more closely (i.e. more continuity). Consistent with this idea, the

interaction between Employees and Review_meanrating was positive and significant such that

big companies appear less sensitive to reviews (they distanced themselves less from styles that

received negative reviews). This result may be because larger companies are less likely to learn

from market feedback or because they believe that despite the reviews, their styles shape the

market (i.e., they possess more power). It is possible that bigger design houses consider

themselves responsible for creating fashion, and thus criticism of larger firms doesn’t carry the

same weight. In other words, these designers are the creative progenitors and customers will

eventually see the light. Finally, we also observe firms with single designers (Type_designer)

introduce styles that differ more from year to year, on average, but they are also more sensitive

to criticism (Review_meanrating*Type_designer). This suggests that firms with a creative lead

or single designer are more daring, but also are more likely to move away from those styles that

do not sit well with reviewers.

       In order to test the effect of critics’ response to competitors’ styles on the degree to which

designers change their styles we ran a regression with a much broader set of style distance

measures. The style distances utilized included every designer’s collection for a particular season

for a particular year and compared it to all other designers’ styles in prior years for that season

(fall/winter or spring/summer). For example, the distance between Givenchy’s styles in 2003, as

the firm introducing the styles, and Chanel’s styles in 2002, were hypothesized in H3 to depend

on how Chanel’s styles were received in 2002 by reviewers. We therefore included two measures

of firm size, one for the design house introducing the style (Employees2, or Givenchy in this

example) and one for the size of the firm upon which the review was made (Employees1, or

Chanel). We also included the same control variables as in our previous model as well as all of

the relevant interactions. Thus our regression included more distance measures as we compared

across designers, and styles were presumed to depend on what other designers did and how the

market responded. The results are summarized in Table 7.

       First and foremost, in support of H3, we observe a tendency for designers to move away

from styles introduced by other designers’ that were reviewed less favorably in the past. The

coefficient for critical reviews (Review_meanrating) is negative and significant (-1.41, p < .01).

This suggests that not only do designers’ attend to how the market responds to the designs they

introduced previously, but they are keenly aware of, and respond to, what the market says about

other designers’ previous styles. Taken together, these results provide the first empirical

evidence that styles evolve based on market forces as well as creative forces, documenting the

dynamic interaction between fashion houses and their customers in a recurring feedback loop.

       Also, a second result emerges that adds depth and sheds further light on these results and

how the industry views market feedback. The coefficient pertaining to the size of the design

house introducing the style (Employees2) was not significant, implying no difference between

big firms and little firms with respect to the degree to which they attend to market feedback.

Given design houses pay attention to how the market responds to competitors’ designs, it is

interesting to note that larger design houses are not inherently less attentive to this information

than smaller design houses. Both seem to learn from how the market responds and change their

styles accordingly. However, we do observe a significant coefficient for the size of the design

house upon which the review was based (Employees1). The coefficient for Employees1 is

negative, which suggests designers at both small and big design houses do not shy as far away

from negative reviews of large design houses as much as they do for small design houses. Again,

what big design houses do seems less sensitive to market criticism.

       We also observe design houses are less likely to shy away from styles introduced by

other design houses with a single designer (Type_designer1). The interaction between the type of

designer and the average rating suggests that designers are also less inclined to distance

themselves from design houses with a single designer when the latter’s styles are reviewed less

favorably. Across designers, we also find larger style changes in spring/summer than winter/fall

(Season) as well as larger style changes among French design houses as compared to Italian ones

(Country). These results are consistent with those obtained when we compared style changes

within designers (i.e., looked at only distances calculated between the styles of a single design



          The idea of considering consumers or what the market wants when making creative

decisions is often seen as anathema to those engaged in artistic pursuits. Consider the famous

work by Komer and Melamid (1997), who attempted to discover what ordinary people wanted to

see in art by surveying the public. The results were paintings that no one person would want,

raising the question of whether artists, who know a lot about art, are better equipped at

determining what types of paintings are visually appealing. These authors worked with David

Soldier composer David Soldier to illustrate the same effect for music. Artists like to believe that

they set the style and eventually the market will catch on. But clearly not everything they do will

catch on, and no work before ours has looked at iterative changes in art that provide the artist

market feedback.

          Our data and our results suggest that fashion designer, as commercial artists, are not only

sensitive to their own past successes and failures when deciding what new designs to introduce,

but shows they also consider competitors’ past work and how those styles have fared in the

marketplace. Ours is the first empirical evidence of market feedback impacting new product

introductions in an industry based on aesthetic innovation. It would be interesting if the same

type of tests could be done for music and art. For example, one could explore whether a band’s

new songs tend to sound more like songs that sold well previously, or whether artists veer

towards works that are well received by art critics. The idea of watching what works is well

known in Hollywood where movies that benefit from market feedback (e.g., adaptation of books,

sequels, advanced screenings) are known to enjoy box office success, while movies created

based on the director’s creative vision alone (auteur theory) are frequently dismal failures.

       Our findings contribute to the literature on fashion in multiple ways. First, we

demonstrate how styles evolve, progressing further away from what has been done in the past as

time progresses, at least in the short run (nine years). Empirical evidence of style changes of any

sort are conspicuously absent in the literature. But we find nine years is too short a time in which

to observe fashion cycles of any significance, at least in high fashion. Second, we show that the

designer as dictator theory of fashion innovation just doesn’t hold true, at least not in a strict

sense. While many fashion theorists have argued for a bottom-up approach based on anecdotal

evidence, they are referring to the substance of those styles that eventually find their way onto

the catwalk. While we do not claim to know or show where designers get their inspiration from,

our work reveals how designers are sensitive to the market; their styles stay closer to what has

worked and further from hasn’t.

       Marketing managers in creative industries should take note of how designers integrate

market feedback and how aesthetic innovation occurs incrementally based on what the market is

willing to accept. In industries where technological innovation drives new product introductions

and functional requirements matter most, whether there is incremental or disruptive change may

depend on the firm’s capabilities. In fashion, change can be implemented much more easily, as

there are fewer operational constraints. Yet what is clearly an impetus to how much innovation

occurs is how readily the market will accept the change. Buyers could look at how designers

respond to past successes and failures to see which houses are market oriented. Different retailers

may want different things; perhaps Macy’s wants a mix of designers who consider the market,

but also wants more avant-garde designers who symbolize novelty and fashion forwardness.

       Our work has its limitations. In our model, a designer’s style, while based on 61

measures, did not include all possible style variables such as those pertaining to specific colors or

the silhouettes favored by a designer. A more comprehensive measure is certainly still possible.

Yet we believe our measure is detailed enough to provide a robust assessment of the extent to

which styles change over time and to provide a meaningful indicator of how styles evolve. The

use of critical reviews as a measure of market response is admittedly only a proxy, but a credible

one for what should eventually succeed. Future research utilizing more direct measures such as

sales, profits or a similar index of market success would surely be welcomed.

       There is plenty of room for future research that explores how styles evolve systematically

over time, especially empirical studies. The paradox of fashion is the conflict between looking

distinctive while giving the impression of a certain degree of uniformity. Understanding how

aesthetic innovation occurs, and what drives its acceptance or rejection is an important area that

should garner a lot of attention in the future.


Bloch, Peter H., Frédéric F. Brunel and Todd J. Arnold (2003), “Individual Differences in the
       Centrality of Visual Product Aesthetics: Concept and Measurement,” Journal of
       Consumer Research, 29, 551-565.

Cappetta, Rossella, Paola Cillo and A. Ponti (2006), “Convergent Design in Fine Fashion: AN
       Evolutionary Model for Stylistic Innovation,” Research Policy, 35, 1276-1290.
Chiles, Adrian (2002), “Inside Chanel” video teaching resource from the BBC. Available at
Craik, Jennifer (2009), “Fashion: The Key Concepts,” Oxford: Berg, Great Britain.
Crane, Diana (1999), “Diffusion Models and Fashion: A Reassessment,” Annals of the Academy
       of Political and Social Science, 566 (November), 13-24.
Belfanti, Carlo and Fabio Giusberti (2000), “Clothing and Social Inequality in Early Modern
       Europe: Introductory Remarks,” Continuity and Change, 15(3) 359-365.
Davis, Fred (1992), Fashion, Culture and Identity. Chicago: Chicago University Press.
Day, George S. (1994), “The Capabilities of Market-Driven Organizations,” Journal of
      Marketing, 58(4), 37-52.
Eliashberg, Joshua and Stephen Shugan (1997), “Film Critics: Influencer or Predictors?” Journal
       of Marketing, 61, 68-78.
Evans, Martin (1989), “Consumer Behavior towards Fashion,” European Journal of Marketing,
       23(7), 7-16.
Freudenberger, Herman (1968), “Fashion, Sumptuary Laws, and Business,” The Business
      History Review, 37 (1/2, Summer), 37-48.
Horn, Diana and Gavriel Salvendy (2009), “Measuring Consumer Perception of Product
       Creativity: Impact on Satisfaction and Purchasability,” Human Factors and Ergonomics
       in Manufacturing, 19(3) 223-240.
Jaworksi, Baernard J. and Ajay K. Kohli (1993), “Market Orientation: Antecedents and
      Consequences,” Journal of Marketing, 57(3), 53-70.
King, Charles W. (1963), “Fashion Adoption: A Rebuttal to the ‘Trickle Down’ Theory,” in
       Toward Scientific Marketing, Stephen A. Greyser, ed., Chicago: American Marketing
       Association, 108-125.
Komer, Vitaly and Alexander Melamid (1997), “Painting by Numbers: Komar and Melamid’s
      Scientific Guide to Art,” Berkeley, CA: University of California Press.
Kroeber, Alfred (1919), “On the Principle of Order in Civilization as Exemplified by Changes in
      Fashion,” American Anthropologist, 21, 235-63.

Lieberson, Stanley (2000), “A Matter of Taste: How Names, Fashions, and Culture Change,”
       Yale University Press, New Haven CT.
Marzal, A. and E.T. Esparza, (2007), “Innovation Assessment in Traditional Industries: A
       Proposal of Aesthetic Innovation Indicators,” Scientometrics, 72(1), 33–57.

McCracken, Grant D. (1985), “The Trickle-Down Theory Rehabilitated.” In The Psychology of
      Fashion, ed. Michael R. Solomon. Lexington, MA: D.C. Hath, Lexington Books.

Miller, Christopher, M., Shelby H. McIntyre and Murali K. Mantrala (1993), “Toward
        Formalizing Fashion Theory,” Journal of Marketing Research, 30 (May), 142-157.

Narver, John C. and Stanley F. Slater (1990), “The Effect of a Market Orientation on Business
       Profitability,” Journal of Marketing, 54(4), 20-35.

O’Connor, Gina Colarelli (1998), “Market Learning and Radical Innovation: A Cross Case
      Comparison of Eight Radical Innovation Projects,” The Journal of Product Innovation
      Management’s special issue on Really New Products, vol. 15 (March), no. 2, pp. 151-

Pesendorfer, Wolfgang, (1995), “Design Innovation and Fashion Cycles,” American Economic
      Review, 85, 771-792.

Polhemus, Ted (1994), Street Style: From Sidewalk to Catwalk, London: Thames and Hudson.

Polhemus, Ted (2007), “Trickle Down, Bubble Up,” In The Fashion Reader, ed. L. Welters and
      A. Lillethun, 327-31, Oxford: Berg.

Robinson, Dwight E. (1958), “Fashion Theory and Product Design,” Harvard Business Review,
      36 (November-December), 126-138.
Robinson, Dwight E. (1963), “The Importance of Fashions in Taste to Business History: An
      Introductory Essay,” The Business History Review, 37(1/2 Spring-Summer), 5-36.
Rocamora,, Agnès (2006), “Paris, Capiale de la Mode: Representing the Fashion City in the
     Media.” In Fashion World Cities, ed. C. Breward and D. Gilbert 43-54. Oxford: Berg.
Simmel, Georg (1904), “Fashion.” Reprint. American Journal of Sociology, 62 (1957): 541-58.
Sproles, George B. (1981) “Analyzing Fashion Life Cycles – Principles and Perspectives,”
       Journal of Marketing, 45 (Fall), 116-124.
Stoneman, Paul (2010), “Soft Innovation: Economics, Product Aesthetics, and the Creative
      Industries,” Oxford University Press Inc., Oxford UK.
Wallendorf, Melanie (1980), “The Formation of Aesthetic Criteria through Social Structures and
      Social Institutions,” in Advances in Consumer Research Volume 7, eds. Jerry C. Olson,
      Ann Arbor: Association for Consumer Research, 3-6.

Weiman, Gabriel (1991), “The Influential: Back to the Concept of Opinion Leaders,” Public
     Opinion Quarterly, 55 (Summer), 267-79.

Yoganarasimhan, Hema (2012), “Cloak or Flaunt? - The Fashion Dilemma,” forthcoming at
      Marketing Science.

Yoganarasimhan, Hema (2012), “Identifying the Presence and Cause of Fashion Cycles in the
      Choice of Given Names,” UC Davis working paper.

                                         Table 1

    Fashion Houses in Paris and Milan (sponsoring at least five Catwalks from 1999-2007)

    French Brands          #Ads          Italian Brands              #Ads
    1. Balenciaga           40           1. Alberta Ferretti         197
    2. Celine               55           2. Blumarine                379
    3. Chanel               90           3. Clips                     50
    4. Chloé               106           4. Dolce & Gabbana          778
    5. Christian Dior       90           5. Etro                     117
    6. Givenchy             43           6. Fendi                    171
    7. Hermes               51           7. Gianfranco Ferre         179
    8. Jean Paul Gautier    50           8. Gucci                    208
    9. Kenzo                34           9. Iceberg                  100
    10. Lagerfeld           33           10. Jil Sander              135
    11. Lanvin              43           11. La Perla                 91
    12. Louis Vuitton       83           12. Mariella Burani          97
    13. Sonia Rykiel        35           13. Missoni                 139
    14. Ungaro              18           14. Miu Miu                 195
    15. Vivienne Westwood 13             15. Moschino                213
    16. Yves Saint Laurent  96           16. Philosophy              118
                                         17. Prada                   296
                                         18. Roberto Cavalli         343
                                         19. Rocco Barocco           132
                                         20. SportMax                142
                                         21. Valentino               106
                                         22. Versace                 254

                                         Table 2

Summary Statistics of Independent Variables

     Variable        N      Mean      Std. Dev.        Minimum               Maximum
Country             684      1.5789     0.4941     1 (France)        2 (Italy)
Season              684      1.5000     0.5004     1 (Fall Winter)   2 (Spring Summer)
Type_designer       640      0.4063     0.4915     0 (Single)        1 (Team)
Designer_change     638      0.9514     0.2152     0 (Fired)         1 (Retained)
Age_firm            640     35.6063    22.7901     1                 118
Employees           500    286.0440   343.6333     4                 2415
Review_meanrating   629      3.4945     0.7683     1                 5
N_reviews           684      2.8885     1.0367     1                 4

                                                      Table 3

                                        Elements of the Style Genome

                           Table 2a: Continuous Variable Coding (By Garment Type)

Garment       Type of Fit         Waistline Neckline                Sleeve              Top                Bottom
Top           slim to oversize                  deep to sculpture   strapless to long   v. short to long
Shirt         slim to oversize                  deep to sculpture   strapless to long                      short to v. long
Caban         slim to oversize                  deep to sculpture   strapless to long                      short to v. long
Sweater       slim to oversize                  deep to sculpture   strapless to long   v. short to long
Skirt         slim to oversize    low to high                                                              short to v. long
Dress         slim to oversize                  deep to sculpture   strapless to long                      short to v. long
Jacket        slim to oversize                  deep to sculpture   strapless to long                      short to v. long
Coat          slim to oversize                  deep to sculpture   strapless to long                      short to v. long
Trench        slim to oversize                  deep to sculpture   strapless to long                      short to v. long
Pants         slim to oversize    low to high                                                              short to v. long
Bluson        slim to oversize                  deep to sculpture   strapless to long                      short to v. long

                      Table 2b: Discrete Variable Coding (For All Garments)

    Color       Fabric               Fabric        Fabric                  Pattern           Sleeve           Sleeve
               (Material)         (Application) (Processing)                                 (Cut)           (Design)

White          Cotton             Feathers           Knitwear           Stripes           Empire           Wide
Neutral        Linen              Bejeweled          Wrinkled           Plaid             Trapeze          One-strap
Pastel         Silk               Fringe             Ripped             Floral            Drape            Bat
Bright         Wool               Flowers            Pleated            Graphics
Dark           Denim              Fur                Decorated          Animal
Black          Leather            Pockets                               Camouflage
Bi-Color       Sequin             Embroidery                            Polka Dots
Fluorescent    Lace                                                     Optical
Multi-color    Chiffon                                                  Patchwork
Metallic       Synthetic                                                Paisley
               Lamè                                                     Shades
               Jersey                                                   Tweed

                                 Table 4

    Average Style Distance as a Function of Year Span

    Yearspan        N           Average            Min     Max 
        1          543           2.7624           0.9663   6.8784
        2          476           2.7609           0.7954   7.0985
        3          407           2.8058           0.8872   7.9373
        4          342           2.8156           0.8844   6.5596
        5          282           2.9946           1.0436   6.5574
        6          215           2.9465           1.1505   6.8475
        7          150           3.0878           1.2693   6.6437
        8           85           3.4764           1.6867   6.4420
        9           30           3.8406           2.1213   6.8557

                                            Table 5

Logistic Regression Analysis of Maximum Likelihood Estimates (N = 525)

          Variable          DF   Estimate   Std. Dev.   Wald Chi-Square      Pr > ChiSq
          Intercept         1     -1.1682     0.4941    1.8351            0.1755
     Review_meanrating      1     -0.5280     0.5004    4.0997            0.0429
        type_designer       1     -0.8926     0.8624    1.0712            0.3007
    Review_me*type_design   1     0.2821      0.2608    1.1700            0.2794

                                        Table 6

Within Designer Regression Parameter Estimates and Type III SS Statistics

                Parameter                     Estimate     F-Value          Pr > F
Intercept                                       2.7335
Yearspan                                        0.0440       10.17          0.002
Season (Fall Winter)                           -0.4359        2.75          0.10
Country (France)                                0.8446        5.60          0.02
Review_meanrating                              -0.0942       12.20          0.001
Review_meanrating*Country (France)              0.0247        0.06          0.81
Employees                                      -0.0014       10.75          0.001
Review_meanrating*Employees                     0.0005       13.40          0.001
Type_designer (Single)                          0.7622        4.23          0.04
Review_meanrating*Type_designer (Single)       -0.2091        3.87          0.05

                                        Table 7

Across Designers Regression Parameter Estimates and Type III SS Statistics

                 Parameter                         Estimate       F-Value    Pr > F
Intercept                                             3.1960
Yearspan                                              0.0041        1.80      0.18
Season (Fall Winter)                                 -0.5673      112.97     <.0001
Review_meanrating                                    -0.1410       44.67     <.0001
Review_meanrating*Season (Fall Winter)                0.1049       48.92     <.0001
Country1 (France)                                     0.5129       53.53     <.0001
Review_meanrating*Country1 (France)                   0.0378        3.47      0.06
Employees1                                           -0.0007       53.78     <.0001
Review_meanrating*Employees1                          0.0002       87.62     <.0001
Type_designer1 (Single)                              -0.0338        0.22      0.64
Review_meanrating*Type_designer1 (Single)            -0.0368        3.12      0.08
Country2 (France)                                     0.2852       18.91     <.0001
Review_meanrating*Country2 (France)                   0.0211        1.32      0.25
Employees2                                          -0.00001        0.03      0.86
Review_meanrating*Employees2                       -0.000003        0.03      0.85
Type_designer2 (Single)                               0.1521        5.00      0.03
Review_meanrating*Type_designer2 (Single)            -0.0146        0.58      0.45


                     APPENDIX 1

    Miu Miu, Autumn/ Winter 200, ELLE_Italia (Larger Change)

    Miu Miu, Autumn/ Winter 2002, ELLE_Italia ((Larger Change)

    Versace, Spring/ Summer 2004, ELLE_Italia (Smaller Change)

    Versace, Spring/ Summer 2005, ELLE_Italia (Smaller Change)

Shared By:
Description: Iconic fashion designers see themselves as artistic visionaries whose creations are impervious to market critique. Using real-world data, this work refutes this notion, and is the first empirical study to document a market response model of fashion. We show empirically how elite European fashion houses introduce new styles based on how previous styles have been received by the market. More specifically, we find designers introduce styles that are more similar to styles that were successful in the past and less similar to styles that were less well-received. Further, the styles a designer introduces are shown to depend on the market’s response to competing designers’ styles. In showing how styles evolve systematically over time, this work illustrates the systematic integration of market feedback on new product introductions, doing so in an industry based on aesthetic innovation.