Journal of Marketing Research Scholarly Insights are produced in partnership with the AMA Doctoral Students SIG – a shared interest network for Marketing PhD students across the world.
The global fashion market has achieved unprecedented growth, being valued at 1.74 trillion USD as of 2023, and it’s expected to expand further to 1.94 trillion USD by 2027 (Statista 2023). In the midst of reaching new heights, two colossal forces of the industry—high-end luxury labels and fast-fashion giants—remain in conflict. As soon as the former spurs a novel trend, the latter produces a copycat almost immediately. Despite the longstanding and growing tension between the two fashion titans, a critical question remains: do these fast-fashion copycats truly threaten the reign of high-end brands, or could they unexpectedly fuel their popularity?
In a recent Journal of Marketing Research study, researchers Zijun (June) Shi, Xiao Liu, Dokyun Lee, and Kannan Srinivasan discuss the dynamics between traditional high-end fashion brands and fast-fashion brands like Zara and H&M, with a focus on the impact of copycats on the high-end market. Despite the growing interest in and concerns about copycats, there has been limited empirical investigation into the subject due to challenges in quantifying fashion and obtaining consumer-level fashion choice data.
The authors overcome these challenges by incorporating deep learning techniques to measure the compatibility and distinctiveness of fashion items and analyzing image-based user-generated content from a fashion-focused online community. Driven from a rich dataset comprising 10,262 users and 64,681 fashion posts over four years, the authors find that fast-fashion copycats may negatively and positively affect high-end brands. Although copycats may harm high-end brands to some extent (a cannibalization effect), they can also be helpful for growth (a market expansion effect). Specifically, the market expansion effect is driven via both static and dynamic mechanisms: mix-and-match styling (a static mechanism), value enhancement via increased popularity on social media, and cost reduction via sponsorship opportunities for popular users (dynamic mechanisms).
Although copycats may harm high-end brands to some extent (a cannibalization effect), they can also be helpful for growth (a market expansion effect).
Furthermore, using a dynamic structural model incorporating an additional data set of 1,380 original–copycat pair images, the authors show that the impact of prohibiting copycats depends on the threshold of similarity. If only extremely similar copycats are banned, high-end brands benefit, as cannibalization seemingly dominates the market expansion effect among remarkably similar copycats. However, if moderately similar items are also banned, high-end brands suffer, as the market expansion effect seemingly dominates the cannibalization effect among moderately similar fast-fashion items. These findings provide insights for managers and policymakers in addressing the magnitude and limitations of copycats’ impact on high-end brands.
We interviewed the authors to discuss further actionable implications, generalizability to other categories, challenges in methodology, and much more:
Q: Based on the trickle-down theory of trend adoption (Simmel 1904), a central role of fast-fashion brands may be the affordable replication of higher-end brands. Considering the fundamental role of imitation in fast fashion, how do you anticipate copyright policies being enforced in this context? Has there been policy implementation addressing the pertinent issue? If public policy is falling short in addressing the imitation problem, what approaches should managers of luxury brands employ to confront the obstacles and potential advantages posed by fast-fashion copycats?
A: Currently the copyright policy does not protect fashion designs. From the managers’ perspective, our research provides helpful implications for luxury brands. That is, copycats would benefit the luxury items if the similarity is not very high. However, the cannibalization effect may dominate if the copycats are too similar to the original design. Therefore, instead of trying to eliminate all copycats, one approach the premium brands can take is to make it difficult for fast fashion to copy their product design to maintain a low similarity.
Q: Do you view your findings to be applicable to other retail categories like consumer electronics, where lower-end brands often replicate designs of higher end brands? If not, what is unique about fashion products that set it apart?
A: It requires additional empirical study to evaluate the effect in other retail categories. Fashion products are unique from several perspectives. First, visual features are key product characteristics, which we quantify and use to gauge the similarity between the original and copycat products in our study. Second, consumers use fashion products differently from other products in the sense that they need to mix and match multiple items to form a complete ensemble. This mix-and-match behavior is also a key mechanism that explains our findings of fast-fashion copycats’ impact on high-end brands.
Q: You used state-of-art machine learning techniques to quantify user-generated visual content. What were some challenges you faced and addressed in quantifying the subjective nature of fashion? Were there alternative ways you have considered that may have yielded different insights?
A: A big challenge was quantifying the compatibility and similarity between two product items. While similarity is relatively more straightforward to measure because we can use many existing tools from the computer vision literature, compatibility is much more challenging. We initially tried transferred learning by collecting human-labeled data and consumer co-purchase data to fine-tune the deep neural network, which unfortunately cannot fully avoid the issue of subject representativeness. The results could be biased if the measurement error is non-negligible. Later we decided to switch to applying a more cutting-edge model trained on similar tasks (i.e., rating compatibility of clothing items).
Q: This study observes data from 2013 to 2017 from a fashion-focused social media platform that has now become obsolete. Since then, the luxury market has grown significantly, and the dynamics of social media have become more diversified. If one were to conduct a similar study in the current market, what additional factors would you consider to reflect the changes? How would you conduct the study beyond social media users to overcome limited generalizability and self-selection bias?
A: The methods, models, and mechanisms we proposed in the paper are generalizable. To understand the copycat effects in more recent markets or beyond social media, one only needs to apply our method to more recently collected datasets or those collected beyond social media.
Q: Luxury brands are increasingly collaborating with fast fashion brands (e.g. H&M x Mugler, Target x Missoni). How do you see these collaborations fitting into your findings? Do you view this trend as a potential solution to the copycat issue or a risk to luxury brands?
A: This could be a valid solution. Alternatively, as discussed in the paper, high-end brands could enable consumers to mix and match items from different price tiers by launching affordable versions of their own high-end styles, capturing the unserved demand rather than leaving it to fast-fashion brands. That said, it is worth noting that both approaches may erode the high-end brand’s value. However, the net effect is unclear without additional information and further empirical study.
References
Simmel, Georg (1904), “Fashion,” International Quarterley, 10, 130-55.
Statista (2023), “Revenue of the apparel market worldwide from 2014 to 2027,” (accessed December 2023), https://www.statista.com/forecasts/821415/value-of-the-global-apparel-market.
Read the Full Study for Complete Details
Read the full article:
Zijun (June) Shi, Xiao Liu, Dokyun Lee, and Kannan Srinivasan, “How Do Fast-Fashion Copycats Affect the Popularity of Premium Brands? Evidence from Social Media,” Journal of Marketing Research, 60 (6), 1027–1051.
Go to the Journal of Marketing Research