I'm preparing for a lunch meeting where my Virginia faculty colleagues are helping me to "incubate" a new paper idea, so at this point I can respond only briefly to Amy's very insightful post. A couple of points:
(1) Amy suggests that if R&D in the fashion industry is cheap, and copying relatively expensive, then perhaps non-IP factors such as first-mover advantage, or perhaps the status value of trademarks, explain why we have innovation in the fashion industry without copyright on designs. We've thought about both these possibilities, and we think they play a role, but they don't tell nearly the entire story.
First-mover advantage. Innovation in the fashion industry is relatively cheap, in that it requires less investment than innovation in, say, pharmaceuticals. Copying is, however, also very cheap. And, perhaps most importantly, the effect of first-mover advantage is limited, in our view, by the fact that copying occurs very rapidly. So any period of de facto exclusivity is likely to be very short.
Trademark. We agree with Amy that fashion trademarks are valuable for the signals they send to both public and buyer. And to the extent that a trademark is valued for the bump it gives to a buyer's self-perception, it is a valuable form of IP without regard to whether it is visible to others. That said, given the ubiquity of copying, there is clearly money to be made in differently-trademarked design copies. In a higher-IP regime, a design's originator would own that derivative market. She either could exploit it though licensing, or via proliferation of her own marks (e.g., bridge lines). It's also true that owners of very valuable marks "reference" one another. So the St. John 4-pocket jacket is essentially a variation on the iconic Chanel 4-pocket jacket bundled with the St. John trademark, which, like the Chanel trademark, is a valuable status-conferring mark. The Chanel and St. John jackets have somewhat different constituencies, but again, in a higher-IP regime Chanel could own both.
If you buy the orthodox IP incentives story generally, but you worry about overprotection as much as underprotection, your mantra is "enough, but not too much". At the moment, this mantra finds adherents mostly within academia, and occasionally within the technology and biomedical/pharmaceutical industries (where people worry about the culture of scientific research and patent thickets). Hollywood, for the most part, sees every innovation problem as demanding an IP solution. Up to now, the fashion industry has behaved unlike Hollywood. More IP is not, in this industry's view, always better. Why? We think because their particular innovation model, aided at the margins by a small first-mover dynamic and the value of status-conferring marks, is best accommodated by a rapid fashion cycle facilitated by relative freedom to copy.
(2) Demsetz and the cost of a high-IP rule for fashion. Amy suggests that we might not see a high-IP rule in fashion because the *cost* of such a rule would outweigh its benefits to firms in the fashion industry, at least relative to the existing low-IP rule and the induced obsolescence/anchoring dynamic that rule facilitates. We agree entirely, but would note that this observation reflects both on the cost of a high-IP rule, and the relative efficiency of a low-IP rule for this particular industry and its particular innovation practice. The point here, and one important point of our paper generally, is that although different industries manifest different innovation cultures, practices, incentives, etc., IP law for the most part proceeds as if all industries are the same. The copyright rules that cover the 10-cent label on the shampoo bottle are about equal to those governing the 25-million-dollar motion picture. This industry-insensitivity of IP rules suggests, in turn, that IP proceeds from a general theory of innovation: (a) that innovation is costly relative to copying, (b) that the copyist will be able to compete away the innovator's returns, and (c) that absent a right to exclude, innovation will suffer.
The fashion industry challenges this model. The challenge arises, in part, because imposition of IP rules on this industry would be expensive. But IP rules are always expensive. The relevant question is "expensive relative to what?" In fashion, IP rules are expensive relative to the low-IP dynamic already in place.