Patents and Vertical Integration: A Revised Theory of the Firm

I'm a big fan of Peter Lee's work, and I'm a big fan of theory of the firm work. Imagine my joy upon seeing Prof. Lee's new article, forthcoming in Stanford Law Review, called: Innovation and the Firm: A New Synthesis. This article is a really thoughtful, really thorough re-examination of patents and the firm. The abstract is here:

Recent scholarship highlights the prevalence of vertical disintegration in high-technology industries, wherein specialized entities along a value chain transfer knowledge-intensive assets between them. Patents play a critical role in this process by lowering the cost of technology transactions between upstream and downstream parties, thus promoting vertical disintegration. This Article, however, challenges this prevailing narrative by arguing that vertical integration pervades patent-intensive fields. In biopharmaceuticals, agricultural biotechnology, information technology, and even university-industry technology transfer, firms are increasingly absorbing upstream and downstream technology providers rather than simply licensing their patents.

 This Article explains this counterintuitive development by retheorizing the relationship between innovation and the firm. Synthesizing previously disconnected lines of theory, it first argues that the challenge of aggregating tacit technical knowledge — which patents do not disclose — leads high-tech companies to vertically integrate rather than simply rely on licenses to transfer technology. Relatedly, the desire to obtain not just discrete technological assets but also innovative capacity, in the form of talented engineers and scientists, also motivates vertical integration. Due to the socially embedded nature of tacit knowledge and innovative capacity, firms frequently absorb entire pre-existing organizations and grant them significant autonomy, an underappreciated phenomenon this Article describes as “semi-integration.” Finally, strategic imperatives to achieve rapid scale and scope also lead firms to integrate with other entities rather than simply license their patents. The result, contrary to theory, is a resurgence of vertical integration in patent-intensive fields. The Article concludes by evaluating the costs and benefits of vertically integrated innovative industries, suggesting private and public mechanisms for improving integration and tempering its excesses.

The abstract does a pretty complete job of explaining the thesis and arguments here, so I'll make a few comments after the jump.

This article begins with a fairly thorough literature review, beginning with Coase's work and discussing economic and legal scholarship through today, including some of my favorite articles. This is a good article to read if you simply want to get a handle on the background in the area, and figure out where you should start.

Much of the literature highlights the benefits of vertical disintegration (I like non-integration - it is not like these firms are falling apart), namely obtaining information and technology not within the firm's skillset. But, if that's true, why do we see so much vertical integration in patent intensive fields?

The answer starts with tacit knowledge - the notion that patents do not disclose information inside the heads of inventors, and the only way to get that is through integration. The withholding of information can be deliberate (as I discuss in my book chapter on trade secret development incentives) or incidental, but either way, patent licenses and purchases are incomplete. Indeed, I have argued that portfolio licenses can be so devoid of information that they should be treated as securities, not information transfers. The article addresses this briefly, noting that patent purchases are often defensive, while firm purchases are often technological. As a result, firms are more likely to vertically integrate in the presence of tacit knowledge; this account builds on prior theory of the firm work in this area.

One nice feature of this article is that it sets up a testable hypothesis: where patents contain all the relevant information then we should be more likely to see licenses, and where they do not then we should be more likely to see integration. Knowing the answer to this question might, for example, help identify strategic behavior in which firms refuse to license patents but the patents nevertheless disclose sufficient knowledge for transfer. It might also aid in evaluating acquisitions in which the purchased product is abandoned and/or the employees are not used. Tacit knowledge stories are harder to make in these cases, and the theory helps explain why.

I look forward to seeing future responses to this important work.