Patent holdup occurs when a patent holder extracts higher royalties ex post than it could have negotiated ex ante, where the difference is not explained by an increase in the technology’s value. To date, the literature principally has focused on—indeed, sometimes conflated—two potential sources of holdup: the sunk costs the user has incurred ex ante to adopt the technology, and the “switching costs” of adopting an alternative ex post.
This paper holds that the common source of holdup is neither sunk nor switching costs as such, but rather path dependence – and in particular the opportunistic exploitation of path dependence effects that magnify the value of the patented invention relative to the best available alternative.
The paper, which can be found here, is structured as follows:
Part II discusses the prior literature on patent holdup, along with the early literature on path dependence.
Many commentaries on patent holdup confuse two important concepts – sunk and switching costs. A reader may reasonably come away with the impression that switching costs contribute to holdup in substantially the same way that sunk costs do. Discussions often omit consideration of how costly the alternative technology’s adoption would have been—or how large its variable profits would have been—if it had been selected ex ante. However, if the advantage of a patented technology over an alternative is that it is much cheaper to implement, then the switching costs that could be extracted ex post could also largely have been extracted ex ante, as part of the legitimate value of the patented technology. In such a case, if holdup arises, it is due to the sunk costs of implementing the original technology, not the fact that switching is expensive. Similarly, even if switching is not particularly costly, holdup may be substantial if the best alternative technology becomes markedly less valuable as a result of ex ante commitment to the patented technology. Consequently, a myopic focus on switching and sunk costs, while ignoring dynamic changes in variable profitability, will tend to produce over- or under-estimates of the holdup problem.
There is some literature that shares this view. It points out that ex post switching from the standard technology to an alternative is often not merely expensive, but strictly more so than hypothetical ex ante adoption of the alternative. Many articles about standards emphasise the interoperability value that is forgone when an implementer unilaterally switches away from the adopted standard, and some scholarship considers particular circumstances in which the cost of adopting the alternative becomes more expensive ex post.
Part III presents a simple model of holdup and switching costs.
This model purports to clarify what phenomena ultimately contribute to patent holdup. Consistent with the canonical theory of (non-patent) holdup, the first factor is simply the technology-specific sunk cost investment in the patented technology. This reflects a form of path dependence, because the choice to invest in a patented technology alters the relative value of the technologies going forward. Furthermore, the relationship between technology-switching and holdup is ultimately driven by path dependence effects. As a result, the model demonstrates that: (i) holdup rents are positive if and only if ex ante commitment to the patented invention generates path dependence; and (ii) switching costs do not independently reveal anything about the magnitude of the holdup rents. Even if the switching cost is quite large, it influences holdup only to the extent that it exceeds the cost of adopting alternative technologies ex ante.
Thus, holdup rents do not increase in tandem with switching costs, but rather in proportion to the size of the opportunity cost of adopting alternative technologies. On the other hand, even if switching costs and sunk costs are zero, holdup may still arise, for ex ante commitment to technology 1 may diminish the post-adoption variable profitability of technology 2.
Part III also provides a number of examples and applications for the model.
An important example concerns standard-setting. The principal value of standard-setting is that it helps to ensure interoperability among a large number of complementary technologies. But, ex post, an implementer enjoys this value only so long as it adheres to the standard adopted ex ante. The model makes it easy to see how this creates a pervasive holdup threat – which may also be affected by other factors, such as sunk costs tending to be larger in this context. In particular, the model explains that, while it may be intuitive to think that holdup serves to extract the network value of the adopted standard—technology 1—that is not the case. It is not the forgone network value associated with technology 1 that constitutes the holdup rents, but rather the forgone network value associated with not having selected technology 2 ex ante.
Commentaries on patent holdup frequently highlight the costs of redesigning an infringing product to replace the patented component with a non-infringing alternative. However, what matters is not the cost of ex post redesign, but rather the amount by which ex post “redesign” is more costly than ex ante “design.” A third example concerns patent injunctions – it is widely asserted that injunctions can facilitate excessive licence fees via holdup. The injunction would produce “lag time” during which the implementer is not able to sell anything while it develops a non-infringing product. If not for the injunction, the implementer would anticipate such development to avoid losing profits during the “lag time”. In short, the holdup rents reflect the opportunity cost of failing to develop a non-infringing product during the time lag.
Part IV concludes.
In short, the authors argue that technology switching contributes to holdup only to the extent that there is a dynamic increase in the cost of adopting an alternative technology, or else a reduction in its variable profitability. An increase in market power by an SEP owner as a result of strong network effects is just one possible embodiment of this problem. On the other hand, while it is true that switching costs can increase royalty rates, it does not follow that this reflects patent holdup: cost savings attributable to the patented invention are simply part of the legitimate advantage of the patented invention over the alternative.
As such, the literature tends to give an oversimplified account of how switching costs enter into the holdup problem. It is incorrect to presume that they affirmatively contribute to holdup whenever they arise. However, this treatment can also cut in the opposite direction, because focusing principally on sunk costs and switching costs may understate the risk of holdup. Indeed, some prominent embodiments of the patent holdup problem, illustrated in the previous section, arise not from these cost concepts, but because of a reduction in post-adoption profitability of the alternative. Consequently, a finding that the act of switching will not impose substantial costs hardly suggests the absence of a holdup risk, even if sunk costs are similarly negligible.
More generally, the authors argue that the link between switching costs and holdup can be categorically attributed to path dependence: some dynamic changes have inflated the relative value of the patented invention for reasons unrelated to technological merit, and the patentee can capture the difference when asserting its patent ex post. Ex ante contracting could have anticipated and avoided this result, but without such timely contracting the patentee can set its price based on a distorted comparison of value.
This is an interesting, clearly reasoned paper. Personally, I was not aware of patent holdup discussions focusing on switching costs to the detriment of path dependency – I always thought of these matters as involving path dependency, even if I thought of this as a form of sunk costs (which, naturally, go beyond the actual costs incurred in adopting a certain technology, and also include related opportunity costs). It may be that I was technically inaccurate in this, but I was not convinced of this by the paper.
In any event, the paper is quite helpful in how it clarifies the underpinnings of patent holdup. On the other hand, I think this paper could have benefitted from enhanced engagement with holdup theory – and, in particular, with the question of at which point holdup triggers antitrust concerns. This was a matter which was the subject of serious criticism (and a subsequent defence) in a couple of articles on patent holdup reviewed in earlier emails / posts, and which I would like to see further discussed (partially because I don’t fully understand it). It is also, I am afraid, a point which I raise in my comments to one of the articles below.