Prof. Robert P. Merges, University of California at Berkeley School of Law
February 12, 2015
Patent pools have captured the attention of lawyers, economists, and historians for a long time. For many, they represent a dangerous solution to the transaction costs associated with clearing and licensing many patents. The danger arises because pools are formed when competitors come together to find a common industry solution – always a dangerous scenario, from an antitrust point of view. For others, they are an interesting solution to the serious tensions that arise when patents are assigned to many dispersed inventors, but products cannot be produced without agreement from all of them.
The antitrust concerns have come to the fore with many of the important pools formed in recent years in the consumer electronics industry. In addition to traditional worries about pooling (e.g., monopoly pricing and collusive opportunities),1 recent commentators have argued that patent pools reduce competition from optimal levels even if they do not quite rise to the level of de facto monopolies.2
Other commentators in recent years have supplemented the traditional concern with pools and firm-level competition. The new strain of scholarship centers more on the dynamic impact of pools on the rate and direction of technical advance within the technologies that are the subject of the pooled patents. One highly respected commentator argues that pools are harmful when they collect together patents for technologies that are substitutes for one another, as opposed to complementary patents (such as blocking patents).3 This concern has led some historians to look into the details of patent pools from earlier in U.S. economic history, with an eye to assessing their impact on technological development in various industries.
History Comes Alive: The Singer Sewing Machine Patent Pool
The most important historical work on patent pools is a detailed study of the innovation-related effects of the Singer Sewing Machine Patent Pool, formed in 1856 among three companies holding key patents on sewing machine technology.4 According to economic historians Ryan Lampe and Petra Moser, the key patents behind the pool gave the three companies that owned them a dominant position in the manufacturing of sewing machines that sewed using the “lockstitch” method. This type of stitch is stronger and less prone to unraveling as compared to the next-best alternative, the chain stitch.5
Lampe and Moser found that the patent pool on lockstitch sewing machines reduced innovation in lockstitch technology and redirected effort in the substitute technology, the chain stitch. Using patent counts, they show that, after the Sewing Machine Combination was formed in 1856, patent counts dropped drastically in the lockstitch field. This was matched by a significant uptick in the number of chain stitch patents:
Patent data reveal a drastic shift toward substitute technologies after the formation of the pool. Until the pool formed in 1856, a comparable number of lockstitch and chain stitch patents were granted each year. Between 1857 and 1862, nearly twice as many patents were granted for the chain stitch technology, compared with the lockstitch (pool) technology. Difference-in-differences estimates imply a 189% increase in chain stitch patenting between 1857 and 1867 compared with other types of sewing machine inventions. Results are robust to controlling for preexisting time trends, Poisson regressions to address the count data characteristics of patents, and a broad range of alternative specifications.6
The effect goes beyond just patent counts however; firm entry was also affected. As the authors argue:
[B]etween 1857 and 1867, 35 new firms entered with chain stitch [sewing] machines, compared with 18 that entered with lockstitch machines. New firms continued to enter with the substitute technology even after the pool’s patent on the lockstitch had expired in 1867. After the pool’s last remaining patent expired on May 8, 1877 and the pool dissolved, only one of 24 firms entered with the chain stitch.7
Assessing the Historical Record
Lampe and Moser have surely added a most painstaking case study to the body of knowledge about patent pools. It seems hard to argue with their conclusion that pooling the lockstitch patents reduced further research in the lockstitch field. And it seems likely that their additional conclusion is also right: research effort did pick up in the substitute chain stitch field.
What is not clear is whether they have really established that the sewing machine pool reduced consumer welfare. There are two reasons why this conclusion does not follow from their data.
First, as their own study shows, entry in the sewing machine industry remained robust during the era of the sewing machine pool. Using this most basic measure of the extent of competition in the industry, the patent pool did not suppress entry in any radical way. This is critical. It means that separate and unrelated inventors, entrepreneurs, and financiers all continued to see the possibility of high economic returns in the promising sewing machine industry. The patent pool did not keep entrants from joining the fray. Indeed, data from the pool’s own records show that attempts to control prices by means of royalty rates on the pooled patents were largely a failure. Despite rhetoric at the time about the outsized power of the sewing machine “Combination,”8 firm-level competition (as measured by entry and pricing pressure) seemed to remain robust throughout the period of the patent pool.
The second consumer welfare indicator centers on invention during the pool period, as measured by patent counts. The table below provides the basic data:
Technology Type |
Chain Stitch |
Lockstitch |
Prepool (1845-1856) |
29 |
30 |
Pool (1857-1876) |
179 |
280 |
Pool (1857-1867) |
142 |
123 |
Pool (1868-1876) |
37 |
157 |
Postpool (1877-1885) |
31 |
192 |
All years |
239 |
502 |
Patent Counts – Sewing Machine Fields (Lampe and Moser, 2013)9
Lampe and Moser emphasize that the fact R&D was deflected to chain stitch sewing machines during the operation of the pool suggests that technological opportunities in the lockstitch field were left undeveloped. As they say, “Between 1857 and 1862, nearly twice as many patents were granted for the chain stitch technology, compared with the lockstitch (pool) technology.”10 Yet the data in the table above show that this statement is carefully crafted, and omits what is arguably the larger picture. Though attention surely shifted to the chain stitch during the early years of the pool (1857 to 1862), patenting in the lockstitch field obviously recovered significantly before the major lockstitch patents expired in 1867. Though the chain stitch surely received attention in the early pool years, invention in the lockstitch field did not fall off a cliff during the operation of the pool. So if patent counts were a valid measure of invention (as is widely assumed), the pool clearly failed to have the effect predicted in much of the antitrust literature. Invention was robust; the pool did not cause a stagnation in the evolution of lockstitch technology.
But what to make of the fact that effort shifted to the chain stitch, an inferior substitute, during the operation of the pool? No doubt the increased price of the lockstitch machines created an incentive for research to focus elsewhere. And yet this is not necessarily a cautionary tale; “inventing around” a patent or group of patents is one of the positive effects of patents in many cases. More importantly, it is not purely a function of the patent pool. The pool began with some successful litigation. And even today, patents that survive the test of litigation command higher royalties. If these royalties stimulated increased inventive activity in related fields, that may reflect greater caution concerning potential infringement liability. And again, it might be just part of the price society pays for the original innovation. For at the end of the day, secondary effects from higher patent royalties could be said to represent not a sign of dysfunction, but instead a sign that patents were working as intended. The incentive called forth inventive effort, which led to a series of patents, which led to a search for alternative technologies.
Conclusion
Far from a horror story, the sewing machine patent pool may represent a success story from the annals of the past. During the existence of the pool, 53 new firms entered the sewing machine market. After it ended, 24 firms did. So even if research was redirected in a suboptimal way, the industry continued to grow and attract new entrants, more so during the patent pool even than afterward.
The larger lesson may therefore not be quite what Lampe and Moser intend it to be. Yes, we need to be on the lookout for anticompetitive conduct conducted under the rubric of a patent pool. Yet we also need to look carefully at the effects a pool might have on inventive effort and firm entry. Inventive effort might be redirected, but not necessarily in a fashion that undermines consumer welfare. And if entry remains robust, then the pool might very well be quite benign. It is, of course, good to learn from history. But we need to have all the data before we can say what the lesson really is.