Fostering the diffusion of general purpose technologies
Fostering the diffusion of general purpose technologies: Evidence from the transistor
There is an ongoing debate about the secular decline in productivity growth. What seems particularly surprising is that the digital revolution does not seem to have spurred productivity growth as much as one would expect. In a recent Vox column, Gordon and Sayed (2020) pointed out that the ICT revolution failed to boost European productivity long after its impact could be seen in the US. Andrews et al. (2015, 2016) documented an increasing productivity divergence between global frontier and laggard firms, pointing to a lack of diffusion of new technologies as a cause of slow productivity growth. Brynjolfsson at al. (2019) similarly argue that the lack of positive productivity effects of new technologies like artificial intelligence and IT innovations may be caused by slow implementation and argue that these technologies need complementary investments like organisational adjustments before they show up in productivity statistics. This raises the question of how the diffusion of new technologies and the development of products building on these technologies can be sped up. In particular, one question is to what extent the patent system might be responsible for the lack of diffusion of new technologies?
Historical accounts suggest that patents on important technologies, and general purpose technologies (GPTs) in particular, are harmful for technological progress and economic growth. The most well-known example is James Watt’s steam engine patent. Mokyr (1994), among others, argued that “because [Watt] held a wide-ranging patent, he succeeded in blocking [the development of high-pressure steam engines] for many years”. Scherer (1965) writes that their refusal to issue licenses “clearly retarded the development and introduction of improvements”. And according to Boldrin and Levine (2008), “by keeping prices high and preventing others from producing cheaper or better steam engines, Boulton and Watt hampered capital accumulation and slowed economic growth”. Similarly, Merges and Nelson (1990) write that there “is good reason to believe that the Wright [brother’s] patent [on an efficient stabilising and steering system] significantly held back the pace of aircraft development in the United States”. And Selden’s patent on an internal combustion engine allegedly slowed automobile development in the early 20th century (e.g. Merges and Nelson 1990).
These narratives of patents hindering the diffusion of key technologies are often used as prime examples for the “case against patents”, suggesting that patenting rights should be weakened or abolished altogether. For example, Merges and Nelson (1994) “(…) come out with the belief that the granting and enforcing of broad pioneer patents is dangerous social policy.”
In the case of general purpose technologies, speeding up their diffusion is particularly warranted because of the positive feedback loops, which set them apart from other technologies. Improvements in the GPT stimulate innovations in the application sector, which in turn give incentives to further improve the GPT. Any delay caused by patents and withholding their licensing is thus particularly costly for society.
Patents have been shown to block follow-on invention in various settings (e.g. Moser and Voena 2012, Galasso and Schankerman 2015). Yet, it is not clear whether this is a relevant concern for GPTs as their potential benefits are so large that they might provide sufficient incentives for efficient technology licensing (e.g. Green and Scotchmer 1995). Understanding the effects of patents for GPTs is important because while GPTs are rare, they are credited with driving sustained economic growth since the industrial revolution.
In a recent paper (Nagler et al. 2021), we study the effect of patents and the licensing of patents on follow-on inventions to the transistor, the defining general purpose technology of the 21st century. From early applications such as hearing aids and radios to modern technology like fast computer chips and smartphones, the transistor and its subsequent developments spread to almost all sectors of the economy. The first working transistor was invented in 1947 by American physicists John Bardeen, Walter Brattain, and William Shockley at the Bell Laboratories. The three shared the 1956 Nobel Prize in Physics for their achievement.
In 1952, the Bell System decided to license its transistor patents at a standardised rate of $25,000 and provided training programmes for all firms who bought licenses. Commentators saw this generous licensing regime as a calculated political move to appease the authorities in an ongoing antitrust case against the Bell System that sought to break up the company (Mowery 2011, Gertner 2012). But according to internal memos at the Bell Labs written a decade later, engineers at the Bell Labs also understood that “by involving engineers around the world in the evolution of the device – making it better, cheaper, more reliable – the hope was that everyone would profit from the advances, especially the Bell System” (Gertner 2012). The standardised licensing opened the transistor technology, reducing the entry barriers to the industry as one commentator vividly described: “If you were going to be a player in semiconductors in the early 1950s, you’d wish you knew the AT&T patent lawyer just as you wish you knew your rich uncle” (Carrick 1982).
We investigate whether the standardised licensing increased follow-on innovation to the transistor. In Figure 1, we show a comparison of follow-on inventions as measured by patent citations to the transistor and a control group of exactly matched patents. We use exactly matched non-Bell patents with the same filing year, the same technology class, and the same number of citations until 1952, i.e. before the standardised licensing started. The figure shows that the standardised licensing of the transistor technology led to a jump in patents building on Bell’s transistor patents.
In 1956, the antitrust case against Bell was settled with a Consent Decree, which obliged Bell to license all of its existing patents royalty-free (Watzinger et al. 2020). Thus, even the licenses on the transistor patents now became available royalty-free. Interestingly, as Figure 1 shows, this did not boost the spillovers of transistor patents any further. This suggests that standardised licensing in itself is more important than the specific royalty fees.
In deeper analyses, we show that the standardised licensing in particular increased cross-technology spillovers. As cross-technology spillovers are a defining characteristic of general purpose technologies, this confirms that patents on GPTs are particularly problematic without a standardised licensing procedure. The impacts are driven by inventors unrelated to the Bell System working in unconcentrated markets. This is in line with commentators’ views on the reasons behind the rapid improvement of the transistor. For example, Merges and Nelson (1994) write that “[m]any companies ultimately contributed to the advance of transistor technology, because the pioneer patents were freely licensed instead of being used to block access.”
There are many stories of how a diverse set of entrepreneurs and inventors benefited from the easily accessible license and the training. Jack Kilby, the eventual co-inventor of the integrated circuit, got his start with the transistor technology when he attended Bell’s ten-day crash course that came with buying a license (Reid 2001: 71-72). Masaru Ibuka licensed the transistor patents to build a transistor radio at SONY, at the time a young company that he had co-founded was struggling to stay in business. By 1957, SONY had issued a pocket transistor radio that sold over 1.5 million units and had become an internationally known company (Nathan 2001).
Our study adds empirical evidence to case studies on the effect of patents on important technologies as recounted in Boldrin and Levine (2008). It shows that the effect of patents on technologies with significant potential for cross-technology spillovers might be particularly harmful. Our study also contributes to the literature on the impacts of patents on follow-on innovation (Williams 2017). We add to this literature by showing that the impact of patent licensing on follow-on innovation is substantially stronger when patents cover a GPT. We also provide evidence that the type of follow-on innovation that is blocked by these patents differs from follow-on innovations blocked by less exceptional patents.
Finally, our paper contributes to the literature on the history of US innovation with the first in-depth analysis of the diffusion of the transistor technology. Already in 1962, Richard Nelson highlighted the importance of the transistor:
“The transistor has had its most significant impact not as a component replacing vacuum tubes in established products, but as a component of products which were uneconomical before the development of the transistor. Very compact computers are the most striking example. Without transistors, computers of a given capability would have to be much larger both because vacuum tubes are larger than equivalent transistors and because cooling requirements are much greater for vacuum tubes. Almost all of our new airborne navigation, bombing, and fire control systems, for example, are transistorized. So are all of our satellite computers. And without transistors our large computers […] undoubtedly would be much more expensive – probably so much so that many of their present uses would not be economically sound. Thus the transistor has stimulated growth, including the invention and innovation on a considerable scale of products which can profitably use transistors as components.”
Although the enormous significance of the transistor technology is widely recognised and the importance of the non-discriminatory licensing by Bell has been suspected to have played a crucial role for its diffusion (e.g. Levin 1982, Merges and Nelson 1994), our paper is the first to provide an empirical analysis of how important the licensing decision of the technology by Bell was for the inventions in the semiconductor industry.
Andrews, D, C Criscuolo and P N Gal (2015), “Frontier Firms, Technology Diffusion and Public Policy: Micro Evidence from OECD Countries”, OECD Productivity Working Papers 2015/02.
Andrews, D, C Criscuolo and P N Gal (2016), “The Best versus the Rest: The Global Productivity Slowdown, Divergence across Firms and the Role of Public Policy”, OECD Productivity Working Paper 2016/05.
Boldrin, M and D K Levine (2008), Against intellectual monopoly, Cambridge University Press.
Brynjolfsson, E, D Rock, and C Syverson (2019), “Artificial Intelligence and the Modern Productivity Paradox: A Clash of Expectations and Statistics”, Economics of Artificial Intelligence: An Agenda, edited by Ajay Agrawal, Joshua Gans, and Avi Goldfarb, University of Chicago Press, 23-57.
Carrick, R (1982), “AT&T Likely to Limit License”, InfoWorld, 32–33.
Galasso, A and M Schankerman (2015), “Patents and Cumulative Innovation: Causal Evidence from the Courts”, The Quarterly Journal of Economics 130(1) :317–369.
Gertner, J (2012), The Idea Factory: Bell Labs and the Great Age of American Innovation, Penguin.
Gordon, R J and H Sayed (2020), “Transatlantic technologies: Why did the ICT revolution fail to boost European productivity growth?”, VoxEU.org, 21 August.
Green, J R and S Scotchmer (1995), “On the Division of Profit in Sequential Innovation.” The RAND Journal of Economics 26 (1):20–33.
Levin, R (1982), “The semiconductor industry”, In Government and Technical Progress: A Cross-Industry Analysis, edited by Richard R. Nelson. Pergamon Press.
Merges, R P and R R Nelson (1990), “On the Complex Economics of Patent Scope”, Columbia Law Review 90 (4):839–916.
Merges, R P and R R Nelson (1994), “On limiting or encouraging rivalry in technical progress: The effect of patent scope decisions”, Journal of Economic Behavior and Organization 25 (1):1–24.
Mokyr, J (1994), “Technological Change, 1700-1810”, in An Economic History of Britain since 1700, vol. 1, edited by Roderick Floud and Donald N. McCloskey, Cambridge University Press, 12–43.
Moser, P and A Voena (2012), “Compulsory Licensing: Evidence from the Trading with the Enemy Act”, American Economic Review 102 (1):396–427.
Mowery, D C (2011), “Federal Policy and the Development of Semiconductors, Computer Hardware, and Computer Software: A Policy Model for Climate Change R&D?”, in Accelerating Energy Innovation: Insights from Multiple Sectors, University of Chicago Press, 159–188.
Nagler, M, M Schnitzer and M Watzinger (2021), “Patents on General Purpose Technologies: Evidence from the Diffusion of the Transistor”, CEPR Discussion Paper No. 15713.
Nathan, J (2001), Sony: The Private Life, Harpercollins.
Nelson, R (1962), “The Link between Science and Invention: The Case of the Transistor”, in The Rate and Direction of Inventive Activity: Economic and Social Factors, Princeton University Press, 549–584.
Reid, T R (2001), The Chip: How Two Americans Invented the Microchip and Launched a Revolution, Random House Trade Paperbacks.
Scherer, F M (1965), “Invention and Innovation in the Watt-Boulton Steam-Engine Venture”, Technology and Culture 6(2), 165-187
Watzinger, M, T Fackler, M Nagler and M Schnitzer (2020), “How Antitrust Enforcement Can Spur Innovation: Bell Labs and the 1956 Consent Decree”, American Economic Journal: Economic Policy 12(4): 328–359.
Williams, H L (2017), “How do Patents Affect Research Investments?”, Annual Review of Economics 9:441–469.