Investments in research and development (R&D) are a key driver of economic growth. An important aspect of research activities is that the knowledge generated by one firm may spill over to other firms that also benefit from it. This notion of knowledge spillovers has been very influential in shaping research and in motivating innovation policies all over the globe, yet it isn’t the only manner in which knowledge (from R&D) diffuses in the current innovation landscape. Knowledge also often diffuses through voluntary technology transfers. FEB-researcher Pere Arque-Castells and Daniel Spulber, a colleague from Northwestern University (US, Illinois), take both channels into account and study their implications on innovation policies and the innovation landscape as a whole.

Knowledge spillovers lead to a well-known problem of underinvestment in R&D because the focal firm does not take into account such social benefits when deciding how much to invest. This market failure has long been at the core of the economics of innovation field and of many growth models. In their study, Arque-Castells & Spulber emphasize that the canonical approach traditionally used to estimate spillovers implicitly assumes knowledge to be non-tradable, which describes the organization of innovation in the mid-twentieth century. The current innovation landscape, however, is characterized by active intellectual property (IP) enforcement and frequent exchange in technology markets. In other words, knowledge diffuses both through unintended spillovers and voluntary technology transfers. Contrary to spillovers, technology transfers allow the focal firm to internalize the social benefits of its research.

Underestimating private return to R&D and overestimating social return

In their study, the researchers develop a general framework that allows R&D to diffuse both through spillovers and voluntary transfers. To operationalize the framework, they created a dataset of interactions in the market for technology between publicly held U.S. firms. “We found that the canonical model substantially underestimates the private return to R&D and overestimates the wedge between the social and private returns. Estimates suggest that gains from trade in market for technology are actually quite important and explain about 10% of total revenue in the sample of US publicly traded firms”, Arque-Castells explains. Overall, the researchers’ results suggest that knowledge diffusion takes place through more articulate channels than implied by the unintended spillovers narrative.

Conflicting innovation policies

Yet, the unintended spillover narrative has carried a great weight in shaping the innovation policies of the last fifty years in all OECD countries. Such policies are largely based on generous R&D subsidies, R&D tax credits and strong intellectual property rights (IPRs) for intangible assets. All with the goal of alleviating the underinvestment problem, either by reducing the costs or increasing the benefits of R&D. “However, some of these policies could be seen as conflicting. For example, certain firms could receive royalties thanks to their IP rights and at the same time claim R&D tax credits, which hinge on the premise that innovators do not receive compensation when other firms use their inventions. Thus, understanding the precise channels through which knowledge is transmitted is necessary to inform sound innovation policies”, Arque Castells states.

Pushing too far and in the wrong direction? 

The researchers’ study indicates that innovation policies are largely based on the results of a model that imposes, by construction, that innovation is more socially appealing than it really is. Argue-Castells: “This makes you wonder, have we been pushing pro-innovation policies too far? Have we been favoring the right firms? This is a timely question, because in the last years we have discovered several interesting, and to some extent concerning, stylized facts about innovation.” He gives the example that in most industries there is an increasing divide between best versus rest firms, with only the former experiencing substantive increases in wages, which generates income inequality. This could be partly explained by the fact that current innovation policies have been reinforcing superstar trends by over-subsidizing R&D investments in firms that already are capable of internalizing the multilateral effects of their research.

Careful with the unintended consequences

These trends have made Arque-Castells realize that innovation policies should not only be about incentivizing more innovation. “It is particularly important to understand the optimal design of innovation policies and their unintended consequences, with a particular focus on competition, dynamism, inequality and redistribution. With our research, we strive to bring some clarity and help in this endeavor.”

More information

The research paper by Pere Arque-Castells and Daniel Spulber has recently been published in the Journal of Political Economy. For more information, please contact Pere Arque-Castells.