Thursday, March 13, 2014

Incentives to Create *Whose* Knowledge?

Guest Blogger

Jim Bessen

For the conference on Innovation Law Beyond IP at Yale Law School

In theory, early stage tech firms need government help. It is commonly argued that without help from the government, inventors and startup tech firms will have insufficient incentives to invest in R&D, they will receive insufficient financing and they will have difficulty commercializing or licensing their technologies. To solve this “market failure” problem, policymakers promote the patent system, innovation prizes, R&D grants, subsidies, tax credits, loans and investments to private companies, all with varying degrees of success. Such policies are assumed to be the central policies for spurring innovation and are the focus of scholarly work at conferences like Innovation Law Beyond IP.

Yet oddly, some of the most successful government programs at spurring emerging technologies were not designed to provide incentives to inventors or startup firms. Computers, semiconductors, the Internet, wireless communications, and many advances in public health and civil engineering were crucially promoted by government programs designed to purchase new technologies needed for national defense, public health, infrastructure, space exploration and other government functions. In an earlier era, government procurement promoted the development of interchangeable mechanical parts, leading to the development of the “American System of Manufactures” that propelled the new nation to international leadership in manufacturing technologies.

Incentives for broad-based learning of new technologies

These histories suggest that incentives for inventors and startup firms are not the only incentives that matter for innovation. Perhaps at times they are not even the most important incentives. Why have government purchasing programs played such pivotal roles? Because implementing major new technologies requires motivating a wide range of people in addition to inventors and entrepreneurs: mechanics, engineers, installers, and ordinary users need to learn new skills and technical knowledge. Such learning is typically not a problem for mature technologies that have well-developed training institutions and labor markets that reward workers who acquire technical skills. But emerging technologies often lack such labor markets and training institutions. In these cases, government can and has played a crucial role developing broad-based new skills.

Why do labor markets fail to provide adequate incentives for early stage technologies? Because early stage technical knowledge is usually highly uncertain and unstandardized. Formal training is difficult under such circumstances. Instead, people typically acquire much knowledge through experience on the job or by exchanging knowledge with people who already have experience. And when technology is not standardized, employers have difficulty hiring experienced workers — their experience is hard to evaluate, they may have learned on somewhat different equipment, etc. The lack of standardization creates “information asymmetries” that give rise to well-known problems. The result is that under such specialized conditions, labor markets do not provide socially optimal incentives for workers to invest in learning.

Consider, for example, interchangeable parts. The early Republic needed to acquire large quantities of firearms for defense, especially firearms made of matching parts that could interchanged. That way, when rifles failed, broken parts could be replaced in the field. But although the US had a sizeable private market for firearms, there were few skilled gunsmiths and early 19th century gunsmiths had not mastered the precision production techniques needed to make parts interchangeable from one gun to the next. The War Department established two armories, hired a diverse range of contractors, required them to exchange knowledge, and developed common standards. The effort took decades, but it generated a large corps of skilled machinists who had the tools and knowledge to do precision work. These machinists subsequently turned their skills to civilian markets, promoting a wide range of advances including the assembly line.

Something similar happened more recently with the Internet. DARPA, the Defense Department agency that promotes new technologies for defense applications, involved diverse sorts of players in early Internet development, they encouraged knowledge sharing and established open standards. This program built a corps of engineers and programmers who were familiar with Internet and web protocols and who could make improvements and build innovative applications based on these standards. These people proved to be a potent source of innovation. They readily surpassed proprietary initiatives such as efforts to develop “Interactive Television” in the early 1990s. And they were responsible for making the initial Internet and web largely US creations even though the original web protocol was developed in Europe. Indeed, the UK, France and the Soviet Union all had internetworking initiatives that were much less open and proved to be much less innovative in the long run.


This last point highlights the policy dimension. Government procurement programs do not inherently promote broad-based skill development. This happens when well-designed programs include diverse players, promote knowledge sharing, and establish open standards. It is notable that since the 1970s, relatively few defense R&D programs have helped spur major new civilian technologies. This decline was accompanied by shifts in the nature of these programs. Research programs became dominated by a few large defense contractors, eliminating many of the small firms and independent academics, and concerns about secrecy limited knowledge sharing and open standards. These changes occurred under the greater political influence of defense contractors and, following 9/11, amid heightened concerns about security, yet the result appears to have impacted innovation. The design of these programs mattered.

Nor is government procurement the only area where policy influences broad-based learning of new technical knowledge. Employee job-hopping has been shown to be important for knowledge sharing; this is affected by the law of employee non-compete agreements and by trade secrecy law. Occupational licensing restrictions and antitrust law affect the scope of players who can participate in learning a new technology. Support for community colleges, vocational training, and on-the-job learning promote skill development as well.

The broader implication of this analysis is that policy needs to consider the complete innovation “ecosystem,” not just the inventors and entrepreneurs. Indeed, in many of the examples where government procurement spurred the early development of major new technologies, inventors and entrepreneurs did not appear to be laboring under insufficient incentives. They often freely shared knowledge and many eschewed patents. And when they did patent, many did not gain much. While patents became much more important once the technologies matured, policy needs to recognize that the relative importance of different incentives changes over the technology life cycle.

Indeed, poorly designed patents can restrict the ability of early stage users to exchange knowledge. For example, some judges promote the notion that "pioneer" patents should be read broadly in order to provide the strongest incentives to early stage inventors. Yet overly broad patents can limit the beneficial effects of employee job-hopping and can deter the beneficial formation of new startups with slightly different technologies. Innovation policy needs to balance these conflicting interests and, so, patent policy should not be decided in isolation. Unfortunately, the current institutional structure makes balanced policy formulation difficult.

Jim Bessen is a lecturer in law at Boston University. He can be reached at jbessen at

Older Posts
Newer Posts