Adam Jaffe

Innovation is central to models in economics, strategy, management, and finance, yet it remains difficult to measure due to its intangible and knowledge-based nature. Recent advancements in natural language processing offer new methods to analyze textual artifacts, providing empirical insights into previously hard-to-measure aspects of innovation. This article provides an overview of the current applications of these methods in empirical innovation research, highlights their transformative potential for reshaping how researchers study and quantify innovation, and discusses the critical challenges that accompany their use.

The arrival of creative machines—software capable of producing humanlike creative content—has triggered a series of legal challenges about intellectual property. The outcome of these legal challenges will shape the future of the creative industry in ways that could enhance or jeopardize welfare. Policy makers are already tasked with creating regulations for a postgenerative artificial intelligence creative industry. Economics may offer valuable insights, and this paper is our attempt to contribute to the discussion. We identify the main economic issues and propose a framework and some tools for thinking about them.

Moving beyond the combination of adoption subsidies, standards, and (albeit limited) attempts at carbon pricing that largely characterized US climate policy over the past decade, recent climate-related legislation has transformed not only the scale of US climate activities but also the policy mechanisms adopted. Newly scaled policy instruments—including demonstration projects, loan guarantees, green banks, and regional technology hubs—are motivated not only by unpriced carbon externalities but also by innovation market failures. This paper maps the economics literature on innovation market failures and other frictions to the stated goals of these policy instruments, with the goal of focusing discussions about how to implement these policies as effectively as possible. The paper also discusses how program evaluation can help to illuminate which market failures are most relevant in a particular context and which policy instruments are most targeted to them.

The ‘Science of Science Policy’ is an interdisciplinary field of scholarly study that seeks to model, measure, and evaluate the interaction of public policies (including funding) and the performance of the science and innovation system. Such study offers insights and findings that can increase the effectiveness of science and innovation policy for New Zealand. New Zealand’s rate of public investment in research is only about two-thirds, and that of business investment in research about one-third of the OECD average. Our low business R&D intensity is not unexpected given our market size, firm size, and industrial composition. Nevertheless, public policy should aim to mitigate barriers to firms’ ability to undertake high-return research investments. As we undertake only 0.2% of the world’s research, most of the knowledge used in New Zealand is created elsewhere, so positioning ourselves to derive maximum benefit from others’ research is likely to have high payoff. The spillover phenomenon should be considered in making decisions about public research support. A re-balancing from competition towards cooperation, and encouraging linkages between research entities and organisations with a commercial orientation will increase the likelihood of spillover benefits. Research output could be increased relatively quickly by increasing the currently low expenditure per researcher, but any sustained increase will require more skilled scientists and engineers. Therefore public financing of research should include a monitored programme of training grants, and immigration policy settings that do not inhibit attracting skilled scientists and engineers. Attraction and retention of research ‘stars’ should be an explicit objective of New Zealand science policy. A culture of innovation requires an attitude that defines success in terms of the global market, not the local market. It requires a social, economic and cultural environment that rewards risktaking and does not see failure as a barrier to undertaking further investment. Policy makers, too, should be willing to take risks, and systematic

Encouraging disclosure is important for the patent system, yet the technical information in patent applications is often inadequate. We use algorithms from computational linguistics to quantify the effectiveness of disclosure in patent applications. Relying on the expectation that universities have more ability and incentive to disclose their inventions than corporations, we analyze 64 linguistic measures of patent applications, and show that university patents are more readable by 0.4 SD of a synthetic measure of readability. Results are robust to controlling for non-disclosure-related invention heterogeneity. The linguistic metrics are evaluated by a panel of “expert” student engineers and further examined by USPTO 112(a) – lack of disclosure – rejection. The ability to quantify disclosure opens new research paths and potentially facilitates improvement of disclosure. •We propose a method using computational linguistics to quantify patent disclosure.•We show that university patents disclose 0.4 SD more than corporate patents.•University-corporate disclosure gap is larger among more-experienced applicants.•Corporations that focus on licensing disclose 0.3 SD more than other corporations.•Expert panel evaluations and USPTO 112(a) rejections show modest support.

Previous research suggests creative ability peaks at ages between the mid 30s and early 40s, but has not focused on the role of age-related changes in cognitive abilities in this pattern. Cognitive processes show aging-related increases in experience-based knowledge (pragmatics or crystallized abilities) and decreases in the ability to process novel information quickly and efficiently (mechanics or fluid abilities). We explore the role of these agerelated changes in the invention process, using a new database created by combining the publicly available patent data with information on inventor ages scraped from directory websites for approximately 1.2 million U. S.-resident inventors patenting between 1976 and 2017. We have made these data publicly available on the Harvard Dataverse and full documentation can be found in Kaltenberg et al. (2021) In the current paper, we present some descriptive statistics, and explore changing patterns of invention as inventor's age. For solo inventors, backward citations and originality increase with age, consistent with their being connected to crystallized intelligence. Forward citations, number of claims, and generality measures, as well as a citation-based measure of disruptiveness decline with inventor age, consistent with a connection to fluid intelligence. A similar pattern was found for performance in teams based on the average age of inventors in the team. Exploration of age diversity showed that teams with a wider age range had patents that are slightly more important (i.e., with more forward citations). Merging of these new data with other data that capture diverse aspects of inventors' environment and incentives offers rich potential for new research on invention.

•Patented inventions vary widely in their relative quality, however this is defined.•A framework for comparing patent-quality-related outcomes is presented.•Common outcomes, such as forward citation counts or patent lifetime, are very inconsistent in their assessment of what constitutes a ’high-quality’ patent.•The relationships between these outcomes and patent characteristics defined at grant are extremely variable across outcomes and technology types.•Neither measurement of patent quality, nor policy responses to a proliferation of poor-quality patents, are likely to be sensibly addressed by a one-size-fits-all approach.
The quality of novel technological innovations is extremely variable, and the ability to measure innovation quality is essential to sensible, evidence-based policy. Patents, an often vital precursor to a commercialised innovation, share this heterogeneous quality distribution. A pertinent question then arises: How should we define and measure patent quality? Accepting that different parties have different views of, and different sets of terminologies for discussing this concept, we take a multi-dimensional view of patent quality in this work. We first test the consistency of popular post-grant outcomes that are often used as patent quality measures. Finding these measures to be generally inconsistent, we then use a raft of patent indicators available at the time of grant to dissect the characteristics of different post-grant outcomes. We find broad disagreement in the relative importance of individual characteristics between outcomes and, further, significant variation of the same across technologies within outcomes. We conclude that measurement of patent quality is highly sensitive to both the observable outcome selected and the technology type. Our findings bear concrete implications for scholarly research using patent data and policy discussions about patent quality.

The extent to which new technological knowledge flows across institutional and national boundaries is a question of great importance for public policy and the modeling of economic growth. This paper develops a model of the process generating subsequent citations to patents as a lens for viewing knowledge diffusion. We find that the probability of patent citation over time after a patent is granted fits well to a double-exponential function that can be interpreted as the mixture of diffusion and obsolescence functions. The results indicate that diffusion is geographically localized. Controlling for other factors, within-country citations are more numerous and come more quickly than those that cross country boundaries