Technological Change
Paper Session
Friday, Jan. 7, 2022 12:15 PM - 2:15 PM (EST)
- Chair: Leila Agha, Dartmouth College
How Can Innovation Screening Be Improved? A Machine Learning Analysis with Economic Consequences for Firm Performance
Abstract
Using USPTO patent application data, I apply a machine-learning algorithm to analyze how the current patent examination process in the U.S. can be improved in terms of granting higher quality patents. Surprisingly and unsurprisingly, combining human examiners' expertise with machine learning predictions would yield a 15.5% gain in patent generality and a 35.6% gain in patent citations for granted patents. My economic analysis in the second part of the paper shows that getting a patent on which humans and machines disagree is a winner's curse. First, these patents are more likely to expire early. Second, patents granted by examiners with higher false acceptance rates cause lower announcement returns around patent grant news for public firms. Third, public firms who get these patents are more likely to get sued in patent litigation cases. Consequently, these firms cut R&D investments and have worse operating performance. Fourth, private firms whose patents are granted by such examiners are less likely to exit successfully by an IPO or an M&A. Overall, this study suggests that the social and economic cost of the current patent screening system is large and could potentially be mitigated with the help of a machine learning algorithm.Shared Culture and Technological Innovation: Evidence from Corporate R&D Teams
Abstract
Given the increasing emphasis on workplace diversity and labor productivity in the modern knowledge economy, we seek to open the “black box” process of corporate innovation production by focusing on the most important input into the firm’s R&D process, namely the individual employees tasked with developing new inventions. Using a novel ‘within-firm’ corporate R&D setting (where we compare individual inventors and inventor teams working in the same firm at the same time in the same office, in combination with information on over two million inventors employed at U.S. public firms), we investigate how individual inventors’ inherited traits (e.g. cultural values and gender) and acquired career experiences (e.g. prior patenting and employment history) affect their desire to collaborate with one another in a corporate R&D team setting and how shared cultural values amongst R&D team members affects a team’s innovation output. We first provide novel evidence that, even amongst groups of comparably experienced inventors working in the same corporate office, inventors who share similar cultural values are 20% more likely to work together on new research projects. Second, using exogenous shocks to inventor team composition arising from premature co-inventor deaths, we find that more culturally homogenous teams produce a higher overall quantity of patents that are more likely to exploit existing technologies and become moderately successful inventions. In contrast, more culturally diverse inventor teams produce a higher share of risky, more exploratory patents that have a greater chance of becoming high impact innovations. Overall, our results have important implications for the promotion of different types of corporate innovation in R&D intensive workplace environments as well as the likely effectiveness of diversity hiring policies.Robots, AI and Immigration - A race for talent or of displaced workers
Abstract
We study the effect of technological change on immigration flows as well as the labor market outcomes of previous migrants versus natives. We look at two different automation technologies: Industrial robots and artificial intelligence. For this purpose, we take advantage of data provided by the Industrial Federation of Robotics as well as online job vacancy data. Our research focuses on Germany, a highly automated economy and one of the main migration receiving countries among OECD countries. We apply an instrumental variable strategy and find that robots decrease the wage of migrants across all skill groups, while having no significant impact on the native population nor immigration flows. In the case of AI we find an increase in the wage gap and also unemployment gap of the migrant and native population. In addition, AI leads to a significant inflow of immigrants. This holds for the low-, middle- and high-skilled and is indicative of migrants facing displacement effects, while natives might benefit from productivity and complementarity effects. Policymakers should pay special attention to the migration population when designing mitigation policies in response to technological change in order to avoid further increases in inequality between migrants and natives.The First Randomized Controlled Trial on Independent Inventors: An Experiment at the US Patent and Trademark Office
Abstract
The markets for technology are inefficient, with high upfront costs, high transaction costs, risks of expropriation, and high uncertainty regarding validity of intellectual property rights. Patent Offices seek to reduce these frictions through policies. Patenting is particularly challenging to the independent inventor with no resources for legal representation, who might have great technologies and entrepreneurial spirit, but lack knowledge of the terminology and processes required to obtain a patent. The White House authorized the USPTO through an executive action to provide additional support to inventors without legal representation. This led to the first ever Randomized Controlled Trial on inventors at the USPTO, where innovators were randomly assigned to a treatment group with examiners providing additional support according to a set of procedures or the regular patent prosecution pathway as the control.There is little prior evidence on the impact of patents on economic outcomes since it is challenging, if not impossible to randomly assign patents to some inventors or firms, and then study outcomes. Therefore, many existing empirical studies suffer from potentially unobserved characteristics between entities receiving patents and those not receiving patents. The novel Pro Se Pilot overcomes this difficulty by exogenously increasing the likelihood of receiving a patent in a randomly selected group of treatment patent applications. To the best of our knowledge, our study is the first to use an RCT to analyze a foundational issue about the basic functioning of the patent system.
Our preliminary results indicate a strong and significant effect of the program on patent outcomes. Further analysis on inventor network measures, career outcomes for the inventors, and entrepreneurial outcomes are underway and will be presentable by the time of the conference.
JEL Classifications
- O3 - Innovation; Research and Development; Technological Change; Intellectual Property Rights