EAGER: University and Corporate Research: Substitutes or Complements?

就在美国公司把重点放在创新作为增长源泉的同时，他们也在稳步地远离内部研究。这并不是因为研究与发明无关。相反，发明活动比以往任何时候都更依赖于研究。一种解释可能是大学研究的规模和相关性越来越大，以及创新劳动的新兴分工，公司和大学专门从事创新所需的不同系列活动。如果大学研究取代了企业研究，相关大学研究的扩大将减少企业研究。然而，公司的不同之处不仅在于相关知识的池，还在于它们所从事的创造性活动的类型。具体地说，正在生产更多新奇发明的公司，以及更依赖于最新科学进步的发明，可能会有更大的动机投资于内部研究，因为他们面临的科学问题可能不那么通用，因此不太可能被大学科学家解决。该项目的更广泛的价值包括显著扩大科学政策科学的数据基础设施。具体地说，该项目在科学和工程知识的使用和生产以及公司发明活动的新颖性方面提供了严格的衡量标准。这些信息将对政策制定者、行业从业者、学者和任何对科学和工程知识在创新中的作用感兴趣的人有用。这个项目利用先进的机器学习技术来探索这个问题。它将1980年至2016年期间的2000万份科学出版物与600万项专利进行匹配，以制定科学和工程研究与发明活动的相关性的衡量标准。它使用这一衡量标准来量化大学研究对4000家总部在美国上市的研发公司的相关性。然后分析了相关大学研究流程的变化与公司内部研究的关系。该项目的第二部分开发了一种衡量总部设在美国的公司申请专利的新颖性程度的指标，以分析拥有更多新奇发明的公司如何应对相关大学研究流程的变化。这些公司级别的数据通过一个易于使用的网络平台向广大公众开放。

Even as they focus on innovation as a source of growth, American corporations have steadily move away from in-house research. This is not because research is not relevant for invention. To the contrary, inventive activity relies on research more than ever before. One explanation might be the growing size and relevance of university research and an emerging division of innovative labor in which firms and universities specialize in different sets of activities that innovation requires. If university research substitutes for corporate research, expansion of relevant university research will reduce corporate research. However, firms differ not only in the pool of relevant knowledge, but also in the type of inventive activity they perform. Specifically, firms that are producing more novel inventions, and inventions that rely more on recent scientific advances, may have greater incentives to invest in internal research because the scientific problems they face are likely to be less generic and hence less likely to be solved by university scientists. The broader value of the project consists of significantly expanding the data infrastructure for the science of science policy. Specifically, this project provides firm level measures on the use and production of scientific and engineering knowledge, and on the novelty of the inventive activity of the firm. The information will be useful to policy makers, industry practitioners, scholars, and anyone interested in the role of science and engineering knowledge in innovation.This project leverages advanced machine learning techniques to explore this question. It matches 20 million scientific publications to 6 million patents for the period 1980-2016 to develop measures of the relevance of science and engineering research to inventive activity. It uses this measure to quantify the relevance of university research for 4000 R&D performing publicly traded American headquartered corporations. It then analyzes how variations in the flow of relevant university research is related to in-house research by a corporation. A second part of the project develops a measure of the degree of novelty of the patents filed by US headquartered corporations in order to analyze how firms with more novel inventions respond to changes in the flow of relevant university research. These firm level data are available to the broad public, through an easy-to-use web platform.