Investing in science, research and technology: where is the biggest bang for the buck?

随着区域经济体在国家和国际层面上日益竞争以发展其创新优势，帮助区域识别和支持创新起源的因素已成为区域经济开发商最重要的任务之一。该项目基于知识生产函数文献，该文献依赖于统计方法来估计研发投资对创新生产的影响。由于该项目关注的是区域层面的创新成果，因此需要明确说明区域之间的知识外部性的作用，这是经济地理学新文献中提出的一种现象。然而，到目前为止，只有少数研究使用适当的统计技术来模拟和衡量区域间知识外部性;当它们这样做时，它们进行全球（而不是地方）测量，掩盖了各区域创新特征的显著差异。因此，这项工作改进了传统的方法，在地方一级，即为每一个区域提供了知识溢出的衡量标准。这项工作还调查了四个“高技术”部门的区域知识创造的动态，以更好地证实各部门之间的知识外部性差异。此外，统计分析是在三个不同的空间尺度上进行的（美国县、大都市统计区和州），使这项工作比传统方法更完整，对地方政府来说更有信息量。另一个贡献是一种衡量研发投资所创造的收入和就业数量的方法，即其社会回报。这种贡献是通过两种互补的技术，称为空间计量经济学和多区域投入产出的区域间外部性建模。最后但并非最不重要的是，本项目中使用的数据集比以前的研究有了显著的改进。调查人员不仅依靠美国专利商标局的最新私人研发投资数据，而且还首次使用人口普查局的综合联邦基金报告来估计联邦资助的研发投资对地区层面知识创造的作用。总的来说，该项目的目的是显示在哪里最大限度地提高研发投资，并帮助区域政府制定适当的政策，以促进知识创造。该项目的广泛影响有三个方面：首先，该项目采用的全面统计方法将帮助当地决策者确定他们应该依靠哪些关键经济特征来刺激当地的知识创造。第二，明确估计公共投资的相对回报对于支持循证财政政策至关重要。因此，在本研究中，私人和社会的研发投资回报率可以直接与教育、交通、国防等其他公共投资的回报率进行比较。最后，除了通常的学术出版物和课堂演示外，该项目的结果将在动态WebGIS（地理信息系统）的框架内在线免费访问。由于它能够让大量受众访问所有结果，因此这种传播方法可以促进手头研究的透明度、客观性和问责制。

As regional economies increasingly compete at the national and international levels to develop their innovative edge, helping regions identify and support the factors at the origin of innovation has become one of the most important tasks of regional economic developers. This project is based upon the knowledge production function literature which relies on a statistical approach to estimate the impact of investments in R&D (Research and Development) on the production of innovation. Because this project focuses on the production of innovation at the regional level, it requires us explicitly to account for the role of knowledge externalities between regions, a phenomenon brought to the fore by the new literature in economic geography. As of today, however, only a handful of studies use the appropriate statistical techniques to model and measure interregional knowledge externalities; when they do, they make global (not local) measurements which mask significant variations in the characteristics of innovation across regions. As such, this work improves upon the traditional approach by providing measurements of knowledge spillovers at the local level, i.e. for every single region. This work also investigates the dynamics of regional knowledge creation for four "high-technology" sectors to better substantiate differences in knowledge externalities across sectors. In addition, the statistical analysis is performed on three different spatial scales (US counties, Metropolitan Statistical Areas and States), making this work more complete and informative to local governments than traditional approaches. Another contribution is a method of measuring the amount of income and jobs created by R&D investments, i.e. their social returns. This contribution is accounted for through the modeling of interregional externalities by means of two complementary techniques called spatial econometrics and multi-regional input-output. Last but not least, the dataset used in this project significantly improves upon previous studies. Not only do the investigators rely on the most recent private R&D investment data from the US Patent and Trademark Office, but they also use the Consolidated Federal Funds Report of the Census Bureau to estimate, for the first time, the role of federally funded R&D investments on knowledge creation at the regional level. Overall, this project aims at showing where to maximize R&D investments and help regional governments tailor appropriate policies to promote knowledge creation. The broader impacts of this project are threefold: first, the thorough statistical approach used in this project will help local policymakers identify the key economic characteristics they should rely on to stimulate knowledge creation in their locality. Second, a clear estimation of the relative returns on public investments is essential to support evidence-based fiscal policy. The private and social returns on R&D investments evaluated in this work can thus be directly compared with the returns from other public investments such as education, transportation and defense, for instance. Finally, in addition to the usual scholarly publications and in-class presentations, the results of this project will be accessible online and for free in the frame of a dynamic WebGIS (Geographic Information System). Because it is able to give a large audience access to all the results, this dissemination method promotes transparency, objectivity and accountability in the research at hand.