KNOWLEDGE FLOWS IN CHINA : A PATENT CITATIONS ANALYSIS Presented

Knowledge Flows Within China: A Patent Citation Analysis Name: Jia LIU Affiliation: Universite de Strasbourg Faculte des Sciences Economiques et de Gestion Bureau d'Economie Theorique et Applique Year of enrollment: 2018 Expected final date: 2021 E-mail: jialiu@unistra.fr 1) Existing state-of-the-art Krugman (1991) emphasizes the importance of knowledge spillover, and he thinks knowledge spillover is hard to track. He has argued, ?Knowledge flows are invisible; they leave no paper trail by which they may be measured and tracked?. Scholars generally think that knowledge spillover is an invisible process and its trajectory is difficult to track. Nonetheless, Jaffe(1993) first use the patent citation as a proxy of the knowledge flow. They find that patent citations can be used as an ?article trail? to measure and track the knowledge spillover. Based on Jaffe’s work, numerous studies start to apply patent data to measure the direction of knowledge spillovers. Criscuolo et al. (2005) use patent citation data from the EU Patent Office to quantify the relative asset augmenting and exploiting character of foreign-located R&D. Criscuolo and Verspagen (2008) try to combine the EPO and the USPTO patent citation data to observe knowledge flows. 2) Research Gap and Research Question The most difficult issue in using patent citation data is the patent data collecting process itself. The novelty of our study lies in that in the existing studies, there is no empirical study of using patent citation to analyze knowledge flows within China at the provincial-level administrative unit level up to now. The patents granted by Chinese Patent Office don’t contain the patent citations information. The United States is China’s most important destination country for foreign invention patent applications, most literature use USPTO to get the citation information. The traditional way of getting citation data is from the NBER. However, the NBER only contain the patents granted between 1963 and 2006. Until 2006, the number of patent applicated by China is small and the number of intra-national citation is much smaller. For the reason above, there is no empirical study of using patent citation to analyze knowledge flows within China. Our key research question has been to examine whether geographical distance, provincial level administrative unit borders knowledge spillovers in China. We also investigate the extent to which knowledge spillovers are confined to regions with particular technological specialization. This paper focuses on the empirical study of the knowledge spillovers in China's provinces, and can more accurately grasp the direction and extent of knowledge spillovers in Chinese provinces by quantitative analysis. 3) Theoretical Arguments We construct a model of knowledge spillovers between province-level administrative units and several empirical explanatory variables. In our study, the number of patent citations between two provinciallevel administrative units as the dependent variable, and we regard geographical distance and province level administrative borders are the main explanatory variables, while other variables, such as technological proximity is also included in the model. 4) Method and Data The citation distribution between province-level administrative units is highly skewed, and many values are zero and small. In addition, patent citation data is count data and appears as an integer. Considering these characteristics of the data, the model is estimated by Tobit regression and negative binomial regression to conduct quantitative studies on relevant factors affecting Chinese patent citations. In the Tobit estimation, the two-step method proposed by Heckman (1979) was applied. As for the data, we use the patent and patent citation data from the U.S Patent and Trade Office of first inventors residing in China in the period of 1977-2017. We resort to the Patents View to collect data. 5) Results Firstly, short geographical distance will increase the probability of inventor citations or, in other words, that knowledge spillovers are localized to some extent. Secondly, technical similarity has a positive impact on knowledge flows in China. Thus, the finding in this paper that technology flows are both industry-specific and confined by geography, provincial-level administrative unit borders, indicates that regional polarization in China may indeed be a reality. KNOWLEDGE FLOWS IN CHINA: A PATENT CITATIONS ANALYSIS

中国的知识流：专利引用分析名称：Jia Liu隶属关系：De Strasbourg University de Sciences des Sciences Economiques et de gestion de gestion bureau d'Chanceie Theorique Theorique et aptrique of Aprilllent of Adrinlllent Inalllent Inalllent Inallland of Grance Inalllent Intrique of Gryalllent of Grestiques：2018预期的最终日期：2021年电子邮件：2021电子邮件：jialiu@unistra ）现有的最先进的克鲁格曼（1991）强调知识的重要性，他认为知识很难他说的是，知识是不可见的； ）首先使用专利引用作为知识流的代理贾夫（Jaffe）的工作开始应用专利数据，以衡量Criscuolo等人（2005年）使用欧盟专利办公室的专利引用数据来量化相对资产的增强和利用外国关注的R＆D Verspagen（2008）尝试将EPO和USPTO专利引用数据结合起来，以观察知识流2）研究差距和研究差距最多。使用专利引用数据的困难问题是我们研究的新颖性的专利数据收集过程。到目前应用程序，大多数文献都使用USPTO获取引文数据。由于上述原因，非国内引用的数量较小。我们要研究中国的地理距离，省级行政部门的知识溢出。 ，可以通过定量分析更准确地抓住中国政策中知识的方向和程度。理论论点我们在我们的研究中构建了省级行政单位和几个经验利用的知识模型。是主要的剥离变量，而其他变量（例如技术接近）也包括在模型中。方法和数据之间的引文分布是高度偏斜的，许多值还为零，而专利的引文数据是计数数据，并且是一个整数。通过TOBIT回归和负二项式回归来估计，以对影响中国专利的相关因素进行定量研究。赫克曼（Heckman）（1979年）是针对数据的。数据5）首先，较短的地理距离会增加发明人引用的可能性，换句话说，知识在某种程度上是局部的。关于中国的知识，本文的发现，技术既特定于行业，又限制了省级行政部门的边界，这表明中国的区域两极分化确实是一个现实。专利引用分析