【作者】 符淼；

【导师】 林少宫；

【作者基本信息】 华中科技大学 ， 数量经济学， 2008， 博士

【摘要】 技术的溢出有多种途径,按来源分有国际的技术溢出和国内的地区间技术溢出,按渠道分可以依赖干中学、看中学、竞争中学、R&D中学或人员交流,不同的行业有不同的渠道特点和溢出强度。技术是否能够通过这些途径溢出依赖于吸收方的技术水平或人力资本水平,当他们的技术吸收能力满足一定的门槛条件,才能产生正的技术溢出效应。技术溢出由于人员流动、信息传播、产品贸易等原因存在空间相关,且技术溢出存在与地理距离相关的传播成本,这正是技术产业高度聚集的原因。技术溢出的研究虽然很多,但是以上领域的研究存在很多空白和分析方法上的不足。我们利用阈值回归来确定技术溢出的门槛条件,利用并改进空间计量方法,分析技术的空间相关性分析估计技术溢出强度随地理距离变化的模式,这些新方法的应用是已有研究所未做的。准确衡量技术是分析技术溢出的基础。本文估算并比较了技术的三种主要衡量形式:索洛剩余法TFP、DEA Malmquist指数法TFP和专利数据。通过分解经济增长投入因素的贡献,发现技术对经济增长的驱动能力越来越强,1997年前主要表现为技术效率提高,1997年后纯技术进步的作用显著而稳定。技术和经济活动都存在局部集聚性,技术的集聚度高于经济;两者的集聚度随时间增强,地理分布高度一致。技术的全局空间相关性低于经济正好解释了技术的强集聚性。技术先进地区和经济发达地区都偏于沿海。阈值回归表明,以受高等教育劳动力占总劳动力百分比表示的人力资本水平存在两个门槛:4.85%和10.99%.当人力资本水平达到4.85%的门槛时,FDI对东道国技术进步负挤出效应减半;当人力资本水平超过10.99%的符号改变门槛时, FDI的负挤出效应变为正溢出效应。从整体上看,我国人力资本水平低于符号改变门槛,但这并不意味着我国不能从FDI技术溢出中获益。由于存在地区差异和显著的省际技术溢出效应,处于门槛之下的相对落后地区可依赖超越门槛的发达地区,由发达地区吸收外部先进技术,而后通过省际技术溢出将技术扩散到相对落后地区。在内部地区间存在经济和技术差异的发展中国家,省际技术溢出和外资技术溢出同等重要。本文的研究发现来自国内的省际技术溢出比来自FDI的技术溢出贡献更大。省际技术溢出强度和技术差距的关系是U型曲线,技术门槛存在但处于值域的上界之外,因此技术差距越小则越容易发生技术溢出。干中学和R&D是内部获取技术的稳健渠道。R&D在吸收FDI技术溢出方面发挥重要的作用。中西部的省际技术溢出效应强于东部,落后地区同等投入的干中学和R&D投入可以获得比先进地区更大的产出,说明存在省际技术追赶的可能性。在发达地区,技术进步的主要来源是干中学、本土R&D以及依赖R&D吸收的FDI技术溢出。随地理距离快速下降的技术溢出效应是导致技术和经济局部集聚的原因,基于空间计量方法和Romer研发模型的分析结果表明,在一到两个省的范围或800公里内为技术的密集溢出区,此范围可用于考虑技术影响力的经济圈划分;800公里以上为快速下降区,技术溢出效应强度减半的距离为1250公里。远距离时快速递减的扩散效应和相对稳定的缪尔达尔回流效应是东西部发展不均问题的原因之一。R&D资本存量与创新产出之间存在理想的正线形关系;空间外部性主要通过误差冲击的空间传递来实现;R&D外部性、人力资本流动以及市场的竞争和合作是导致创新溢出的主要原因。强化以上因素作用以及促进信息高速公路、学术信息库和技术交易市场的建设将有助于区域的平衡发展。外贸和外资的技术溢出在不同行业存在差异。在不考虑吸收因素的交叉作用时,外资对技术创新产生负的挤出效应,其挤出效应超出了其本身作为知识生产者的正面作用;内资是否发挥作用的关键在于行业市场是否充分竞争和是否重视研发的投入;外资技术溢出效应显著的部门是适宜“看中学”的行业和因为技术差距较大而未与外资直接竞争的企业;出口贸易和干中学拉动创新的关键在于生产的产品是否有较高的技术含量。区域的专利数据由于人口流动和知识传播的原因,存在着空间相关。采用空间面板模型提高了模型的拟合度和系数的正确性。人口的流动和流动导致的人口的增量对专利创新有重要的影响,人口流入较多的地区专利创新多;经济增长对创新的影响是正面稳定的且具有一定的局部性;高校教育有正面影响且空间相关性较大,但其作用尚待发展。更多还原

【Abstract】 The channels of tehnology spillovers are multiple. From the perspective of technology senders, there are internatioal technology spillovers and intranational technology spillovers. As technology recipients, they can depend on learning by doing, learning by observing, learning by competition, learning by R&D or by personnel flows. The utilities of channels and strength of spillovers are variable across industries. Although channels are provided, technology spillovers occur or not depending on the absorptive capacity represented by technological level and human capital. Positive technology spillovers take place only if the recipients satisfy a certain threshold. Tehnology activities are spatial correlated because of personnel flow, information distribution and trade. The increasing technology transfer cost related to geographical distance is the important reason that causes the high concentrations of technology industries. Although literaures related to technology spillovers are abundant, studies on above fields are insufficient and lack appropriate econometric methodology. We estimate the thresholds for technology spillovers with threshold regressions. Based on spatial econometrics and our new methodological development, we analyse the spatial technology correlations and the decling style of spillovers as the geographical distance increases. These researches will give new developments in technology spillover studies.The exact estimation of technology is the basis of technology spillover studies. This paper estimates and compares the three major measurements of technologys, that is, TFP estimated by the Solow surplus, TFP estimated by the DEA Malmquist index. By decomposing the economic growth into three parts of imputs, we find the impacts of the technology on Chinese economic growth are increasin. Before 1997, it is technology efficicency improvement that play important role in economic growth, and after 1997, pure technology changes have stable and positive effects over economic growth. Both technologic and economic activities are geographically localized. The concentration of technology is stronger than that of economy. Both become more concentrative over time and distribute consistently in geography. The lower global spatial correlations of technology compared with economy explain the stronger concentration of technology well. Both technologically advanced areas and economically advanced areas are sticked to the coast area.The results of threshold regression suggest that the human capital has two thresholds: 4.85% and 10.99%, in terms of percentage of labors received higher education. When human capital quality surpasses 4.85%, the negative technology crowding-out effects of FDI halve. When it surpasses 10.99%, negative crowding-out effects of FDI change to positive spillover effects. China as a whole doesn’t meet the sign-change threshold, and this doesn’t mean that we can’t benefit from FDI technology spillovers. Because of regional diaparities and significant inter-provincial technology spillovers, the regions below the threshold can depend on technologically advanced regions above the threshold. Foreign advanced technologies are adopted by advanced regions first, and later they are transferred to the backward regions through inter-proviancial spillovers.In a country with heterogeneous economic and technological characteristics, the inter-provincial technology spillovers are as important as international spillovers caused by FDI. It finds that inter-provincial technology spillovers, that is technology transfer from technologically advanced provinces to less advanced provinces, contributes more than the international technology transfer via FDI for the backward regions. The relationship between the strength of inter-provincial technology spillovers and technology distances is U-shaped with the technology threshold falls outside the upper bound of technology distance, suggesting that technology spillovers takes place more effectively when technology distance is small. Learning by doing and R&D are robust internal approaches for technical progress. R&D also plays a key role in the assimilation of foreign technologies. Inter-provincial technology transfer effects are stronger in the middle and west regions of China than in the east region. In backward areas, the knowledge output elasticities of learning by doing and R&D are larger than those in technologically advanced regions, indicating the possibilities for inter-provincial technological catch-up. In technologically advanced regions, learning by doing, indigenous R&D and technology transfer from FDI all play a significant role in the technical progress of these regions.The spillover effect declining with distance is the major reason for the concentrations of technology and economy. Based on spatial econometrics and Romer’s R&D model, we find that the range of 800 km or one to two provinces is the intensive area for technology spillovers. This range gives an applicable benchmark for the segmentation of economic zones with consideration of technology spillovers. Above 800 km, the degree of technology spillovers declines rapidly. The distance at which the amount of spillovers is halved is about 1250 km. The quickly declining spread effects and relative stable Myrdal’s backwash effects cause the inequalities of the east and west areas. There is perfect positive linear relationship between innovative output and R&D capital stock. Spatial externalities are transferred spatially by error shocking. The externalities of R&D, human capital flow, competition and corporation of market are the major reasons for innovative spillovers. For balanced development across regions, it is important to strengthen the effects of above factors and facilitate the constructions of information super highway, academic database and technology exchange market.The technology spillover effects of FDI and exporting are industry specific. Without considerring the interaction effects of absorptive factors, it finds that for technology innovation, the negative crowding-out effects of FDI overweigh its positive knowledge production. The innovative efficiency of domestic firms depends on the adequate competition of the industry market and the industry investment of R&D. The industrial sectors with significant FDI spillover effects are those industries that are appropriate for learning by observing or those firms who don’t compete with FDI directly as their technology gaps from FDI are so big. Exporting and learning by doing facilitate innovative capacity when the exporting or producing goods are technologically advanced.Because of the migration of the population and the diffusion of knowledge, regional patent data are spatially correlated. The spatial panel data method improves the determination of the model and the correctness of the coefficients. Innovations are directly related to population growth caused by migration. Regions with more immigrated population are more creative. The effects of economic growth on technology progress are robust and somewhat localized. The effects of university education are highly spatial correlated and leave much to be desired..更多还原