Abstract: Use panel data sets collected from five high-tech industries in Chinese 31 provinces through year 1995-2010, this study examines effects of geographical proximity, technological proximity and combined proximity on inter-provincial knowledge spillovers respectively by spatial panel data analysis. The empirical results indicate that, first, estimations of spatial panel data analysis is more precise than that of common panel data analysis. Second, R&D capital contributes more than R&D labor on knowledge production. Third, whether under geographical proximity or technological proximity, the presence of Chinese inter-provincial knowledge spillovers in high-tech industries is much evident, in which technological proximity has a little more effects, but the effects of combined proximity is not sufficient as presumed. The estimation varies with different model specifications and its implications are discussed.

摘要： 采用我国31个省份中5个高技术产业1995-2010年的创新投入与产出数据，基于地理邻近、技术邻近和交互邻近三种空间权重矩阵，利用空间面板数据分析，分别考察了不同邻近性下我国高技术产业的省际知识溢出效应。实证结果表明：相比普通面板分析而言，空间面板分析的估计更为准确；研发资本对创新产出的贡献要高于研发人员；不管是在地理邻近还是在技术邻近维度下，我国高技术产业的省际溢出效应均很明显，其中技术邻近效应略高，然而基于二者的交互效应并没有明显增强；不同设定方式下模型系数存在差别，文章据此探讨了其背后的可能发生机制。

关键词: 省际知识溢出, 地理邻近, 技术邻近, 空间面板数据模型, 高技术产业, inter-provincial knowledge spillovers, geographical proximity, technological proximity, spatial panel data analysis, high-tech industries