产业特性、区域特征与中国制造业省区集聚

产业地理集聚存在产业差异和区域差异。本文结合产业特性和区域特征揭示产业集聚的形成机制。采用2004 年第一次经济普查数据，本文发现交通密度、政府支出比重、开发区数量以及贸易壁垒强度等区域特征显著影响产业集聚。在产业特性与区域特征交叉分析中发现，市场化促进了农业投入依赖性强、产业联系强、劳动密集型及研发密集型的产业的空间扩散，交通网络的完善有利于产业联系强的、规模经济显著的产业的分散布局，在市场潜力较大的省区规模经济显著的产业更为集聚；地方政府的产业和区域政策促进了产业联系较强的产业集聚，而拉动了利税率高的、国有企业主导的、属于国家重点计划的产业分散布局；各级政府建立的开发区和园区则尤其不利于国家计划重点产业的集聚和规模经济的实现。本文研究表明，贸易理论与新经济地理理论对于中国的制造业集聚具有一定的解释力，而中国的政策制度环境也是产业集聚的重要影响因素。产业集聚需要一定的产业和区域条件，并非所有产业和所有区域内都能够实现高水平的产业集聚。

Industrial Attributes,Provincial Characteristics and Industrial Agglomeration in China

This paper explains industrial agglomeration by combining industrial attributes and provincial characteristics. Based on data from the first economic census in 2004, this study found that industrial agglomeration significantly varied by industry and across province. Overall, industrial agglomeration in the western provinces is much higher than that in the coastal provinces. Indidvidual industries however show significant different patterns. Statistical analysis indicates that provincial characteristics such as per captita GDP, road and raiway density, ratio of non-state-owned economy, ratio of governmental expenses to GDP, number of development zones, social capital and law enforcement, and trade barrier indeed influence the extent of industrial agglomeration. Controlling for provincial dummies, industrial attributes such as agricultural input intensity, labor input intensity, average enterprises size, industrial linkages, ratio of foreign capital in total capital, ratio of exports in gross output, ratio of profits and value added taxes in total sales, ratio of state capital in total capital, transportation input intensity and planned key industries are associated with industrial agglomeration. The significance of many interactions between industrial attributes and provincial characteristics suggests that industrial agglomeration is not only industry-specific but also province-specific. The policy implication is that local governments shall carefully choose industries to develop industrial clusters.