关键词: 产业转移, 知识溢出, 区域创新, 空间计量模型

Abstract: Industrial transfer is an effective way to form a rational industrial division system and optimize the spatial distribution of economy. At the present stage in China, it is an inevitable requirement to promote the transformation of economic development mode, accelerate the adjustment and upgrading of industrial structure and promote the coordinated development of regional economy. On the one hand, the spatial distribution of industries, especially manufacturing industry, is an important factor affecting regional economic disparities; on the other hand, China is in a critical period of growth momentum transformation, and the transformation and upgrading of manufacturing industry as the main body of the national economy is particularly important. Especially in recent years, China’s manufacturing industry is facing many problems, such as structural overcapacity, backward technology, rapid rise in labor wages, and sustained decline in profits, while developed countries have implemented the strategy of "re-industrialization" to reshape the new competitive advantages of manufacturing industry, which makes the development of China’s manufacturing industry facing the "two-way squeeze" from developed countries and other developing countries. The above situation urgently requires China to re-optimize the spatial allocation of industries, give full play to regional comparative advantages and strengthen technological innovation, so as to achieve the goal of promoting industrial upgrading in the eastern region and accelerating the process of new industrialization in the central and Western regions.
On the other hand, compared with the spatial agglomeration of economic activities, the geographically unbalanced distribution of innovation is more obvious. From the perspective of agglomeration economy and externality, industrial distribution is an important factor affecting the spatial and temporal evolution of innovation output. Industrial agglomeration can bring about significant knowledge spillover, thus reducing the cost of production and R&D, and forming a "public knowledge pool". Knowledge spillover is a process of knowledge unconscious transmission by different subjects in direct or indirect information exchange, and knowledge spillover has the characteristics of attenuation as distance increases. The spatial distribution of industry will affect knowledge spillover, and knowledge spillover will further affect the spatial pattern of industrial activities, which presents an endogenous cumulative cyclic causality. Academic circles generally agree that agglomeration and knowledge spillover can promote regional innovation, but there are many debates about which type of agglomeration structure can promote innovation growth. “MAR externalities” think that knowledge spillovers mainly occur between enterprises within the industry, emphasizing the similarity or at least correlation of technology; while “Jacobs externalities” think that knowledge spillovers between complementary and differentiated industries rather than similar industries, emphasizing the impact of diversity on innovation. A large number of empirical studies have explored this issue. However, due to the differences in measurement indicators, geographical units, industrial choice and inspection time, there are still many divergent conclusions. 
Based on spatial panel data of innovation in Chinese provinces from 2001 to 2014, the paper measures space-time evolution characteristics of regional innovation and analyzes influences brought by manufacture transfer and knowledge spillover to regional innovation space evolution.In 2001, the patent shares of the eastern, central, western and northeastern regions were 63.05%, 14.00%, 12.39% and 10.57%, respectively. Since then, the share of innovation output in the eastern region has been increasing, while the share in the central, western and northeastern regions has been declining gradually. By 2007, the share of innovation output in the eastern region was 74.66%, while that in the central, western and northeastern regions was 9.99%, 8.58% and 6.77%, respectively. Subsequently, the share of innovation output in the eastern region began to show a downward trend, while the share of innovation output in the central and western regions showed an increasing trend. By 2014, the share of innovation output in the four regions will be 65.05%, 14.89%, 15.35% and 4.71% respectively. From 2001 to 2004, the share of manufacturing industry in the eastern region increased from 67.50% to 70.86%, indicating the transfer of manufacturing industry to the eastern region; while the share of manufacturing industry in the central, western and northeastern regions decreased from 13.41%, 10.69% and 8.40% in 2001 to 12.10%, 9.70% and 7.34% in 2004, respectively, indicating that manufacturing industry was transferred from the above three regions. From 2004 to 2011, the situation was just the opposite: the share of manufacturing industry in the eastern region fell to 60.65%, and manufacturing industry turned out; while the share of manufacturing volume in the central, western and northeastern regions rose to 18.56%, 12.51% and 8.28%, respectively, with the manufacturing industry turning in. During the investigation period, China’s manufacturing industry experienced a spatial shift from the eastern region to the central and western regions and then to the Northeast region.
With 2007 as the time boundary point, innovation output share in eastern areas increased firstly and then decreased, presenting a reverse U-shaped variation characteristic. Correspondingly, innovation output share in middle and western areas decreased firstly and then increased, presenting a U-shaped variation characteristic. During this period, the manufacture industry in our country presented the space-time characteristics that with 2004 as the time boundary point, it transferred towards eastern areas and then transferred to middle and western areas as well as the northeast China region. Relative variation in regional manufacture shares caused by transfer of manufacture exerted prominent positive spillover influences on regional innovation, narrowing the huge gaps between eastern areas and middle and western areas in the innovation output to a certain extent. Professionalization and diversification of regional manufacture exerted obviously positive spillover effects on regional innovation. Investment in research and development capitals, regional development level and activity degree of technical market brought obvious influences to regional innovation.
Policy enlightenment: From the perspective of promoting regional coordinated development, we should give full play to the comparative and competitive advantages of various regions. To further promote the industrial transfer from the eastern region to the central and western regions is conducive to the industrial agglomeration and knowledge spillover effect in the central and Western regions, thus promoting the growth of innovation output. From the perspective of promoting regional innovation growth, first of all, we should strengthen the construction of industrial clusters, which can make full use of the knowledge spillover effect between enterprises and industries, whether in the same industry or in diversified industries. Secondly, we should pay attention to R&D capital investment to improve innovation ability and knowledge absorption in the central and Western regions. In order to better absorb the knowledge spillover brought about by industrial transfer in foreign countries and Eastern regions, the ability to collect knowledge can be improved.

Key words: manufacturing transfer, knowledge spillover, regional innovation, spatial econometric model