关键词: 科技服务业, 集聚, 工业效率, 门限回归, 空间计量

Abstract:    With the transformation of China′s economy to high-quality development, the improvement of industrial efficiency has become a more pressing task. We try to prove that promoting spatial agglomeration of S&T service is an effective way to improve industrial efficiency. There are two deficiencies in existing literatures. In the theoretical dimension, the mechanism of S&T service agglomeration affecting industrial efficiency is not yet clear and needs further exploration. Empirical studies on the influence of S&T service agglomeration on industrial efficiency are also very scarce, and more causal inference analysis is needed. Compared with the existing literatures, our marginal contributions lie in not only the construction of a theoretical model to illustrate the influence mechanism of S&T service agglomeration on industrial efficiency, but also the regression analysis using provincial data to confirm that the agglomeration of S&T service would significantly improve industrial efficiency. 
   We demonstrate that the spatial agglomeration of S&T service can produce strong economic linkage, which is manifested as competition effect and trust effect. On one hand, when S&T service companies agglomerate in the same region, they compete with each other. Competition urges enterprises to innovate constantly to improve their efficiency. Meanwhile, competition forces enterprises in S&T service industry to provide high-quality and inexpensive services to industrial enterprises, and timely adjust the service content to meet the changing needs of industrial enterprises. On the other hand, industrial enterprises can effectively alleviate the adverse selection problem caused by information asymmetry and identify trustworthy upstream suppliers by observing and comparing S&T service enterprises in the agglomeration area. Once the cooperative relationship is established, even if there are commercial disputes in the execution of the contract, they are easy to be settled through friendly negotiation. This trust saves industrial firms the transaction costs of searching, negotiating and monitoring, thus increasing their efficiency. Agglomeration of S&T service also improves industrial efficiency through knowledge linkage. The spatial agglomeration of S&T service forms an informal place for knowledge exchange, effectively overcoming the spatial limitations of tacit knowledge and greatly reducing the barriers to its transfer. As long as industrial enterprises enter the agglomeration area of S&T service, they can obtain tacit knowledge of S&T service enterprises through casual contact and communication. The tacit knowledge not only helps to expand the knowledge stock of industrial enterprises, but also promotes the creation and accumulation of new knowledge, which will greatly improve the industrial efficiency. 
    The sample period used in the empirical analysis is from 2008 to 2016, and the sample individuals are 30 provincial administrative regions in China. The core variable S&T service agglomeration is measured by location entropy, the explained variable industrial efficiency is measured respectively by labor productivity and total factor productivity, and control variables include industrial labor stock, the level of economic development, industrial structure, trade dependency, degree of government intervention, fixed effect of year and fixed effect of province. In addition to the linear baseline regression model, we set a threshold regression model to explore the nonlinear influence of the agglomeration of S&T service, and analyze whether the influence of S&T service agglomeration has spatial overflow characteristics by using the spatial econometric model. Finally, regional heterogeneity is tested by sub-sample regression. The baseline regression shows that the agglomeration of S&T service significantly promotes the improvement of industrial efficiency. From the threshold regression result, the agglomeration of S&T service has no nonlinear effect on industrial efficiency. Next, spatial econometric analysis finds that the spatial spillover effect of S&T service agglomeration is not significant, but the industrial labor productivity among neighboring provinces would be mutually inhibited. The heterogeneity analysis shows that the improvement effect of S&T service agglomeration on industrial efficiency mainly exists in the eastern region, but is not significant in the central and western regions. 

Key words: S&T service, agglomeration, industrial efficiency, threshold regression, spatial econometrics