多维邻近性对不同知识基础产业创新的影响——基于ANN和OLS回归双重检验

Multi-dimensional proximity effects on industry innovation of different knowledge base——Double test based on ANN and OLS regression

采用人工神经网络和普通最小二乘回归结合的方法,从全国层面以及东、中、西三大地区层面对地理邻近和认知邻近在合成型和解析型知识基础产业创新中的作用进行实证检验,得出地理邻近对合成型知识基础产业创新呈正效应显著的半倒"U"型影响,作用强度高于认知邻近的弱正向效应;地理邻近对解析型知识基础产业创新呈狭窄倒"U"型影响,作用强度低于认知邻近的平缓倒"U"型影响;东部地区两种知识基础产业创新中地理邻近初现饱和甚至抑制效应;中部地区合成型知识基础产业创新中地理和认知邻近正效应突出,解析型知识基础产业创新中地理和认知邻近效应为显著倒"U"型;西部地区两种知识基础产业认知邻近极其缺乏,对创新的影响不显著。

Use Artificial Neural Network and Ordinary Least Squares regression method to explore the effect of geographic proximity and cognitive proximity effects in the synthetic knowledge based and analytical knowledge based industry innovation in the national and the three regional levels. We find that geographical proximity effect on the synthetic knowledge based industrial innovation is significantly accelerated with the type of half inverted＂U＂,and the influence degree is higher than the weak promotion effect of cognitive proximity.Geographical proximity effect on the analytical knowledge based industrial innovation is like narrow inverted＂U＂,the influence degree is lower than the cognitive proximity effect. Geographical proximity effects are over saturated in the eastern region. Geographical proximity and cognitive proximity effects are positive in synthetic knowledge based industries but showing inverted＂U＂in analytical knowledge based industries in the central region. Cognitive proximity effects need to be more cultivated in both synthetic and analytical knowledge based industries in the western region.