Research On Technology Innovation Efficiency Of High-tech Industry Based On DEA Network Structure

With the rapid development of high-tech industry,the efficiency of technological innovation is playing a more and more important position in promoting the international competitiveness of high-tech industries.The method of decomposing the efficiency of technological innovation also needs to change from the "black box"structure of the previous single system to the multi system "grey box" structure.Aiming at the problem of measurement and decomposition of technological innovation efficiency in high-tech industry,this paper mainly studies the network structure model based on DEA with shared input and free intermediate output,measures and decomposes efficiency from two-stage model and extended three-stage model respectively.In addition,an improved algorithm from a single variable to a multivariable is used to solve the problem that the optimal solution is not unique due to the different priorities of each stage.Through the empirical analysis,it provides the reference for the high-tech industry to improve the efficiency of technological innovation.First of all,according to the existing research results of scholars,it builds a two-stage network structure model based on DEA with shared input and free intermediate output;And in the same conditions,it puts forward the three-stage network structure model based on DEA with shared input and free intermediate output.Secondly,scholars often consider the weight adding method or product method alone,ignoring the systematic error of the calculation model.Meanwhile,it uses two methods to explore the error effect of the model calculation results.Thirdly,in order to solve the problem that the optimal solution of the whole efficiency is not unique due to the different priorities of each stage,this paper explores and puts forward a multivariable algorithm replacing the single variable algorithm to solve this problem.Finally,the index data are collected,from the two situations of non shared input and free intermediate output,shared input and free intermediate output,it analysis two-stage network structure model based on DEA and three-stage network structure model based on DEA.In the empirical analysis,it compares and analysis the effects of these two combinations on the measurement results;The advantages of weighted summation and product method in the decomposition of overall efficiency and the error effect of measurement results are compared and analyzed;The optimality of multivariate algorithm and univariate algorithm on measurement results is compared and analyzed.It provides some suggestions for studying the efficiency of other complex network structure and improving the efficiency of technological innovation in the future.