关键词: 高技术产业, 创新效率, 双目标DEA模型, 两阶段, 数据包络分析

Abstract: Innovation is a core driving force of economic growth and a critical factor to promote national competitiveness. High-tech industry, based on knowledge-intensive technologies and the integration of multi-disciplinary technological achievements, is a main source of innovation, an important pillar industry for China’s economic development, and an important force to enhance national scientific and technological competitiveness and industrial core competitiveness. To scientifically and reasonably evaluate the innovation efficiency of high-tech industry is of great significance for optimizing the allocation of innovation resources of high-tech industry, enhancing its innovation competitiveness, and promoting its position in the global industrial chain and supply chain.
Based on the innovation value chain theory, the high-tech industry innovation activity is generally divided into two successive sub-processes: R&D process and commercialization process. To be specific, R&D process measures the ability to transform R&D inputs into scientific and technological outputs, and commercialization process measures the ability to marketize the scientific and technological achievements.The two-stage data envelopment analysis (DEA) model has been widely used in innovation efficiency evaluation of high-tech industry. Previous studies usually portray the relationship between the two sub-processes of R&D and commercialization according to the assumption of “leader-following”, and then evaluate the innovation efficiency and sub-process efficiency of high-tech industry, ignoring the cooperative relationship between the R&D and commercialization sub-processes. Strengthening the mutual cooperation between the two sub-processes of R&D and commercialization is important to enhance the innovation efficiency and innovation capacity of high-tech industry, and maintain its sustainable and efficient development. Incorporating the idea of win-win cooperation, this study extends the traditional two-stage DEA model and develops a bi-objective DEA model from the perspective of the cooperation between the R&D process and commercialization process. The proposed bi-objective DEA model treats the two sub-processes equally and equitably, that is, it maximizes each sub-process’s efficiency simultaneously and autonomously and determines the optimal stage weight corresponding to each sub-process. By traversing the Pareto optimal efficiency of two sub-processes, the optimal solution of the bi-objective DEA model can be obtained. Then, taking 29 provincial-level high-tech industries in mainland China as the research object (Qinghai and Tibet are excluded because of data missing), the proposed model is used to evaluate and analyze their overall innovation efficiency, R&D efficiency, and commercialization efficiency of 29 provincial-level high-tech industries in mainland China from 2013 to 2015, which corresponds to three consecutive innovation periods i.e., 2013 represents 2013—2015; 2014 represents 2014—2016; 2015 represents 2015—2017.
The empirical results show that: First, the overall innovation efficiency of China’s high-tech industry is low, and the innovation efficiency of high-tech industries in half of the provinces is lower than the national average, which indicates much room for improvement. Second, for the vast majority of provinces, the R&D efficiency is greater than the commercialization efficiency, and the low commercialization efficiency is the main reason for the low overall innovation efficiency. Third, there exists a difference in the overall innovation efficiency, R&D efficiency, and commercialization efficiency of high-tech industries between provinces as well as regions. Specifically, the eastern area has the best overall innovation efficiency and sub-process efficiency, while the northeastern one performs the worst in terms of overall and sub-process efficiency. The central area has better overall innovation efficiency and commercialization efficiency than the western one, but performs worse in R&D efficiency than the western one. Last, we divide 29 provincial-level high-tech industries into four categories according to R&D efficiency and commercialization efficiency: high R&D efficiency and high commercialization efficiency, low R&D efficiency and high commercialization efficiency, low R&D efficiency and low commercialization efficiency, and high R&D efficiency and low commercialization efficiency. And then we propose specific strategies for improving the innovation efficiency of each provincial high-tech industry. This study enriches the theory and method of innovation evaluation, and provides a new perspective for the evaluation of innovation efficiency of high-tech industries, which has an important theoretical value and practical significance.

Key words: high-tech industry, innovation efficiency, bi-objective DEA, two-stage, data envelopment analysis