网络DEA模型在科技创新投资效率测度中的应用研究

An Applied Study of Network DEA Model to the Efficiency Measurement of Scientific and Technologic Innovation Investment

引入网络数据包络分析模型测度科技创新投资过程的纯技术效率,以适应科技创新投资转化过程中上游的科技研发过程和下游的技术转化过程构成的关联网络结构,借此消除因忽略科技创新投资转化过程中两个子过程的关联性以及在规模效应与生产技术上的差异性而引致的效率估计偏误,并且使每个参评单元各过程之间的效率值直接可比。本文基于传统数据包络分析比率型模型最乐观构建的思想建立了包含关联子系统的网络型生产系统在可变规模报酬假设下的纯技术效率测度与分解模型,并提出网络型生产系统的整体与局部规模收益状态的判断准则。将该模型用于分析中国省域科技创新投资效率发现,在省域平均水平上,上游的科技研发和下游的技术转化两个子过程的效率绩效存在严重的不匹配,技术转化效率水平在整个省域科技创新效率水平中起到更为突出的作用。

This paper introduces a network data envelopment analysis model to estimate the pure technique efficiency of scientific and technological （S＆T） innovation investment processes. This method can accommodate for the relational network structure composed of the upstream research and development （R＆D） process and the downstream technological transformation process, by which the estimation errors in efficiency scores due to neglecting the linkage between two subprocesses as well as the cross-subprocess difference in scale effects and production technologies can be removed. In this way, different processes are comparable in the efficiency score of each deci- sion-making unit. This study extends the optimization-based thinking of DEA modeling to the pure efficiency measurement and decompo- sition of network production systems with relational subsystems under variable returns to scale assumption, and presents the judgment rules of returns to scale of the whole production system and its subsystems. Our model is applied to China＇ s province-level S＆T innova- tion investment. The empirical results show that there is a significant unmatching relationship between the upstream R＆D process and the downstream technological transformation process, and the latter plays a more important role in the overall innovation efficiency of prov- inces at the average level.