本文放松环境同质性的假定条件,构建包含环境变量的网络DEA模型,测度环境变量对生产前沿及生产决策单元效率的影响效应。本文构建的模型放松了投入产出变量集合与环境变量集合间的独立性假定,并且不需要先验性地判断环境变量的作用方向,弥补了现有处理环境变量方法的不足。本文提出的距离函数是一种非径向非导向的效率测度类型,能够有效地测度所有潜在松弛。Monte Carlo模拟结果表明,本文构建的模型能够很好地测度环境变量对整体生产过程及其子生产过程的作用效应,创新生产过程分析结果表明出口依存度与技术创新效率间存在负相关关系,不考虑环境变量作用效应下测度的省际技术创新效率排名存在较大偏误。
In order to detect the effect of environmental variables, this paper constitutes a network DEA model with environmental variables, which is fully characterized by the extended production frontier estimator and proposed network directional distance functions. The established estimator combines the input-output space and the space of environmental variables without requiring a priori specification of the role of environmental variables, avoiding the unrealistic assumptions involved in most of the extant approaches. Besides, the defined weighted network directional functions both for conditional and unconditional measures are non-radial and non-oriented measure types. The proposed model is carefully illustrated and exemplified with some simulated data with univariate and multivariate scenarios. An application on real data using the China's high-tech innovation production process data illustrates how this model of detecting the effect of environmental variables on production frontier and efficiency of DMUs works well in practice.