Abstract: The government supports the corporation innovation through direct R&D subsidies, tax exemption and policy inducement. However, there is litter literature pay attention to the efficiency of government R&D support. This article proposes a three-stage DEA model based on Super-SBM method to measure the efficiency of government R&D support and find out that technological and operational capability among factors of enterprises as well as characteristics of corporations among factors of external environment exert tremendous influences on the input and/or output slacks. With the help from econometrics technique, we shed more light on the effect of enterprises’ R&D expenditure on the efficiency of government R&D support. After controlling the ratio of government R&D subsidies to enterprises’ R&D input, the R&D input itself perform positive influence on efficiency of government R&D support, while the ratio plays a negative moderating role in this relationship. it concludes by presenting that government should the ratio lower than a certain value; only if the R&D input lower than the critical value can the government increase that ratio favorably.

摘要： 政府通过R&D补贴、税收减免和政策引导扶持企业创新，但政府R&D扶持效率很少被深入研究。本文提出基于Super-SBM的三阶段DEA模型测度政府R&D扶持效率，研究发现企业因素中的技术能力和经营能力以及外部环境因素中的企业特征对投入、产出松弛变量产生了较大影响。利用计量回归分析企业R&D投入对政府R&D扶持效率的影响，发现企业的R&D支出对政府R&D扶持效率具有正影响，但政府补贴占企业R&D投入比例对其产生负向调节。这表明政府必须保证R&D补贴不能超过一定比例限额；只有当企业R&D水平在临界值以下时，政府提高这一比例才是有益的。

关键词: 政府R&D扶持效率, 企业因素, 环境因素, Super-SBM方法, efficiency of government R&D support, corporation factors, external environment factors, Super-SBM method