Abstract: Based on threshold panel data model with industrial enterprises in China, this paper analyzed the non-linear effects of R&D intensity on firm performance. R&D promotes firm performance only if R&D intensity exceeds the first threshold. While R&D intensity exceeds the second threshold, increasing R&D can hardly bring performance growth. The threshold effects of R&D intensity on firm performance vary across industries. The optimum interval of R&D intensity ranges from 1.1% to 4.13% to effectively promote TFP, and 1.17%~4.16% to promote ROA in high-tech industries. This optimum interval for low-tech industries ranges from 0.74% to 3.73% to effectively promote TFP, and 0.93% ~ 3.8% to promote ROA. R&D intensity for 51.46% enterprises involved in R&D activity is still smaller than the first threshold. Therefore, encouraging R&D investment and promoting R&D intensity should still be an important goal for S&T policy in China.

摘要： 基于我国工业企业数据，使用门槛面板数据模型，研究了研发投入强度对企业绩效的非线性影响：研发投入强度只有达到第一门槛值时，才能对企业绩效产生显著促进作用；而超过第二门槛值时，对企业绩效的作用变得不明显；研发投入强度对企业绩效影响的门槛效应存在行业差异。最佳的研发投入强度区间是：对于高技术行业而言，研发强度处于1.1%~4.13%时，研发对资产收益率的促进作用最明显；研发强度处于1.17%~4.16%时，研发对全要素生产率的促进作用最明显；而对于低技术行业，研发强度在0.74%~3.73%时，研发对资产收益率的促进作用最强；而在0.93%~3.8%时，对全要素生产率的促进作用最强。我国有研发活动的企业中，超过51.46%的企业，研发投入强度仍低于第一门槛值。因此，引导企业研发创新，提高研发投入强度仍将是我国科技政策的重要目标。

关键词: 研发投资, 企业绩效, 门槛效应, R&D, Firm Performance, Threshold Effects