企业研发效率测度与比较——以中国各地区大中型工业企业数据为例

摘要

研发效率的高低是影响企业研发决策和经济增长效率的重要因素.已有研究在选择分析方法的时候往往只是根据主观判断,并没有给出明确和客观的理由.而且由于在样本区间、指标选择和数据处理等方面存在较大差异,不同研究之间难以进行直接比较,其核算结果也不能作为判断哪种方法有效的依据.本文在相同的样本区间、指标选择和数据处理前提下,针对中国各地区大中型工业企业1996-2005年的面板数据样本,利用索洛剩余核算、随机前沿分析和数据包络分析等方法对中国企业的研发效率进行评价.通过不同指标和方法的分析比较,可以得到相对更为全面和准确的测度,从而对中国企业的研发行为及效率做出较为客观的评价.%A firm's R&D decisions, and a government's policy-making decisions need to take R&D productivity into consideration.Although the measurement of total factor productivity of R&D activities based on the knowledge production function is an effective and popular method, findings of other estimation methods remain inconsistent and elusive. In order to better understand the rigor of different estimation methods, we use different proxy indicators, such as the number of new product projects, the number of patent applications, and owners. Three estimation methods, including SRA, SFA and DEA, are adopted to analyze the same province-level panel data collected from medium- and large-sized industrial enterprises. These comparisons enable us to objectively evaluate R&D behaviors and efficiency of Chinese enterprises.Our analysis results show that the number of R&D personnel and new products are reliable indicators to address multi-product problems, and so is the DEA measurement of R&D productivity. The trend of changes in the total factor productivity is consistent across different provinces. Results found in different methods are highly correlated with each other. Another finding is that R&D efficiency decreases along with the increased number of new products, but increases with the increased number of patent applications.The most important finding is that the whole R&D efficiency level is low. The productivity gap between different provinces shows that there is room for efficiency promotion. R&D efficiency is reduced when we use the number of new products as R&D outputs. This finding corroborates with the decrease of rankings in Chinese industrial competitiveness this year. As to the contribution of input factors, we find that the human capital's output elasticity is quite lower than the physical capital's. We also find that human capital's output elasticity is much lower than physical capital's. These findings have an important policy implication; that is, an effective method to increase R&D efficiency of industrial enterprises should be adopted to sustain industrial and national economic development.