中文摘要：

本文基于1998-2005年中国1813家研究所构成的均衡面板数据,首次对中国研究机构的科技资源利用效率进行了评价研究。在采用FE和IVE方法控制双向固定效应和潜在的选择性偏误的基础上,估计了研究所投入要素对总收入和科技收入的贡献份额,由大到小依次为：非研究生科技人员、科研业务费、研究生科技人员、科研设备费,研究所的科技投入总体上不存在规模效应。采用GPS方法控制了投入要素的条件概率密度分布,估计了研究所投入要素对边际产出的动态影响,研究发现,随着投入要素规模的不断增大,非研究生科技人员的边际产出呈先升后降的趋势,其他三种投入要素则呈现总体上升趋势。本文的研究结论对研究所提高投入要素的利用效率具有重要的启示和指导意义。

英文摘要：

For the first time, we evaluate the efficiency of S＆T inputs of China＇ s research institutes （RIs） based on a panel dataset consisting 1813 RIs from 1998 to 2005. By controlling the individual and time fixed effects, the instrumental variable estimate shows that S＆T personnel without graduate degree has the highest output elasticity, which is followed by S＆T intermediate inputs, S＆T personnel with graduate degree, and S＆T equipment. In general, the constant or increasing return to scale doesn＇t exist in the sector of China＇ s RIs. Through including the conditional density of S＆T input in the GPS regression, the input-varying output elasticity and marginal products can be evaluated. For S＆T personnel without advanced degrees, medium-size institutes exhibit somewhat higher marginal products. Institutes with the largest inputs of S＆T personnel with graduate degrees, equipment, and intermediate inputs exhibit the highest returns to each of these inputs. Several suggestions on how to improve the S＆T inputs＇ efficiency of RIs are provided.