大中小微企业规模划型统计标准的实证研究  

An Empirical Study on the Statistical Standard for the Size Classification of Large,Medium,Small and Micro-sized Enterprises

作　　者：高敏雪[1,2] 孙庆慧 胡强 甄峰[1,2] Gao Minxue;Sun Qinghui;Hu Qiang;Zhen Feng(Center for Applied Statistics,Renmin University of China;School of Statistics.Renmin University of China)

机构地区：[1]中国人民大学应用统计科学研究中心 [2]中国人民大学统计学院

出　　处：《数量经济技术经济研究》2022年第2期164-180,F0003,共18页Journal of Quantitative & Technological Economics

摘　　要：研究目标:引入机器学习方法讨论大中小微企业规模划型标准问题。研究方法:基于国家统计局公开的第四次经济普查企业数据,采用Adasampling方法,结合基本描述统计方法判断企业规模划型问题。研究发现:多数行业规模划型采用单一指标可达到较好划分效果,缓解小微企业划分过于宽松的痼疾,简化分类识别难度;现行小微企业分类阈值较为合理,但中小企业间和大中企业间分类阈值有明显偏差;可细化当前的企业规模划型标准,基于行业大类特征进行企业规模划型;分类阈值可建立机器学习与经济普查相结合的定期动态调整机制。研究创新:引入Adasampling方法讨论企业规模划型指标选取和划分阈值标准。研究价值:从理论上论证我国企业规模划型标准的科学性和优化方向。Research Objectives: To measure the statistical standard for the size classification of large, medium, small and micro-sized enterprises. Research Methods: By using the micro data of enterprises in the fourth National Economic Census, to discuss the topic with adasampling method and descriptive statistical methods. Research Conclusions: Firstly, a proper single indicator can achieve a better classification effect, which can help alleviate the chronic disease of too loose identification of small and micro-sized enterprises. Secondly, the measurement results verify the rationality of the classification thresholds of small and micro-sized enterprises in the current standard, but there are obvious deviations in the classification thresholds of medium and small-sized enterprises and large and medium-sized enterprises. Thirdly, it is suggested that the current enterprise size classification standard can be specified and classified based on the characteristics of 2-digit level industries, and a regular dynamic adjustment mechanism can be funded by combining Machine Learning Methods and the National Economic Census. Research Innovations: To select and measure classification indicators and the determination of classification thresholds among large, medium, small and micro-sized enterprise by using adasampling method. Research Values: To support official standard and its rearrangement of the size classification of large, medium, small and micro-sized enterprises in theory and statistical methods.

关 键 词：企业规模划型 大中小微企业划分标准 小微企业 Adasampling方法 

分 类 号：F222.1[经济管理—国民经济]