基于稳健回归的经济增长数据可靠性评估

Reliability Evaluation of Economic Growth Data Based on Robust Regression

为了顺应改革开放以来经济规模和结构的不断调整,我国的统计体系发生了较大的变化.部分经济指标在不同层面的汇总结果存在差异,导致一些学者和组织对我国公布的经济增长数据质量提出质疑.因此,对我国经济增长数据的可靠性进行检测,成为学界持续关注的热点话题.在过去的研究中,很多学者使用了传统的回归方法,但这些方法容易受到异常值的影响,造成结果的可靠性较低.本文提出一种基于MRCD估计和MM估计的稳健回归方法,使用2019年中国内地31个省级行政区域的GDP增长率和14个经济增长相关指标的增长率数据对中国的经济数据质量进行了评估.研究结果表明,该模型不仅提高了对异常值的识别能力,还降低了异常值对回归估计值的影响,因而同时提高了结果的可靠性和实际应用能力.实证结果表明,我国的经济增长数据是有质量保证的.

In order to adapt to the continuous adjustment of economic scale and structure since the reform and opening up,China's statistical system has undergone major changes.The differences in the aggregate results of some economic indicators at different levels have led some scholars and organizations to question the quality of my country's published economic growth data.Therefore,testing the reliability of my country's economic growth data is a hot topic that academia continues to pay attention to.In past studies,many scholars have used traditional regression methods,but these methods are susceptible to outliers,resulting in low reliability of the results.This paper proposes a robust regression method based on MRCD estimation and MM estimation,and uses the GDP growth rate of 31 provincial administrative units in China's mainland in 2019 and the growth rate data of 14 related indicators to evaluate the quality of China's economic data.The research results show that the model not only improves the ability to identify outliers,but also reduces the infuence of outliers on regression estimates,thus improving the reliability of the results and practical application capabilities simultaneously.The result of empirical analysis shows that the economic growth data of China is of quality assurance.