GDP统计数据质量实证研究:基于Benford法则和空间面板模型

Empirical Study on the Data Quality of GDP:Benford's Law and Spatial Panel Model

GDP统计数据的质量问题一直是舆论的焦点。本文运用Benford法则,对1993-2016年中国31个省域GDP数据的前三位数字的真实性进行检验;创新性地设计了税收、货运量、电力消耗量以及贷款发放量的评价指标体系,然后借助空间面板模型,以1995-2016年GDP数据质量为因变量进行逻辑匹配检验。结果表明:(1)从Benford法则的4种检验方法来看,1993-2016年各省区GDP数据的首位数字和第三位数字的4种检验结果呈现不一致性;(2)从空间面板模型的逻辑匹配结果来看,1995-2016年的GDP数据与4大评估指标之间匹配性较好,基本符合经济发展的规律。因此,相关GDP核算部门需要注意各省市区GDP数据的第一位和第三位数字的质量,保证GDP数据的准确真实。

The data quality of GDP has always been the focus of public opinion.In this paper,the first three figures of GDP data from 1993 to 2016 are tested according to the Benford’s law.Moreover,the evaluation index system of tax revenue,freight volume,electricity consumption and loan amount is innovatively designed.The spatial panel model was used to check the quality of GDP data from 1995 to2016.The results show that:(1)From the four test methods of Benford’s law,the first and third digits of GDP data in the provinces from 1993 to 2016 are inconsistent;(2)From the logical matching results of the spatial panel model,the matching between GDP data from 1995 to 2016 and the 4 evaluation indexes is relatively good,which is basically in line with the laws of economic development.Therefore,the relevant department of discipline inspection needs to focus on the first and third digits of the GDP data of the provinces to strictly verify it and ensure the accuracy of GDP data.