基于首末位质量因子的BP神经网络财务风险预警模型  

Financial Early Warning Model of BP Neural Network Based on the First and the Last Quality Factors

作　　者：杨贵军 杜飞[1] 贾晓磊 Yang Guijun;Du Fei;Jia Xiaolei(School of Statistics,Tianjin University of Finance and Economics,Tianjin 300222,China;China Center for Economic Statistics Research,Tianjin University of Finance and Economics,Tianjin 300222,China)

机构地区：[1]天津财经大学统计学院,天津300222 [2]天津财经大学中国经济统计研究中心,天津300222

出　　处：《统计与决策》2022年第3期166-171,共6页Statistics & Decision

基　　金：国家社会科学基金重点项目(20ATJ008);国家社会科学基金青年项目(18CTJ008);天津市哲学社会科学规划重点课题(TJTJ19-001);天津市自然科学基金青年项目(18JCQNJC69600);内蒙古纪检监察大数据实验室开放课题(IMDBD202002;IMDBD202004)。

摘　　要：上市公司往往存在粉饰财务数据来美化企业经营状况的动机,这会降低财务风险预警模型预测的准确性。文章利用Benford律和Myer指数两种数据质量评估方法,构建Benford和Myer质量因子,引入BP神经网络模型,构造BM-BP神经网络财务风险预警模型;并进一步利用2000—2019年中国A股上市公司数据,评价数据质量因子对财务风险预警模型预测准确性的影响,分析新模型预测准确性的稳定性。实证分析结果显示:Benford和Myer质量因子提高了BP神经网络财务风险预警模型预测的准确性;在不同质量因子的比较结果中,包含评选指标Benford和Myer质量因子的BP神经网络财务风险预警模型具有较高的预测准确率和较低的二类误判率,稳定性良好;利用决策树算法筛选指标有效提高了新模型的预测准确性。Listed companies often have the incentive to embellish their financial data to beatify their business, which reduces the accuracy of financial risk warning models. This paper uses the two data quality assessment methods of Benford’s law and Myer index to construct Benford and Myer quality factors, bringing in BP neural network model to establish BM-BP neural network financial early warning model. Then the paper further uses the data of Chinese A-share listed companies from 2000 to 2019 to evaluate the impact of data quality factors on the accuracy of financial risk warning model prediction, and analyzes the stability of the prediction accuracy of the new model. The Empirical analysis results show that Benford and Myer quality factors improve the accuracy of BP neural network financial risk early warning model, that among the comparison results of different quality factors, the BP neural network financial risk early warning model including the selection index Benford and Myer quality factor has higher prediction accuracy and lower second-class misjudgment rate, with good stability, and that the prediction accuracy of the new model is improved effectively by using decision tree algorithm to screen the indexes.

关 键 词：财务风险预警 数据质量因子 Benford律 Myer指数 BP神经网络 

分 类 号：F830.9[经济管理—金融学] O211.3[理学—概率论与数理统计]