主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2024, Vol. 32 ›› Issue (1): 1-12.doi: 10.16381/j.cnki.issn1003-207x.2021.1570

• •    下一篇

基于混频数据的中国上市公司财务困境动态预测研究

闫达文1,李存2,迟国泰2()   

  1. 1.大连理工大学数学科学学院, 辽宁 大连 116024
    2.大连理工大学经济管理学院, 辽宁 大连 116024
  • 收稿日期:2021-08-09 修回日期:2022-01-07 出版日期:2024-01-25 发布日期:2024-02-08
  • 通讯作者: 迟国泰 E-mail:chigt@dlut.edu.cn
  • 基金资助:
    国家自然科学基金项目(72271040);教育部人文社会科学研究规划基金项目(22YJAZH125);来华留学研究课题重点项目(DUTLHLX202304)

Dynamic Financial Distress Prediction for Chinese Listed Companies Based on the Mixed Frequency Data

Dawen Yan1,Cun Li2,Guotai Chi2()   

  1. 1.School of Mathematical Sciences, Dalian University of Technology, Dalian 116024, China
    2.School of Economics and Management, Dalian University of Technology, Dalian 116024, China
  • Received:2021-08-09 Revised:2022-01-07 Online:2024-01-25 Published:2024-02-08
  • Contact: Guotai Chi E-mail:chigt@dlut.edu.cn

摘要:

本文以年度财务指标、年度失业率、季度GDP增长率以及月度通货膨胀率为解释变量,以年频的财务困境状态变量为响应变量,结合指数Almon多项式赋权和逻辑回归方法,构建了不同时间窗口(t-k年)的中国上市公司财务困境低频预测模型。进一步地,本文捕捉了上市公司摘戴帽的季度时间信息,将财务困境状态设置为季频变量,又构建了中国上市公司财务困境的高频预测模型,揭示混频宏观和财务因素对企业未来每季度发生财务危机的预警功能。本研究创新和特色在于:构建年度的失业率、季度GDP增长率、月度CPI增长率与企业财务困境状态的非线性函数关系,反映不同频率的关键宏观因素变化对企业清偿能力的影响。本文研究结果表明:(1)通过引入高频宏观经济指标,能有效提高中国上市制造业企业财务困境的年度预测表现。以t-3年模型为例,混频模型相比同频的年度数据模型的AUC值提高了7.32%。(2)不同频率的关键宏观因素在不同年度的预测表现不同。月度的通货膨胀率仅在t-2年模型中具有显著预测功能。季度的GDP增长率、年度的失业率在不同窗口下的模型中,均具有统计意义下的显著影响。(3)与低频的年度财务困境预测相比,年度失业率指标不再对企业季度财务困境状况的变化具有显著影响,而行业景气指数却表现出明显的预测功能。季度GDP增长率和月度通货膨胀率数据的经济意义更加明显,在企业财务困境风险的季度高频预测中,表现出更强的时效性。

关键词: 财务困境预警, 混频数据, 混频数据抽样方法, 高频宏观经济变量, 中国上市公司

Abstract:

In this paper, a mixed-frequency data driven financial distress prediction problem is considered. The exponential Almon weighting scheme of mixed-data sampling is introduced in the context of a Logistic regression, which allows for an individual weighting of high-frequent increments, hence distinguishing the importance of different macroeconomic period to explain the change of a listed companies’ financial status. In applying the suggested model to the period from 2007 to 2017, a total data set is used that includes 4 macroeconomic indicators of mixed annual, quarterly and monthly frequency and 15 potentially significant financial indicators of annual frequency for 350 Chinese listed manufacturing companies. The results show that (1) the model correctly classifies more financially distressed companies than classic Logistic model, in particular for forecasting horizons of 2-3 years. (2) The macroeconomic indicators observed at different frequency show their importance for both low frequency (annually) and high frequency (quarterly) financial distress prediction, while quarterly GDP growth rate and monthly inflation rate have more significant influence, both statistically and economically, on quarterly financial distress risk. All findings indicate that this paper provides a suitable approach incorporating mixed-frequency data for prediction of financial distress for listed companies in China.

Key words: financial distress prediction, mixed-frequency data, mixed data sampling method, high- frequency macroeconomic indicators, Chinese listed companies

中图分类号: