Scholarlink

<p>The data uncertainty in financial statements limits the reliability of assessments of a company’s financial condition, which is a challenge widely acknowledged in financial context. This study pioneers the application of robust support vector machines (RSVMs) to model the complex uncertainties in financial reporting that may involve both random errors and systematic distortions through tailored uncertainty sets within a robust optimization framework. Empirically, we apply our method to predict financial distress (FD) of Chinese listed manufacturing companies, analyzing how uncertainty set configurations affect RSVMs performance. The selected optimal form of uncertainty sets, varying across the different prediction horizons illustrate that short-term FD forecasts are more sensitive to ST company data noise, while longer horizons expose systemic gaps between reported and actual financial data, suggesting historical disclosure inflation. Further, comparative analysis demonstrates that RSVMs outperform all benchmark models&#xa0;that include classical <i>L</i><sub>1</sub>-norm SVM and its major noise-anti variants in terms of predictive accuracy and stability, and the performance advantage widens with longer forecast horizon.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Prediction of Financial Distress for Chinese Listed Manufacturing Companies given Uncertain Financial Accounting Data Environments

  • 期刊论文

  • 发表日期：

    2025年11月19日

查看全文
  • Dawen Yan,
  • Manran Zhang

摘要

The data uncertainty in financial statements limits the reliability of assessments of a company’s financial condition, which is a challenge widely acknowledged in financial context. This study pioneers the application of robust support vector machines (RSVMs) to model the complex uncertainties in financial reporting that may involve both random errors and systematic distortions through tailored uncertainty sets within a robust optimization framework. Empirically, we apply our method to predict financial distress (FD) of Chinese listed manufacturing companies, analyzing how uncertainty set configurations affect RSVMs performance. The selected optimal form of uncertainty sets, varying across the different prediction horizons illustrate that short-term FD forecasts are more sensitive to ST company data noise, while longer horizons expose systemic gaps between reported and actual financial data, suggesting historical disclosure inflation. Further, comparative analysis demonstrates that RSVMs outperform all benchmark models that include classical L1-norm SVM and its major noise-anti variants in terms of predictive accuracy and stability, and the performance advantage widens with longer forecast horizon.