Scholarlink

<p>The structural dominance of Chain Leader firms is intrinsically linked to the compliance risks of their upstream and downstream partners, yet the cross-organizational transmission of financial fraud remains underexplored. Using a sample of Shanghai and Shenzhen A-share listed firms from 2010 to 2022, this study identifies Chain Leaders via social network analysis and employs the LightGBM algorithm with SHAP interpretability to detect predictive signals linking Chain Leader attributes and executive traits to financial fraud. The findings reveal: (1) Nonlinear Risk Profiles: Machine learning captures critical threshold and interaction effects where linear models fail; (2) Predictive Asymmetry: Customer-side models exhibit superior performance, reflecting robust risk signals driven by rigid resource dependence; (3) Signal Hierarchy: Firm-level structural attributes dominate risk identification, with executive traits playing a secondary, contextual role; (4) Divergent Indicators: Supplier risks are signaled by “structural power” metrics (e.g., cooperation concentration, analyst coverage, market power), whereas customer risks are anchored in “operational state” (e.g., operational efficiency, financial leverage, external opinion).By leveraging external Chain Leader information to penetrate the opacity of internal financial manipulation, this study extends fraud research into cross-organizational contexts. The findings offer actionable pathways for regulators to construct early-warning systems based on Chain Leader attributes, and for supply chain participants to optimize risk-sharing and information coordination.</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

Uncovering cross-organizational risk patterns: a machine learning approach to predicting financial fraud via chain leader attributes

  • 期刊论文

  • 发表日期：

    2026年4月3日

查看全文
  • Linjing Yang

摘要

The structural dominance of Chain Leader firms is intrinsically linked to the compliance risks of their upstream and downstream partners, yet the cross-organizational transmission of financial fraud remains underexplored. Using a sample of Shanghai and Shenzhen A-share listed firms from 2010 to 2022, this study identifies Chain Leaders via social network analysis and employs the LightGBM algorithm with SHAP interpretability to detect predictive signals linking Chain Leader attributes and executive traits to financial fraud. The findings reveal: (1) Nonlinear Risk Profiles: Machine learning captures critical threshold and interaction effects where linear models fail; (2) Predictive Asymmetry: Customer-side models exhibit superior performance, reflecting robust risk signals driven by rigid resource dependence; (3) Signal Hierarchy: Firm-level structural attributes dominate risk identification, with executive traits playing a secondary, contextual role; (4) Divergent Indicators: Supplier risks are signaled by “structural power” metrics (e.g., cooperation concentration, analyst coverage, market power), whereas customer risks are anchored in “operational state” (e.g., operational efficiency, financial leverage, external opinion).By leveraging external Chain Leader information to penetrate the opacity of internal financial manipulation, this study extends fraud research into cross-organizational contexts. The findings offer actionable pathways for regulators to construct early-warning systems based on Chain Leader attributes, and for supply chain participants to optimize risk-sharing and information coordination.