企業財務危機預警模型— 集團關聯性指標對於財務危機之影響

Abstract

本研究探討企業集團關聯性指標與財務危機預測之關聯，以臺灣經濟新報(TEJ)資料庫中2011至2021年的企業資料為樣本，運用特徵值分解(Eigenvalue decomposition)方法，透過企業集團內部各公司營業毛利率的相關係數矩陣提取最大特徵值作為集團關聯程度的量化指標。實證結果顯示，集團關聯程度與財務危機發生機率呈顯著負相關，即集團關聯程度越高，企業發生財務危機的可能性反而越低。此發現與Khanna and Yafeh (2005)關於台灣企業集團具風險分攤特性的觀察相符，支持Myers and Majluf (1984)與Fazzari et al. (1987)提出的內部資本市場理論。本研究進一步透過敏感性分析證實此結果對不同窗期長度與標準化方法的穩健性，並採用產業固定效果檢測與集群標準誤等方法驗證其一致性。研究發現拓展了Billio et al. (2012)的系統風險理論至非金融企業集團，為投資者、債權人和監管機構提供更全面的風險評估工具，同時為企業集團風險管理策略的制定提供理論支持。

This study investigates the relationship between intra-group financial correlation and financial distress prediction by analyzing a sample of companies from the Taiwan Economic Journal (TEJ) database spanning from 2011 to 2021. Using eigenvalue decomposition methodology, the study extracts the maximum eigenvalue from the correlation matrix of gross profit margins across group companies as a quantitative indicator of intra-group correlation. Empirical results demonstrate a significant negative relationship between group correlation and the probability of financial distress, indicating that higher group correlation is associated with lower likelihood of financial crisis. This finding aligns with Khanna and Yafeh's (2005) observations on the risk-sharing characteristics of Taiwanese business groups and supports the internal capital market theory proposed by Myers and Majluf (1984) and Fazzari et al. (1987). Through sensitivity analyses, the study confirms the robustness of these results across different time windows and standardization methods, and verifies their consistency using Heckman selection models, industry fixed effects, and clustered standard errors. The findings extend Billio et al.'s (2012) systemic risk theory to non-financial business groups, providing investors, creditors, and regulatory authorities with more comprehensive risk assessment tools while offering theoretical support for business group risk management strategies.