Book

In today's rapidly changing and complex business environment, the ability to predict business failure has become a critical issue for stakeholders, including investors, creditors, regulators, and managers. Business failure can have widespread and often devastating consequences, including job losses, reduced shareholder value, disruption of supply chains, and broader economic instability. The collapses of major corporations such as Enron, Lehman Brothers, and WorldCom have clearly demonstrated the importance of timely identification of business failure risks. As a result, business failure prediction models have attracted increasing attention in academic and professional circles, providing valuable tools for decision-makers to reduce risk and prevent corporate crises (Chen, 2024).

Research on business failure prediction models dates back to the 1960s, beginning with univariate techniques. Over time, with methodological improvements, multivariate techniques, conditional probability methods, and artificial intelligence approaches were gradually introduced. These developments have led to gradual improvements in the results and reliability of prediction models (Romero Martinez et al., 2025).

However, traditional business failure prediction models, which are largely based on financial ratios such as Altman's Z-score or Ohlson's O-score, often fail to fully capture indicators that signal business crises. These models primarily rely on historical financial data and frequently overlook non-financial factors that contribute to corporate decline, such as accounting anomalies and corporate governance mechanisms (Nour et al., 2024). While these dimensions of corporate health may be subtle, they can provide early warning signals that financial ratios alone are unable to detect. For example, accounting anomalies such as revenue manipulation and weak governance structures like inadequate board oversight can indicate deep structural issues that may lead to failure (Hartati et al., 2022).