Accounting Information Usefulness in Credit Risk Evaluation by Banks

Abstract

This research investigates the evolving utility of traditional accounting information in the context of bank credit risk evaluation, proposing a novel, hybrid analytical
framework that integrates conventional financial statement analysis with emerging,
non-parametric pattern recognition techniques inspired by ecological succession models. While accounting data has long been the cornerstone of credit assessment, its
standalone predictive power in contemporary, volatile economic environments is increasingly questioned. This study posits that the diminishing marginal informativeness of historical cost-based accounting figures can be counteracted by a synergistic
methodology that re-contextualizes them within a dynamic, systems-based model of
firm resilience and adaptive capacity. Our primary research questions explore: (1) To
what extent can a bio-inspired, successional analysis of financial statement sequences
enhance the predictive accuracy of default risk beyond standard ratio analysis? (2)
How does the integration of qualitative, narrative disclosures—processed through a
novel semantic trajectory mapping—alter the weight assigned to quantitative accounting metrics in a holistic risk score? We develop and test the Ecological Succession Credit
Evaluation Model (ESCEM), which treats a firm’s multi-period financial statements as
an ecological community, identifying successional stages (pioneer, establishment, climax, disturbance) and calculating resilience indices based on the volatility and recovery
patterns of key accounting variables. Concurrently, we apply trajectory divergence algorithms to management commentary to assess strategic consistency. Testing on a
longitudinal dataset of corporate borrowers from 1995-2004, our results demonstrate
that the ESCEM framework improves default prediction accuracy by approximately
18% compared to a benchmark Z-score model, with the most significant gains observed
for firms in transitional or distressed successional stages. The findings challenge the
prevailing dichotomous view of accounting information as either useful or obsolete, arguing instead for its reconceptualization as a core input into a more complex, adaptive
system of firm health. This represents a fundamental shift from static, point-in-time
assessment to a dynamic, process-oriented evaluation of creditworthiness, offering a
novel pathway for enhancing the robustness of bank lending decisions in the face of
economic uncertainty