The dilemma of accuracy in bankruptcy prediction: a new approach using explainable AI techniques to predict corporate crises

Abstract

Purpose Our aim is to develop a highly precise corporate crisis prediction model that surpasses previous versions, rooted in the forefront of technological advancements. Design/methodology/approach Artificial Intelligence (AI) for corporate default prediction with a novel approach based on a mix of techniques, enabling it to achieve a higher accuracy. We investigated models with sequence lengths that were both fixed and variable, and we chose the best variable sequence length model. Findings Our findings demonstrate that the artificial techniques implemented lead to very high accuracy in predicting business crises compared to previous research efforts, even those utilising long-time sequences or a high volume of observations. Research limitations/implications We highlight the key variables with a higher predictive power that need monitoring to prevent business crises. We also aim to open a new avenue of research that, starting from the use of these techniques and our results, can implement models incorporating non-accounting variables to prevent business crises. Practical implications We provide a model/tool that assesses a possible business crisis in advance through a monitoring and alert system. Policymakers can use our research’s output as a tool to combine with current credit-scoring systems and to assess the effectiveness of the new corporate crisis reforms that are upcoming in many European countries. The results of our research can be useful also to banks, public entities, and consulting firms that interact with companies and are interested in the evaluation of a firm’s financial health and stability. Originality/value Our innovative work leverages cutting-edge methodologies such as deep Recurrent Neural Networks and explainable AI. This choice is driven by the rapid evolution of AI techniques in practical application.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 19 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.