Purpose: This paper investigates the application of Machine Learning (ML) models to predict insolvency in Italian SMEs. The study identifies key factors leading to financial distress, aiming to equip managers and policymakers with tools for timely intervention to prevent insolvency and ensure business continuity. Methodology: The study employs ML models, including Logistic Regression, XGBoost, Gradient Boosting, and Random Forest, to analyze extensive financial and non-financial data from 10,000 Italian SMEs. Findings: ML models, particularly XGBoost with an accuracy rate of 0.87, significantly outperform traditional methods in predicting SMEs insolvency. The study emphasizes the importance of model interpretability, ensuring data-driven insights are actionable and comprehensible for managers. Key predictive features include "Number of current representatives & managers," "CIAA Number," and "Structural Margin." Research Limitations/Implications: The reliance on historical data from 10,000 Italian SMEs may not capture all relevant variables or industry-specific nuances. Also, models may require adjustments for different regions or industries. Ethical concerns such as data biases and model transparency also pose challenges. Managerial implications are profound. ML models provide tools for continuous financial health monitoring and early distress detection, enhancing resilience, business continuity, and stakeholder protection. Academically, this study advances predictive analytics by demonstrating the efficacy of ML in insolvency prediction and encourages an interdisciplinary approach. Originality/Value: This study underscores the practical benefits of ML for insolvency and crisis prediction, offering Italian SMEs managers robust tools to enhance decision-making, mitigate risks, and promote sustainable growth. By leveraging AI-driven insights, managers can, in fact, better monitor performance, make informed decisions, and ensure business resilience.
Too Small to Fail? Leveraging AI for Early Insolvency Detection in Italian SMEs / LO CONTE, DAVIDE LIBERATO; Antonini, Valerio; Sancetta, Giuseppe. - (2024). (Intervento presentato al convegno 27th Excellence in Services International Conference (EISIC) tenutosi a Bergamo, Italy).
Too Small to Fail? Leveraging AI for Early Insolvency Detection in Italian SMEs
davide liberato lo conte
;giuseppe sancetta
2024
Abstract
Purpose: This paper investigates the application of Machine Learning (ML) models to predict insolvency in Italian SMEs. The study identifies key factors leading to financial distress, aiming to equip managers and policymakers with tools for timely intervention to prevent insolvency and ensure business continuity. Methodology: The study employs ML models, including Logistic Regression, XGBoost, Gradient Boosting, and Random Forest, to analyze extensive financial and non-financial data from 10,000 Italian SMEs. Findings: ML models, particularly XGBoost with an accuracy rate of 0.87, significantly outperform traditional methods in predicting SMEs insolvency. The study emphasizes the importance of model interpretability, ensuring data-driven insights are actionable and comprehensible for managers. Key predictive features include "Number of current representatives & managers," "CIAA Number," and "Structural Margin." Research Limitations/Implications: The reliance on historical data from 10,000 Italian SMEs may not capture all relevant variables or industry-specific nuances. Also, models may require adjustments for different regions or industries. Ethical concerns such as data biases and model transparency also pose challenges. Managerial implications are profound. ML models provide tools for continuous financial health monitoring and early distress detection, enhancing resilience, business continuity, and stakeholder protection. Academically, this study advances predictive analytics by demonstrating the efficacy of ML in insolvency prediction and encourages an interdisciplinary approach. Originality/Value: This study underscores the practical benefits of ML for insolvency and crisis prediction, offering Italian SMEs managers robust tools to enhance decision-making, mitigate risks, and promote sustainable growth. By leveraging AI-driven insights, managers can, in fact, better monitor performance, make informed decisions, and ensure business resilience.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
