Understanding corporate insolvency in Portugal: concepts, regulatory framework and analytical approaches

Purpose This paper aims to provide a structured and integrative review of the literature on insolvency prediction in small and medium enterprises (SMEs), addressing the gap between analytical modelling approaches and legal-institutional perspectives. This study focuses on the evolution of predictive models, from classical statistical techniques to more recent machine learning and data-driven methods, while examining how the Portuguese legal and institutional framework, including mechanisms such as the Special Revitalization Process (PER), influences the identification and interpretation of financial distress. Design/methodology/approach This study adopts a structured literature review approach combining a narrative review of seminal contributions with a systematic search of recent empirical studies. The search was conducted using the Scopus database, based on predefined keywords, time frame and inclusion criteria, focusing on SMEs and insolvency prediction models. The analysis integrates both international and Portuguese studies and is complemented by a comparative and conceptual synthesis, enabling a critical evaluation of modeling approaches and their interaction with legal and institutional factors. Findings This study identifies a clear evolution from traditional statistical models to more advanced machine learning approaches, while highlighting that financial ratios remain the core variables used in most models. The findings reveal a trade-off between predictive accuracy and interpretability, leading to the emergence of complementary and hybrid approaches. In addition, the results show that legal and institutional factors, particularly in the Portuguese context, influence how financial distress is defined and interpreted, affecting the signals captured by predictive models. Research limitations/implications This study is based on a structured literature review and does not involve original empirical testing, which limits the direct comparison of model performance within a unified framework. In addition, the reviewed studies rely on heterogeneous data sets, variables and validation procedures. These limitations suggest that future research should explore more standardized empirical designs and further investigate the integration of financial, qualitative and institutional variables in insolvency prediction models, particularly in SME context. Practical implications The findings suggest that insolvency prediction models can support decision-making by banks, financial institutions and managers, particularly in the early identification of financial distress in SMEs. The results highlight the importance of combining financial indicators with qualitative and institutional factors. In addition, understanding the influence of legal frameworks, such as the Portuguese Special Revitalization Process (PER), can improve the interpretation and practical application of predictive models in real-world contexts. Social implications From a public policy perspective, improved insolvency prediction can support earlier and more targeted intervention strategies, contributing to the preservation of viable firms and employment. In the Portuguese context, integrating predictive tools with legal mechanisms such as the PER may enhance the effectiveness of restructuring processes, reduce the economic and social costs of business failure and promote greater stability in SME-based economies. Originality/value This paper contributes to the literature on corporate insolvency by providing a structured and integrative review that combines legal, accounting and financial perspectives, with a particular focus on SMEs. Its originality lies in systematizing the evolution of analytical and predictive approaches, from classical statistical models to more recent machine learning and data-driven methods, while also offering a context-specific analysis of the Portuguese insolvency framework, including the Special Revitalization Process (PER). By explicitly linking predictive models with legal and institutional factors, this study adds value by clarifying both the methodological trajectories of insolvency prediction research and the context in which these models are applied.