Frame of the research. Strategic management increasingly conceptualizes firm default as a cumulative and path dependent process of organizational decline. Within this perspective, early warning depends on the ability to detect weak and evolving signals of vulnerability, yet this remains challenging in contexts characterized by uncertainty and rare adverse events. Purpose of the paper. This study conceptualizes unsupervised anomaly detection as an analytical mechanism that supports strategic sensing. Rather than predicting default outcomes, it identifies persistent deviations from financial normality, making emerging vulnerability observable before formal failure occurs. Methodology. The analysis is based on a longitudinal sample of 32,801 Italian SMEs (2016-2024). A rolling three year window design is implemented to capture medium-term dynamics, and three complementary unsupervised anomaly detection techniques are applied to accounting data without relying on prior default labels. Results. The results show that anomaly scores are temporally stable and identify firms whose financial deviations persist over time. These firms exhibit substantially higher ex post default rates, indicating that deviations from financial normality provide a meaningful signal of emerging risk. Research limitations. The study focuses on Italian SMEs and accounting data. Future research may extend the analysis to different institutional contexts and incorporate non-financial indicators to capture additional dimensions of vulnerability. Managerial implications. The findings suggest that anomaly detection can support early warning systems by identifying firms that warrant attention before deterioration becomes observable through formal outcomes. Originality of the paper. The study shifts the focus from default prediction to anomaly-based sensing, offering a process-consistent empirical approach to financial vulnerability and providing a concrete operationalization of strategic sensing under conditions of limited labeled data.
Detecting Strategic Decline: Anomaly Detection as an Early Warning System for Corporate Default in SMEs / Ariante, I., Cavicchia, C.. - (2026), pp. 381-392. (Sinergie-SIMA Management Conference 2026 ).
Detecting Strategic Decline: Anomaly Detection as an Early Warning System for Corporate Default in SMEs
Irene Ariante
Primo
;Carlo CavicchiaUltimo
2026
Abstract
Frame of the research. Strategic management increasingly conceptualizes firm default as a cumulative and path dependent process of organizational decline. Within this perspective, early warning depends on the ability to detect weak and evolving signals of vulnerability, yet this remains challenging in contexts characterized by uncertainty and rare adverse events. Purpose of the paper. This study conceptualizes unsupervised anomaly detection as an analytical mechanism that supports strategic sensing. Rather than predicting default outcomes, it identifies persistent deviations from financial normality, making emerging vulnerability observable before formal failure occurs. Methodology. The analysis is based on a longitudinal sample of 32,801 Italian SMEs (2016-2024). A rolling three year window design is implemented to capture medium-term dynamics, and three complementary unsupervised anomaly detection techniques are applied to accounting data without relying on prior default labels. Results. The results show that anomaly scores are temporally stable and identify firms whose financial deviations persist over time. These firms exhibit substantially higher ex post default rates, indicating that deviations from financial normality provide a meaningful signal of emerging risk. Research limitations. The study focuses on Italian SMEs and accounting data. Future research may extend the analysis to different institutional contexts and incorporate non-financial indicators to capture additional dimensions of vulnerability. Managerial implications. The findings suggest that anomaly detection can support early warning systems by identifying firms that warrant attention before deterioration becomes observable through formal outcomes. Originality of the paper. The study shifts the focus from default prediction to anomaly-based sensing, offering a process-consistent empirical approach to financial vulnerability and providing a concrete operationalization of strategic sensing under conditions of limited labeled data.| File | Dimensione | Formato | |
|---|---|---|---|
|
SinergieSIMA-Conference-Proceedings-2026-2-405-416.pdf
solo utenti autorizzati
Tipologia:
Versione Editoriale (PDF)
Licenza:
Accesso privato/ristretto
Dimensione
305.94 kB
Formato
Adobe PDF
|
305.94 kB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


