Uncovering financial crime patterns: a clustering approach to offender typologies

Abstract

Purpose As financial crimes continue to expand globally, examining this phenomenon to understand its mechanisms, enhance prevention and detection strategies and mitigate its impact has become essential. Prior research attempted to identify the socio-professional characteristics of white-collar criminals who commit occupational fraud using traditional statistical techniques. While powerful, these methods rely on manually identifying offender types. This task becomes more complex as more variables are taken into consideration. This study aims to use more advanced techniques to analyze criminal characteristics. Design/methodology/approach In this study, financial criminals data is collected from the Court of Appeal in Tunis, Tunisia. Each observation contains the crime type and six socio-professional factors, namely, age, marital status, education level, hierarchical level, seniority and shareholding of the criminals within the victim company. This paper uses Artificial Intelligence (AI) techniques to discover hidden patterns and construct nuanced typologies. Each type is then associated with certain crime using a newly introduced Affinity Score. The study findings are then validated using chi-square test of independence. Findings Five unique offender types related to occupational fraud were identified, each associated with particular crime. These types reveal more nuanced associations compared to previous research, offering insights into the correlation between socio-professional characteristics and criminal behavior. Originality/value This research offers a new AI-driven framework to reveal criminal characteristics. This approach overcomes the limitations of traditional manual clustering by enabling a data-driven, systematic identification of high-risk offender groups and offers new insights into how organizational context, role and career stage shape criminal behavior.