This paper presents a data-driven complex network approach, to show similarities and differences—in terms of financial risks—between the companies involved in organized crime businesses and those who are not. At this aim, we construct and explore two networks under the assumption that highly connected companies hold similar financial risk profiles of large entity. Companies risk profiles are captured by a statistically consistent overall risk indicator, which is obtained by suitably aggregating four financial risk ratios. The community structures of the networks are analyzed under a statistical perspective, by implementing a rank-size analysis and by investigating the features of their distributions through entropic comparisons. The theoretical model is empirically validated through a high quality dataset of Italian companies. Results highlights remarkable differences between the considered sets of companies, with a higher heterogeneity and a general higher risk profiles in companies traceable back to a crime organization environment.

本文提出了一种数据驱动的复杂网络方法，以显示涉嫌有组织犯罪业务的公司与未从事犯罪活动的公司之间在财务风险方面的异同。为此，我们在高度关联的公司拥有与大型实体相似的财务风险状况的假设下，构建和探索了两个网络。公司的风险状况由统计上一致的总体风险指标捕获，该指标是通过适当汇总四个财务风险比率而获得的。通过执行等级规模分析并通过熵比较研究其分布特征，从统计角度分析网络的社区结构。该理论模型已通过意大利公司的高质量数据集进行了实证验证。