Structural studies of the global networks exposed in the Panama papers

Abstract

AbstractIn recent history, the Panama Papers have comprised one of the largest and most influential leaks detailing information on offshore entities, company officers and financial (and legal) intermediaries, and has led to a global exposé of corruption and tax evasion. A systematic analysis of this information can provide valuable insights into the structure and properties of these entities and the relations between them. Network science can be applied as a scientific framework for understanding the structure of such relational, heterogeneous datasets at scale. In this article, we use an existing, relational version of the Panama Papers to selectively construct various networks, and then study the properties of the underlying system using well-defined analytical methods from network science, including degree properties, country assortativity analyses, connectivity and single-point network metrics like transitivity and density. We also illustrate significant structural features in these networks by conducting a triad census and exploring the networks’ core-periphery structure. Together, these results are used to show that the Panama Papers constitute a distinct class of networks that differ significantly from ordinary social and information networks. We also propose, construct and analyze ‘higher-order’ networks from the raw data, such as a ‘social’ network of officers. We confirm that some of these higher-order networks also show significant non-random deviations from expected or typical behavior, including in their degree distributions.