Modeling resilience to sanctions: potential for agent-based models ABBA and CANVAS in the analysis of cryptocurrency flows

Abstract

Tightening international sanctions has highlighted the need for analytical tools capable of capturing adaptive strategies of economic agents under restrictions on cross-border capital flows. Agent-based modeling (ABM) has emerged as a promising methodology and has been actively used by central banks to study macro-financial processes. This article provides a conceptual review of two prominent ABM architectures – an Agent-Based Model of the Banking System (ABBA) and the Canadian Behavioral Agent-Based Model (CANVAS) – focusing on their potential to assess sanction circumventing via cryptocurrency channels. We systematize the main features of both models (agent typologies, adaptive mechanisms, reproducibility, documentation availability), identify their limitations in the context of on-chain transactions, and outline avenues for further research. Our findings suggest that despite their methodological sophistication, neither ABBA nor CANVAS adequately incorporate the specifics of cryptocurrency infrastructures (exchanges, P2P platforms, stablecoins), which limits their applicability for analyzing the resilience of financial systems under sanctions. The contribution of the paper lies in refining the suitability criteria for ABM in sanction-related research and setting a research agenda for the development of specialized models integrating behavioral and technical aspects of crypto-financial transactions.