This paper studies the effects that connectivity and centralisation have on the response of interbank networks to external shocks that generate phenomena of default contagion. We run numerical simulations of contagion processes on randomly generated networks, characterised by different degrees of density and centralisation. Our main findings show that the degree of robustness-yet-fragility of a network grows progressively with both its degree of density or centralisation, although at different paces. We also find that sparse and decentralised interbank networks are generally resilient to small shocks, contrary to what so far believed. The degree of robustness-yet-fragility of an interbank network determines its propensity to generate a too-many-to-fail problem. We argue that medium levels of density and high levels of centralisation prevent the emergence of a too-many-to-fail issue for small and medium shocks whilst drastically creating the problem in the case of large shocks. Finally, our results shed some light on the actual robustness-yet-fragility of the observed core-periphery national interbank networks, highlighting the existing risk of systemic crises.

本文研究了连通性和中心化对银行间网络对产生违约蔓延现象的外部冲击的反应的影响。我们在随机生成的网络上运行传染过程的数值模拟，其特点是不同程度的密度和中心化。我们的主要发现表明，网络的稳健性和脆弱性程度随着其密度或集中程度的增加而逐渐增加，尽管速度不同。我们还发现，稀疏和分散的银行间网络通常对小冲击具有弹性，这与迄今为止的看法相反。银行间网络稳健而又脆弱的程度决定了它产生“太多而不能倒”问题的倾向。我们认为，中等水平的密度和高度集中化可以防止中小型冲击出现太多而不能倒闭的问题，同时在大型冲击的情况下会严重造成问题。最后，我们的结果揭示了观察到的核心-边缘国家银行间网络的实际稳健性和脆弱性，突出了系统性危机的现有风险。