Market integration and diversification, processes that increase the interaction between banks, may drive the financial system towards instability, reports a study in Nature Communications. These processes were thought to increase financial stability, but the new research suggests that arising complex cyclical structures in the bank financial network have the potential to quickly amplify financial distress.
The stability of the global financial system is assessed by regulatory authorities through stress tests that typically consider each institution in isolation. However, the increasingly connected nature of the banking network may lead to distress propagation and amplification, collective effects that would not be predictable on an individual basis.
Using mathematical modelling and raw data from around 50 bank’s balance sheets, Marco Bardoscia and colleagues demonstrate that increasing the complexity of the financial system may not increase its inherent robustness. They show that the formation of specific features in the network formed by financial institutions and their mutual contracts can have a destabilizing effect. By enlarging the number of banks participating in the financial system and increasing the connections between them, the individual institutions become part of multiple cycles of contracts that may amplify the effects of financial shocks. The authors draw comparisons between these findings and observations in ecosystems, in which increased complexity (more interactions between species, once thought to improve stability) can also undermine stability.
Although tests to predict potential stresses in the banking system are becoming increasingly sophisticated, they usually consider financial systems in isolation. The authors suggest that their approach is more agile as it considers consequences of distress propagation across the network of contracts established among them.
Biotechnology: Engineering human cells to become transparentNature Communications
Machine learning: Model identifies three biomarkers associated with COVID-19 mortalityNature Machine Intelligence