Researchers at the US Office of Financial Research investigated whether banks’ counterparty choices in OTC derivative markets contribute to network fragility. They use novel confidential regulatory data and show that banks are more likely to choose densely connected non-bank counterparties and do not hedge such exposures.
Banks are also more likely to connect with riskier counterparties for their most material exposures, suggesting the existence of moral hazard behavior in network formation.
Finally, researchers show that these exposures are correlated with systemic risk measures despite greater regulatory oversight after the crisis. Overall, the results provide evidence of risk propagation in bank networks through non-bank linkages in opaque markets.
The findings have a number of potentially important policy implications. First, bank regulators primarily focus on direct counterparty exposures to calculate regulatory capital charges, overlooking broader network information related to how counterparties may be connected to other banks. As the analysis suggests, this may provide opportunities for leakage, as banks may be able to increase indirect interconnections without the same degree of regulatory scrutiny. In the paper, researchers demonstrated that it is possible to monitor these types of interconnections using existing regulatory data.
Second, the results are consistent with predictions of theoretical models that bank risk-taking behavior can exacerbate fragility in dense network structures. One potential criticism of these models is that banks were quite resilient during the significant shocks to the financial system in March 2020. It also casts doubt on the validity of skepticism of post-crisis reforms designed to mitigate systemic risks.
At the same time, the very meaningful regulatory interventions that were implemented throughout 2020 suggest that there are deficiencies in the post-crisis regulatory framework that should be addressed, and the network connections documented in this paper can be an area requiring regulatory attention.
Finally, our analysis focuses on the uncleared derivatives markets. The mechanisms studied are not mutually exclusive to these markets and could exist in the cleared markets as well. One implication is that risks to CCPs may be understated based on most conventional metrics. However, what helps allay concerns to some extent is that certain types of derivative instruments used by counterparties to increase risk through their derivative positions cannot be readily cleared, including total return swaps.