Working Paper: CEPR ID: DP16458
Authors: Andrew Ellul; Dasol Kim
Abstract: We provide evidence on how banks form network connections and endogenous risk-taking in their non-bank counterparty choices in the OTC derivative markets. We use confidential regulatory data from the Capital Assessment and Stress Testing reports that provide counterparty-level data across a wide range of OTC markets for the most systemically important U.S. banks. We show that banks are more likely to either establish or maintain a relationship, and increase their exposures within an existing relationship, with non-bank counterparties that are already heavily connected and exposed to other banks. Banks in such densely-connected networks are more likely to connect with riskier counterparties for their most material exposures. The effects are strongest in the case of (non-bank) financial counterparties. These findings suggest moral hazard behavior in counterparty choices. Finally, we demonstrate that these exposures are strongly linked to systemic risk. Overall, the results suggest a network formation process that amplifies risk propagation through non-bank linkages in opaque financial markets.
Keywords: counterparty risk; financial networks; bank interconnectedness; over-the-counter markets; derivatives
JEL Codes: G21; G22; D82
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| interconnectedness of counterparties (F65) | riskiness of bank choices (G21) |
| banks' counterparty choices (G21) | systemic risk (E44) |
| network formation process (D85) | risk propagation through nonbank linkages (F65) |
| interconnectedness (F60) | counterparty leverage (G32) |
| riskiness of counterparty (G33) | exposures to riskier counterparties (G32) |