Executive Summary
This research note provides an empirical analysis in response to the Federal Reserve’s notice of proposed rulemaking to tailor enhanced prudential standards for large banks. The proposal would tailor regulatory capital and liquidity requirements to the risk profiles of U.S. banking organizations with over $100 billion in assets.[1] Specifically, it would create four different categories based on nominal thresholds for a series of proposed risk-based indicators: size, cross-jurisdictional activity, reliance on short-term wholesale funding, nonbank assets, and off-balance-sheet exposure. The main objective of this research note is to assess the empirical importance of the proposed risk-based indicators in explaining SRISK, which we utilize as a proxy for a market-based measure of losses, and the suitability of the corresponding nominal thresholds. This assessment is based on linear and nonlinear regression models that relate SRISK to the proposed risk-based indicators. Before we summarize our main results, we should note that there are important limitations to our analysis. Specifically, any individual measure’s estimate of systemic risk, including the SRISK measure used in the analysis, is insufficient by itself and fundamentally flawed. First, SRISK does not capture the spillover effects on the real economy that may arise if a bank were to fail. Second, SRISK is based on a simple model used to estimate the expected equity loss conditional on a single shock to the equity market and also places undue reliance on bank size. Finally, for the avoidance of doubt, measures such as SRISK do not capture the effect of the regulations prescribed to mitigate systemic risk. Nevertheless, we use SRISK as a proxy for systemic risk here because it uses a consistent metric across banks and it is also readily available, making it straightforward to replicate the results of our analysis.
The results of our empirical analysis indicate that of the five risk-based indicators, only cross-jurisdictional activity and reliance on short-term wholesale funding have a strong positive correlation with a market based measure of losses, i.e. SRISK.[2] That is, after controlling for these two risk-based indicators, bank size, nonbank assets and off-balance-sheet exposures are not important drivers of a bank’s SRISK measure. In addition, we decompose cross-jurisdictional activity into its two main components: cross jurisdictional claims and cross-jurisdictional liabilities. While our results indicate that cross-jurisdictional claims are positively correlated with SRISK (our proxy for systemic risk}, we find that cross-jurisdictional liabilities are negatively correlated with SRISK across all specifications considered. The negative correlation between cross-jurisdictional liabilities and SRISK holds using both data from U.S> banks as well as data available for the international banks that are included in the BCBS GSIB assessment sample and have an SRISK measure. For this reason, our results indicate that the dollar-based threshold for cross-jurisdictional activity should be set at a significantly higher level than proposed, or at a minimum, if kept at the $75 billion level, cross-jurisdictional liabilities should be excluded.
There are several economic reasons to exclude cross-jurisdictional liabilities from the regulatory proxy of global footprint and complexity. First, it is desirable for banks to diversify their funding sources and banks have ample experience in hedging foreign exchange risk. Second, for foreign subsidiaries of U.S. firms, raising local funding is a sound asset-liability management and risk-management practice. Finally, cross-border borrowing can be less expensive than other forms of funding.[3]
With respect to the nominal thresholds, our nonlinear regression analysis sets the thresholds of cross-jurisdictional claims, nonbank assets, short-term wholesale funding and off-balance-sheet exposures at $70 billion, $80 billion, $75 billion and $50 billion, respectively. However, these estimated thresholds are not statistically different from the $75 billion thresholds advanced in the proposal at the 10 percent confidence level.
1 See Federal Reserve System, Notice of Proposed Rulemaking, Prudential Standards for Large Bank Holding Companies and Savings and Loan Holding Companies (31 October 2018).
2 Note that the proposal provides no empirical analysis demonstrating that cross-jurisdictional activity, reliance on short-term wholesale funding, nonbank assets, and off-balance-sheet exposure are the risk-based indicators that should be used to tailor prudential standards.
3 Ahnert, Toni, Kristin Forbes, Christian Friedrich, and Dennis Reinhardt, “Macroprudential FX Regulations: Shifting the Snowbanks of FX Vulnerability?” NBER Working Paper No. 25083, September 2018.