U.S. banking regulators have proposed changes to the Enhanced Supplementary Leverage Ratio (eSLR), which is applied to the nation’s largest financial institutions. Supporters of the proposal argue that the eSLR should act as a backstop rather than a binding constraint. But a backstop that never acts as a binding constraint doesn’t make for much of a backstop. More fundamentally, how leveraged are big banks?
Overview
The current proposal would revise the eSLR calculation applying to the nation’s largest banks and bank holding companies (BHCs). The Fed estimates that the proposal would lead to an average reduction of the eSLR of 23% at the BHC level and by 36% for bank subsidiaries. Regulators are also considering excluding Treasury securities held for trading and reserves held at the Federal Reserve from the denominator. Adding these exclusions would translate into eSLR reductions at the BHC and bank levels of 34% and 46%, respectively. The change in overall capital requirements would be less, since risk-based capital (RBC) would become the binding requirement for more banks. Beyond these specifics, the proposal raises a broader question: are capital requirements for megabanks too strict? Or not strict enough?
The country’s largest banks, or G-SIBs, are subject to a dizzying array of capital requirements and constraints. These include risk-based capital under standardized and advanced approaches, leverage, supplemental leverage, and stress buffers. Risk weighted requirements include capital charges for market and (sometimes) operational risk. Before we bring out the violins, though, it’s important to recognize that these constraints aren’t that constraining.
Leverage at Big Banks
Defining capital relative to “risk-weighted assets” (RWAs) was an effective piece of branding by banks and their regulators. For the vast majority of banks, RWAs are considerably less than (unweighted) total assets, as reported under Generally Accepted Accounting Principles (GAAP). That’s even after including off-balance sheet exposures, derivatives, and market risk. Representing capital in terms of RWA makes the ratio look a lot higher. JP Morgan’s CET1 ratio looks great at 15.41%. But its actual leverage ratio (Equity/Assets) is only 7.16%.
That’s not just the case with JPMC. G-SIBs have among the lowest leverage ratios in the banking industry. The chart below compares leverage ratios for the G-SIBs to all bank holding companies (BHCs) with assets over $10 billion. Each G-SIB was among the lowest 10 percent in this peer group.
It’s All About the Denominator
The difference between risk based and leverage ratios boils down to how regulators define assets. Not only do risk-weighted assets tend to be much less than GAAP assets, but the GAAP assets figure also ignores much of the leverage that arises from derivatives and other off-balance sheet exposures. That’s where the eSLR comes in. It introduces a new denominator, total leverage exposure. The table below contrasts the three measures.
BNY and State Street benefit from a custodial bank carve-out in the eSLR rules. For the other BHCs, however, TLE is between 15% and 22% higher than total assets under GAAP. RWA ranges from 31% to 55% of total assets. Wells Fargo, which has primarily a retail focus, came in highest at 55%. Both Goldman Sachs and Morgan Stanley are below 40%. Current risk-based capital rules are friendlier to Wall Street than to Main Street.
What’s driving TLE and ultimately, eSLR? TLE adjusts for derivatives, repo-style transactions, and assorted other off-balance sheet exposures. Only the fair value of a derivative (roughly corresponding to its mark-to-market gain) counts as an asset under GAAP. This amounts to much less than the derivative’s potential exposure, let alone its notional value. These activities allow banks to increase their leverage without the activity showing up in the traditional leverage ratio. How much? The nation’s largest banks have tens of trillions in derivatives. There have been enough derivative-related blowups to show that this activity is hardly risk-free. The eSLR add-on is quite modest, representing a tiny fraction of their notional values, as shown below:
These ratios are just for the denominator. Think in terms of marginal capital requirements. Under the supposedly draconian 5% eSLR requirement, $1 billion in additional capital at BAC or Citi would support $3.8 trillion in additional derivatives. Also bear in mind that these levels of derivatives activity apply to only a handful of megabanks. Contrast them with the country’s seventh largest bank, U.S. Bank, which has less than $1.3 trillion (notional) in derivatives. USB is hardly a mom and pop. Its total assets are nearly one-third higher than those of Microsoft.
Some banks and regulators dismiss the importance of notional values. It’s true that you’d see much bigger market value swings in a ten-year swap than in a one-year swap with the same notional value. But that doesn’t make notional values irrelevant. Derivatives are often designed to replicate the return and risk of an on-balance sheet transaction and can allow for a tremendous amount of leverage. Collateral, netting agreements, central clearing and other actions to mitigate the risk of derivatives may work well under benign conditions but can break down during a financial crisis.
The argument that the eSLR should be a backstop and not a binding constraint is a curious one. Meaningful backstops should become binding constraints, at least on occasion. Otherwise, what is their point? Proponents of the eSLR changes argue that risk-based capital rather than leverage should act as the primary capital constraint. The trouble is that risk-based capital rules leave some serious gaps. What’s worse, proponents of changes to eSLR have also resisted efforts to close risk-based capital gaps.
What are these gaps? The RBC framework does not consider interest rate risk at all. It’s not as though IRR is not an issue at large banks. At Bank of America, unrealized losses on held-to-maturity securities still represent nearly half of Tier 1 capital. Current rules are supposed to account for market risk, but the market risk rule relies on a value-at-risk approach that has proved inadequate to deal with severe market swings. Operational risk charges only apply under the Advanced Approach (AA), meaning they only apply to one BHC (Citi).
These gaps in the RBC framework look to get worse, not better. The Basel Endgame proposal included a new framework for measuring market risk that applied some of the lessons from the global financial crisis. It would have also changed some risk weighting under the AA to make meaningful capital charges for operational risk more likely. The Basel Endgame looks dead in the water. At the same time, regulators also plan to reduce the stress capital buffer by watering down supervisory stress tests.
The situation can be likened to speed limits and air bags. As with risk-based capital, speed limits are designed to curb riskier behavior. As with the eSLR, air bags serve as a backstop in case speed limits alone aren’t enough to prevent accidents. Opponents of air bags and of the eSLR both argue that these backstops might encourage riskier behavior. For eSLR, by making leverage rather than RBC the binding constraint. For airbags, by making high speed accidents more survivable. But suppose the government removed airbag requirements and raised the speed limit to 100 m.p.h. Doesn’t sound very safe, does it?