Size is Overrated

Any firm whose failure would pose a systemic risk must receive especially close supervisory oversight of its risk-taking, risk management, and financial condition, and be held to high capital and liquidity standards.” Ben S. Bernanke, Speech at the Council on Foreign Relations, March 10, 2009

“The history of bank regulation in the United States is of progressive dilutions of core regulatory requirements over a number of years, leaving the banking system as a whole vulnerable to crisis.”
Paul Tucker, Chair of The Systemic Risk Council, Letter to Senate Banking Committee leaders, February 21, 2018.

This month, in the guise of supporting community banks, the U.S. Senate passed a bill (S.2155) that eases regulation of large banks. Looking at the details of the legislation, analysts have identified a broad range of shortcomings. These include: (1) compelling regulators to relax leverage ratio requirements on key custody banks; (2) easing capital requirements on the largest banks by modifying the supplementary leverage ratio; (3) allowing reduced frequency of stress tests; (4) obliging the Federal Reserve to tailor prudential standards—rather than simply giving them the option—thereby inviting litigation aimed at weakening oversight of the largest firms; and (5) compelling regulators to treat certain municipal securities as eligible in meeting liquidity requirements (Title IV Sec. 403 of the bill).

We share the critics’ views that this wide-ranging dilution of existing regulation will reduce the resilience of the U.S. financial system.

In its best known and most publicized feature, the Senate bill raises the asset size threshold that Dodd-Frank established for subjecting a bank to strict scrutiny (such as the imposition of stress tests, liquidity requirements, and resolution plans) from $50 billion to $250 billion. At the same time, S.2155 gives the Fed the option to maintain strict oversight of banks larger than $100 billion provided the Board “takes into consideration … [their] capital structure, riskiness, complexity, financial activities (including financial activities of subsidiaries), size, and any other risk-related factors.”

In the remainder of this post, we examine the role of asset size in determining the systemic importance of a financial intermediary. Is size a sufficient proxy for the damage a failing institution can inflict on the financial system? Our conclusion is that (aside from the very largest institutions, where it does in fact dominate) balance sheet size is not a terribly useful indicator of the vulnerability a bank creates. As we will explain, some institutions with assets well below the $250 billion threshold could easily pose significant systemic risks were they to become dysfunctional. The reverse is also true. That is, some larger institutions are unlikely to lead to broader disruptions even if they were to fail.

Our conclusion is that Congress should ease the strict oversight burden on institutions that pose little threat to the financial system without raising the Dodd-Frank threshold dramatically. Judge makes an elegant proposal for accomplishing this. For institutions with assets between $100 billion and $250 billion, Congress should just flip the default. Rather than obliging the Fed to prove a mid-sized bank’s riskiness, give the bank the opportunity to prove it is safe. This approach has two extremely appealing properties. First, it gives institutions the incentive to limit the systemic risk they create in ways that they can verify. And second, it sharply reduces the risk of litigation by banks that the Fed deems risky.

Turning to our main arguments, recall that some moderately sized institutions catalyzed the 2007-2009 crisis because they became dysfunctional precisely when there was a broad-based shortage of capital. IndyMac, Countrywide, National City, and GMAC―all of which collapsed in 2008 with assets well below the Senate’s $250-billion threshold―were like lit matches in a forest of dry tinder (see, for example, Wilmarth).

Second, consistent with the goals of many legislators, federal regulators already tailor their regulations and supervision to circumstances. At the small end of the spectrum, major rulings to implement Dodd-Frank typically exempt banks with less than $10 billion in assets—these are the community banks that account for 98 percent of the 5,670 FDIC-insured institutions (see here, Table 2). In fact, stricter Dodd-Frank-mandated supervision currently applies to only 38 bank holding companies. And, the Fed further tailors supervision, imposing tighter standards on a mere 15 large banks, with the most demanding regime reserved for the 8 U.S. banks designated as global systemically important banks (G-SIBs) using the methodology developed by the Basel Committee on Banking Supervision (BCBS). (See page 24 of the recent Congressional Research Service report for details.)

This prompts us to ask how it is that an institution’s size per se affects the threat to the financial system. For some measures of systemic risk, the answer is straightforward. In the case of SRISK—the systemic risk measure of the NYU Stern Volatility Lab that we cite frequently—an increase in debt-financed assets adds proportionally to an institution’s capital shortfall (see Brownlees and Engle). However, features other than a firm’s capital shortfall—such as the lack of substitutability for its activities—also influence the threat to the financial system.

This leads us to consider the array of risk indicators that the BCBS uses to assess systemic vulnerability in designating G-SIBs (see Table 1 here). Like most measures of systemic risk, the BCBS starts with size. However, rather than simply using assets, they measure exposures including off-balance sheet items related to derivatives positions, credit commitments, and the like. Beyond the size of exposures, the BCBS considers four categories with 11 individual indicators:

  • Interconnectedness (intra-financial sector assets, intra-financial sector liabilities, and securities outstanding)
  • Substitutability (payments activity, assets under custody, and underwriting activity)
  • Complexity (over-the-counter derivatives, securities held for trading or available for sale, and certain highly illiquid assets)
  • Cross-jurisdictional activity (foreign claims and total cross-jurisdictional liabilities)

Behind this multi-dimensional array of indicators lies two simple ideas. First, the larger a bank’s exposures, and the more interconnected, complex and internationally active it is, the more likely that its failure will spill over to others in a crisis. Second, the less substitutable the activities of a bank, the more likely its impairment will disrupt critical financial functions such as payments and credit supply. To determine the list of G-SIBs, the BCBS calculates a systemic importance score as a weighted sum of the five categories (the four above plus size), subject to a cap on the contribution of substitutability.

Using the latest estimates for 40 bank holding companies (BHCs) from the Office of Financial Research, we plot this systemic importance score against total assets—the latter being the measure of size written into S.2155. Looking at the chart, the blue line represents the Senate’s $250-billion threshold, while the red line is the BCBS’s threshold of 130 used for G-SIB designation. We highlight the eight U.S. G-SIBs with red dots and with labels. While one U.S. G-SIB is just below the Senate threshold (State Street), all eight clearly exceed the BCBS systemic importance threshold for designating a G-SIB.

Systemic risk score vs. asset size (billions of U.S. dollars), end-2016

 Source: Based on bank holding company data in Treasury  Office of Financial Research  update of  Size Alone is Not Sufficient To Identify Systemically Important Banks .

Source: Based on bank holding company data in Treasury Office of Financial Research update of Size Alone is Not Sufficient To Identify Systemically Important Banks.

The chart suggests that size is an excellent indicator of systemic risk. In fact, a simple regression of the score on size—plotted as the black line in the chart—accounts for 87 percent of the variation in the full data set. But, things look very different if we remove the six behemoths with assets over $800 billion.

To see how different, we redraw the chart for the 34 BHCs with assets under $500 billion. Now asset size accounts for only 22 percent of the variation of systemic risk scores. Furthermore, some relatively large firms that easily exceed the Senate’s $250-billion threshold―Capital One, PNC, TD Group U.S., and U.S. Bancorp (in blue)―exhibit relatively low systemic risk scores. Conversely, several BHCs that fall into the $100 billion to $250 billion range―Northern Trust, Deutsche Bank Trust, Barclays U.S., and Credit Suisse U.S. (in red)―exhibit substantial risk scores, albeit below the G-SIB threshold.

Systemic risk score vs. asset size (billions of U.S. dollars), end-2016

Clearly, size is far from definitive. But, there is more. Perhaps unsurprisingly, banks can pose material risks to the entire system even if their systemic risk score falls well short of the Basel Committee’s G-SIB threshold. In some recent work, Passmore and von Hafften (here and here) seek to quantify the level of capital surcharge needed to eliminate public bailouts of risky banks during a severe financial crisis. Their analysis implies that—to avoid public bailouts—the initial G-SIB threshold should be set at a systemic importance score between 16 and 52 rather than the current 130. In the U.S. sample of 40 banks, this would lead to designating between 12 and 22 banks as systemic, not 8. (See the work of the Office of Financial Research for a discussion of ways to improve the assessment of systemic importance.)

On top of this, raising the Dodd-Frank threshold to $250 billion could lead to a gradual increase in risk-taking by banks below that threshold. A 2015 study for the Financial Services Roundtable shows that regional banks with assets below the existing Dodd-Frank threshold of $50 billion frequently have systemic risk scores that are as high as banks with assets in the range of $50 billion to $250 billion (see Figure 3 here). If this pattern were to repeat following the weakening of scrutiny prescribed by the Senate bill, financial vulnerability would rise significantly over time.

Our conclusion is that using size alone―and raising the threshold sharply―would almost surely reduce systemic resilience. Not only that, but fragility should be expected to rise if the Fed were to bear the burden of proving that banks with fewer than $250 billion in assets merit strict scrutiny. The likely result of this change will be court proceedings that legal scholars worry will result in a reliance on a cost-benefit analysis (see, for example, Gordon). We have already seen how legal challenges can diminish the effectiveness of oversight in the case of nonbank systemically important financial intermediaries (SIFIs). In the aftermath of MetLife’s successful SIFI de-designation suit, the Financial Stability Oversight Council (FSOC) appears unwilling or unable to designate nonbanks as systemic. (And the lone remaining nonbank SIFI is lobbying to reduce oversight despite evidence that it has not de-risked.)

So, the Senate has it only partly right. Size is overrated as a metric for determining whether large banks deserve strict prudential oversight—at least for those smaller than the six largest U.S. institutions. However, that is no reason to raise the Dodd-Frank threshold dramatically or to place greater legal burdens on regulators struggling to safeguard the resilience of the U.S. financial system and to prevent future public bailouts. Rather, Congress should allow the Fed—at its option—to exclude from strict oversight those institutions with assets in excess of $100 billion that demonstrate their safety using an agreed-upon array of risk indicators.

We expect that the Fed would welcome this approach, as it would allow regulators and supervisors to focus their scarce resources where they are most needed. Unfortunately, the best we may get from this Congress is stalemate, slowing the regulatory dilution that historically precedes financial crises. That dilution would accelerate if, instead, the House of Representatives inserts major components of its Financial CHOICE Act into a joint House-Senate “regulatory relief” bill that becomes law (see our earlier post).