Systemic risk represents the potential impact that the failure of an individual firm can lead to a broader contagion. Such a contagion can endanger the overall financial sector and even the economy as a whole. Efforts to control such contagions, especially within the financial sector, can be expensive. Actions taken during the 2008 Global Financial Crisis as well as during the more recent failures of SVB, Signature, and First Republic are two prominent examples. While most agree that systemic risk is important, measurement of that risk has proved more elusive.
Systemic risk consists of two separate but related components. The first relates to the likelihood that an individual firm might fail and the potential cost of that failure. That risk is a function of the firm’s leverage, liquidity, and composition of its balance sheet. The second relates to the risk that the firm’s failure can impact other firms or the broader economy. The second risk can reflect a high degree of interconnectedness to other firms, or the failure could expose a vulnerability common to other firms.
The FDIC must resolve bank failures at the least cost to the insurance fund. However, a systemic risk exception allows resolutions that are more costly, at least in the short run. The SVB resolution provides a case in point. The FDIC estimates a resolution cost of $17.8 billion (on total assets of $209 billion). However, insured deposits were only about $24 billion, with collateralized borrowings (FHLB advances) another $15 billion. The bank’s cash balance and fair value of its investment portfolio was nearly $115 billion, more than enough to cover the cost, even at fire sale prices.
The resolution of Signature Bank also fell under the systemic risk exception. The failure of First Republic did not – officially. However, regulators made extraordinary efforts to try to save First Republic. It’s hard to imagine the Treasury Secretary personally intervening on behalf of a small community bank. Its resolution cost ($16.5 billion) was nearly as high as SVB’s. The direct costs don’t account for the potential impact on behavior. Spending billions to go above and beyond clearly defined insurance limits amounts to moral hazard on steroids.
Tailoring and Systemic Risk
The resolutions in 2023 run counter to earlier legislative and regulatory changes that lightened regulatory requirements, especially for banks with less than $250 billion in assets. The rationale for the change was that while some of these banks were quite large, they posed little systemic risk. When laying out the new tailoring framework, the Federal Reserve was careful to describe these firms as having less systemic risk than the megabanks. Some members of Congress were less careful. Seven senators wrote to Fed Vice Chair Randy Quarles urging him to further extend relief to firms in the $250-750 billion range since “where your data indicates that they do not pose systemic risks.” In financial modeling parlance, the tailoring model had three big back-test exceptions in 2023.
While regulators retain some flexibility in designating firms as systemically risky, they have shown little appetite to do so. The Financial Stability Oversight Committee (FSOC) is responsible for designating Globally Systemically Important Banks (GSIBs) and other financial firms. Staffing levels at FSOC’s research arm, the Office of Financial Research (OFR) fell from 214 to 107 between 2016 and 2020, right at the time regulators were given more discretion in systemic risk determination. Treasury Secretary Steve Mnuchin justified the staffing cuts as “saving taxpayer dollars.” This is a good illustration of the shortsightedness with respect to regulation that I discussed in an earlier post.
AIG, General Electric, and Metropolitan Life were designated as “systemically important financial institutions” (SIFIs) following the Global Financial Crisis. The FSOC later rescinded that designation. AIG’s case is especially striking. The rescission came less than a decade after a $180 billion bailout. The Treasury Secretary at the time, Hank Paulson, justified the bailout on the grounds that an AIG collapse “would’ve buckled our financial system and wrought economic havoc on the lives of millions of our citizens.” The last time I checked, AIG still sells insurance. The impact of the AIG bailout on Paulson’s former employer may have played a role as well.
Current Regulatory Measures of Systemic Risk
The Federal Reserve still collects and OFR analyzes information related to systemic risk, based on the FR Y-15 Systemic Risk report. The reporting is more detailed for those banks already designated as GSIBs. A bank’s GSIB score determines how much additional capital a current GSIB has and can also be used to designate new GSIBs. Although regulators collect FR Y-15 data quarterly, a bank’s G-SIB score is only calculated as of December 31st, and only becomes effective twelve months later.
FR Y-15 reports showed SVB with a fairly high systemic risk profile. SVB ranked 15th as of the end of 2022 and tenth among US-headquartered banks. That’s higher than US Bank, PNC, and Truist. Signature and First Republic didn’t make the list at all. They were not bank holding companies and were not required to file an FR Y-15 Systemic Risk report. Banks shaded in gray are not subject to G-SIB charges. Nevertheless, exempt Charles Schwab had a higher GSIB score than BNY-Mellon, State Street, and even retail giant Wells Fargo.
Academia Weighs In
Are there better ways to measure systemic risk? Brownlees and Engle developed SRISK, which measures systemic risk of firms based on their size, leverage, market value of equity (based on stock price rather than a mark to market of the balance sheet), and stock beta. The idea is to estimate the potential capital shortfall in the event of a financial crisis. Engle prepared a YouTube video that provides a handy explanation. The advantage of SRISK is that it can be updated frequently and takes into consideration the market’s perception of the firm’s condition and risk.
That may also be a downside, since publicly available information on items like asset quality and interest rate risk is spotty at best. The table below shows the US financials with the ten highest SRISK levels. The list includes some large banks with relatively low G-SIB scores (US Bancorp and Truist) and three insurance companies.
Rajgopal and Peddireddy developed an alternative systemic risk measure called CRISK, which focuses more on financial statement measures. For example, CRISK looks at book equity, and backs out market-based measures such as other comprehensive income. However, it also focuses more on funding risk, including the level of uninsured deposits. One strong selling point of CRISK is that it would have identified SVB as one of the five highest risk U.S. banks in both 2022 and 2021. However, the authors do not provide a detailed breakdown of the CRISK drivers for SVB, so its appearance on the list remains a bit of a black box. Data availability limitations also mean that the CRISK calculation must rely on proxies. Curiously, while Citigroup has the highest SRISK rating, it wouldn’t have made the CRISK top five since 2010. Citi’s low price-to-book ratio (<50% of book) may be driving its SRISK rating. CRISK does not take such market perceptions into account.
Proposed Changes to Systemic Risk Reporting
On September 1, 2023, the Federal Reserve issued a Notice of Proposed Rulemaking (NPR) to make changes to FR Y-15 reporting and the GSIB calculation. The proposal addresses two gaps in the GSIB designation process. Using a single date makes it much easier for a bank to game its GSIB score. Both anecdotal and empirical evidence suggest that such gaming occurs. The NPR proposes incorporating more quarter-end and daily averages into the calculation. Another problem is the long lag between the GSIB “as of” date and the timing of the surcharge. The SVB debacle showed that regulators lacked an appropriate sense of urgency in addressing problems. A twelve-month lag between the as of date of the calculation and its implementation as a GSIB surcharge is ridiculously long. The NPR also proposes moving up the timing of the GSIB surcharge, suggesting either April 1 or October 1 as alternative dates.
These changes make sense in principle, but I see a couple of potential problems. The use of daily averages will make it harder to game the GSIB score but will also increase the banks’ reporting burden and likely make the data more prone to error and harder to verify. Increasing the complexity of the report calculation may also lead to less timely reporting.
The NPR calls for only modest revisions to FR Y-15 line items. SVB, Signature, and First Republic all relied largely on uninsured deposits. Current systemic risk reporting does not adequately capture this exposure. At best, current GSIBs report short-term wholesale funding. That amount can include non-operational deposits, but that’s not the same as uninsured. Moreover, the designation of operational deposits is highly subjective. Current GSIB rules also exclude banks that are not holding companies but two of the large failures this year fit that category. Extending the reporting requirements to non-BHCs would take systemic risk reporting and GSIB designations outside of the Fed’s domain. But isn’t that what FSOC and OFR were created for?
Douglas Holtz-Eakin of the center-right American Action Forum claimed in 2018 that “there has never been a convincing definition of systemic risk or any operational way to measure it…It should be jettisoned and FSOC disbanded, and the focus returned to efficient and effective prudential supervision.” To his credit, Holtz-Eakin also opposed coverage for SVB’s insured depositors in 2023. Could the FDIC have saved $17.8 billion by sticking to the $250K insurance limit? Maybe. Such an action could have also sent uninsured depositors at other banks heading for the exits and triggered a broader financial crisis. We may never know. Regulators made the systemic risk determination over the space of two days in what seems like a seat of the pants fashion. Better tools for monitoring and measuring systemic risk, and better use of existing tools, could have led to a more thoughtful and rigorous decision process.