Recent regulatory initiatives, including proposed changes to the CAMELS ratings for banks, treat risk stripes as separate and independent from one another. In practice, however, a bank’s risks, both financial and nonfinancial, are often interdependent.
The CAMELS Proposal
A recent proposal to change the CAMELS rating system for banks itself reads like a game of Mad Libs that tries to jam as many regulatory cliches as possible into one document. The proposal uses the words “modernize”, “transparency” and “process-related” three times each. “Material” appears 63 times with the specific phrase “material financial risk” appearing 38 times. Interestingly, the proposal never defines “material financial risk” or even the words “material”, “financial”, or “risk.” As noted in an earlier post, the phrasing may sound impressive without being especially meaningful or informative.
More substantive, if similarly wrongheaded is Comptroller of the Currency Jonathan Gould’s statement on the CAMELS proposal. Gould sees each CAMELS factor as separate and distinct from one another. In his view, the Management rating should “serve as a standalone assessment rather than a secondary reflection of other components.” Gould further argues that each rating component should provide “distinct incremental value.” Finally, “no single component rating should disproportionately drive the composite rating.” While these assertions have some surface appeal, they are not supported by the evidence.
Management as a Standalone Assessment
It makes little sense to view the assessment of management as entirely separate from the other CAMELS components. If a bank has deficient capital, poor asset quality, excessive interest rate or liquidity risk, or weak earnings, how does that not reflect on management? If a bank’s management isn’t responsible, who is? The CAMELS proposal doesn’t even go that far since the ubiquitous “material financial risk” would supposedly also drive the Management rating. The material financial risk assessment is incredibly vague but attempts to define it would generally require a likely (not just elevated) threat the bank’s survival. A management downgrade at that point would represent more a CYA exercise than any attempt at preventive supervision.
There can be elements of a management assessment that are not reflected in the other CAMELS components. Noncompliance with consumer protection laws or anti-money laundering (AML) requirements comes to mind. Another example is operational risk, including potential losses from cyber threats and fraud. None of these risks even receive a mention in the CAMELS proposal.
Even if we believe examiners should base CAMELS assessments entirely on quantitative measures of risk, management can affect the quality and reliability of those risk measures. Data and system weaknesses can lead to less reliable risk measures. Assumption changes can improve the risk numbers without reducing the risk. For example, Silicon Valley Bank (SVB) exceeded its EVE limits in early 2022. The solution was not to hedge, sell long-term securities, or otherwise reduce its interest rate risk profile. Instead, the bank changed its model assumptions for nonmaturity deposits, lengthening their maximum average life from 5.5 years to 12 years.
Distinct Incremental Value
The focus on the “distinct incremental value” of each CAMELS component ignores the interdependence of these factors. The interaction between interest rate risk and liquidity risk presents a good case in point. The table below shows how four banks fared under two key risk indicators as of December 2022. “Uninsured/Total Deposits” shows the percentage of the bank’s total deposits were above the FDIC insurance limit of $250,000. This is a good indicator of liquidity risk. “Unrealized/CET1” shows unrealized losses on investment securities as a percentage of common equity tier 1 capital. This is a good indicator of interest rate risk (IRR). There are many indicators of IRR and liquidity risk, but these are two of the most prominent.
SVB showed extreme levels of both liquidity and interest rate risk. Nearly all its deposits were above the FDIC limit and unrealized losses exceeded capital. Signature Bank also had very high liquidity risk (89% uninsured) but a more moderate level of IRR. Conversely, Charles Schwab Bank and Bank of America had very high levels of unrealized losses, but also relatively low levels of uninsured deposits. Bank of America’s status as a systemically important (too big to fail) bank probably also discouraged depositor outflows.
Now let’s look at what happened to these banks. SVB failed at a cost to FDIC of nearly $19 billion. Signature also failed, though the cost to the FDIC was considerably less. Schwab and BAC survived. These banks were stuck with underwater assets but could wait things out since they weren’t facing a liquidity crunch. Combining high IRR with high liquidity risk turned out to be much worse than either high IRR or high liquidity risk alone.
A Risk Problem or a Capital Problem?
Exam reports often express problem assets or IRR relative to the bank’s capital. This makes sense. Banks with larger capital cushions can afford to take on more risk. However, it also makes sense to measure these risks relative to a capital-neutral denominator, such as total assets. It assists supervisors and the bank’s management in distinguishing between a risk problem or a capital problem. Consider again the case of SVB. The unrealized losses illustrated the bank’s exposure to rising interest rates. In addition, the bank failed to hedge its IRR or take other steps to contain its unrealized losses. However, the unrealized losses were also a capital problem.
Supervisors have historically placed little weight on unrealized losses when assessing capital. Supervisors consistently rated SVB’s capital a satisfactory “2”, even after unrealized losses started to exceed CET1. First Republic had similarly high unrealized losses, though on mortgages rather than securities, and likewise received a “2” rating for capital. Even by March 2023, when the bank’s condition became dire, the FDIC only saw fit to downgrade capital to a “3.”
Neither current nor proposed CAMELS rating criteria mention unrealized losses. They both mention “quality of capital,” and unrealized losses should be part of that assessment. Unfortunately, they usually are not.
Risks are Not All Created Equal
The notion that a bank’s CAMELS rating should simply reflect an arithmetic average of each of the individual components reflects a fundamental misunderstanding of how most banks fail. Multiple risks and weaknesses certainly make things worse and increases the opportunities for things to go wrong. But high across-the-board risks are rare. Most banks pick a poison when it comes to taking on credit risk vs. interest rate risk. Moreover, low or moderate risk in one area rarely offsets outsized risk in other areas. Pristine asset quality at SVB, Signature, and First Republic turned out to be about as relevant as Mrs. Lincoln’s opinion of Our American Cousin. The same goes for high credit risk. WaMu was rated “2” for Sensitivity as late as September 2008, but that good component rating didn’t ward off failure.
Leave a Reply