As noted in an earlier post, efforts to reduce the role of supervisory guidance included limiting the use of numerical bright lines. While regulators need to develop and apply these bright lines with some care, their absence can undermine effective supervision.
The “guidance on guidance” specifically cautions against the use of quantitative thresholds or bright lines when promulgating supervisory guidance. No further rationale is provided, however. This may reflect the idea that clear, quantitative guidance is easier for examiners to interpret as a rule than as a more general point of reference. However, regulations also frequently avoid numerical thresholds or clear definitions in favor of a more “principles-based” approach.
There are several characteristics of a well-constructed bright line. It should minimize both false positives and false negatives, avoid undue complexity, and leave room for examiner discretion. Some dismiss bright lines as a one size fits all approach, but having common understanding of risk promotes consistency across banks and across supervisors. I’ll explore the use of bright lines in three areas: interest rate risk, credit risk, and liquidity risk.
Interest Rate Risk
Section 305 of the FDIC Improvement Act of 1991 required each Federal banking agency to revise risk-based capital standards to account for interest rate risk. The commercial banking regulators essentially blew off the requirement. The agencies indicated that techniques for measuring IRR were still evolving and did not wish to impede progress but instead encourage industry efforts to improve risk management techniques. (Spoiler alert: it didn’t work.) The Office of Thrift Supervision, which regulated savings institutions, took a shot at it.
The OTS’s IRR Capital Component used the agency’s supervisory IRR model as a starting point with the sensitivity of net portfolio value (essentially economic value) as the relevant risk measure. If your IRR was above normal (in this case changing by more than 2% of assets following a 200bps rate shock), you’d have an additional risk-based capital requirement. The rule would be implemented on a six-month lag. If your IRR fell in the interim, you could use the lower number. If it increased, well, good for you. You could appeal based on internal model results.
OTS repeatedly delayed implementation of the rule and eventually dropped it altogether. This decision reflected concerns that the new rule would disadvantage thrifts relative to banks. There was, however, a more fundamental problem with the rule. When it comes to risk management, financial institutions usually pick their poison between credit risk and IRR. Banks and thrifts with high IRR tend to load up on MBSs and other low risk weighted assets. As a result, their risk-based capital ratios were very high. I’m aware of cases where a 200 or 300bps rate shock would completely wipe out NPV but the thrift remained well capitalized, even after applying the IRR component. These sorts of nonsensical results tended to undermine rather than promote effective supervision of IRR.
OTS tried a different approach with Thrift Bulletin 13a. The idea here was to provide guidance on exam ratings for IRR, based on the level of IRR (Sensitivity Measure) and the ability to absorb risk (Post-Shock NPV Ratio). Examiners also considered management of IRR. TB-13a provided a handy matrix that showed presumptive ratings based on risk profiles, shown below.
These guidelines were only a starting point. Examiners were to consider the bank’s internal results as well as a host of other factors before settling on the rating. I found examiners more willing to adversely rate IRR once TB 13a thresholds were established. TB 13a and the supervisory model went away with the dissolution of OTS. Quantitative thresholds would be difficult to implement today in the absence of a supervisory model or at least a common set of assumptions.
Credit Risk
Residential mortgages are typically assigned a 50% weight under risk-based capital (RBC) regulations. Qualification for the 50% risk weight has become less strict and more subjective over time. When RBC regulations first went into effect in 1989, OTS required an original loan-to-value ratio of 80% or lower to qualify for the 50% risk weight. Years of experience show that more borrower equity usefully translates to fewer defaults and lower losses given default.
The regulation also stipulated that the mortgages be “prudently underwritten.” This provision didn’t have much teeth, as subprime, low doc, and negatively amortizing loans started to proliferate. Even after the Global Financial Crisis, financial regulators were reluctant to be nailed down. The CFPB defined Qualified Mortgage (QM) loans for Truth in Lending purposes based on a set of ability to pay criteria. Lenders needed to verify borrower income or assets, avoid excessive rates or fees, and avoid higher risk features, such as negative amortization or balloons. Banking regulators clarified that “Residential mortgage loans will not be subject to safety-and-soundness criticism based solely on their status as QMs or non-QMs.” Risk based capital gets no mention at all.
OTS loosened the LTV threshold to 90% in 2002. Banking agencies didn’t establish a threshold in the regulation itself, but Interagency Lending Guidelines generally required an LTV of no more than 90%. Current capital regulations merely specify making loans in accordance with prudent underwriting standards” including LTV standards. The current approach has a distinct “know when you see it” quality. If history is any guide, these prudent underwriting standards will be blindingly obvious – after something goes wrong.
Liquidity Risk
Despite recent bank failures, rules around liquidity risk are fundamentally sound and require relatively modest tweaks. The liquidity coverage ratio is clear, conceptually straightforward, and intuitive. Net outflows during times of stress should not exceed cash on hand and other liquid assets. An adequate LCR is a necessary, but not sufficient condition when assessing liquidity risk. Supervisors can address institution-specific exposures through internal liquidity stress testing requirements. The ILST regulation doesn’t get very specific, but it can provide a credible backstop to the bright lines of LCR.
The current LCR framework has two key weaknesses. First, the definition of high-quality liquid assets fails to take IRR into account. Revisiting the HQLA definition would address that issue. Second, large but not gigantic companies use a “modified” LCR that assumes only 70% of stressed outflow in the denominator. The rationale was that these smaller firms “would likely not have as great a systemic impact as larger, more complex companies if they experienced liquidity stress.” Well, that assumption is out the window.
It’s worth noting that SVB fell below its 100% LCR threshold in December 2022, but remained in compliance with modified LCR. SVB’s ILST buffer was similarly identified as insufficient. The failures of SVB and others were less a matter of examiners failing to understand and appreciate liquidity risk. Rather, they were too slow to address the deficiencies they had already identified.
The use of bright lines is not a strictly supervisors vs. regulators issue. Many examiners don’t like to be boxed in by quantitative thresholds and prefer to retain as much discretion as possible. But history shows that the absence of bright lines emboldens more aggressive and pugnacious banks and provides plausible deniability to their enablers.