The Stress Testing Road Ahead: How to Prevent Future Bank Failures

“Happy families are all alike; every unhappy family is unhappy in its own way,” reads the opening sentence of Leo Tolstoy’s Anna Karenina.

This is a fitting analogy for stress scenarios. Normal market conditions are generally favorable for everyone, but trouble and unhappiness arise in financial markets in unique and varied ways.

Alla Gil

Time and again, over the past 15 years, the largest U.S. banks have proven their ability to withstand severe (hypothetical) shocks, passing the Federal Reserve’s annual stress test with flying colors. While this year was no different, the problem is that the Fed’s so-called CCAR exercise is limited in scope, and has always been more reactive than proactive.

As we saw in 2023, regional banks, just like their larger counterparts, are vulnerable to failure – but they are not currently subject to CCAR. Moreover, whether we’re talking about, say, interest-rate risk, inflation, liquidity risk or commercial real estate hazards, the Fed’s stress scenarios do not typically consider the factors fueling bank fiascoes before they actually happen.

While the 2024 CCAR results once again demonstrated the stability of the largest banks, there are hundreds of smaller banks with concentrated, undiversifiable exposures. Consequently, as Federal Reserve Vice Chair of Supervision Michael Barr recently pointed out, there is a need for dynamic stress testing.

What scenarios could push smaller banks into default in 2025? How many of them are at risk, and what specific business practices make them most vulnerable?

The Need for Expansion and Customization

Expanding standard stress tests to regional and community banks could place additional pressure on their already scarce resources, without necessarily revealing their specific pockets of risk. To prevent further bank failures, next year’s exploratory Fed scenarios must be tailored to peer groups of banks based on their size, geography and business profile.

Banks that want to move toward dynamic stress tests must first answer the following questions: (1) What unique risks does your business model present, and how can you mitigate them? (2) How would prolonged high-interest rates and an inverted yield curve impact your financial health? and (3) What strategies can you employ to manage liquidity pressures effectively?

Supervisors, on the other hand, should ask: (1) How can we tailor stress tests to better reflect the unique risks of smaller, regional banks? (2) What early warning signs should we be monitoring to prevent future defaults? and (3) How can we ensure that stress tests evolve in response to changing market conditions and emerging risks?

The answers to these questions will shape the next set of CCAR exploratory scenarios and should help safeguard the stability of banks of all sizes.

Dynamic Stress Testing: A Case Study

To illustrate how dynamic stress testing can work in practice, consider a group of regional banks with assets under management between $50 billion and $200 billion. Using call report data from the FDIC database, we linked their balance sheet segments to macroeconomic scenarios, using machine-learning-based regression with regularization methods. The estimated equations were then applied to thousands of auto-generated future scenarios.

For each scenario at each time point, we calculated two key ratios representing credit and liquidity risk. This allowed us to construct empirical distributions of these ratios for any selected time step, yielding the probability of identified scenarios and enabling peer group benchmarking for financial institutions and their regulators.

Figures 1 and 2 demonstrate benchmark analysis examples conducted for selected sample banks. Credit and liquidity ratios were calculated on 1,000 forward-looking scenarios over a three-year horizon, on a quarterly basis. Liquidity ratios were calculated as demand deposits and short-term borrowings over total deposits and borrowings, while credit ratios were calculated as delinquent loans and leases and net charge-offs over total loans and leases.

Source: Straterix

Source: Straterix

All banks’ ratios were calculated on the same set of scenarios. To compare their risk behaviors, we examined their distributions at a 12th-quarter time horizon.

For each bank, we calculated the average value of such ratios from that distribution, as well as the 99th percentile. We introduced “increase factors” as quotients of the adverse 99th percentile over the average value of liquidity and credit ratios, measuring the banks’ relative sensitivity to extreme outcome scenarios.

Plotting average liquidity and credit ratios against their respective “increase factors” allowed for comparison across the benchmark sample. One notable observation was the presence of two types of outliers.

Some banks showed significantly higher expected levels of risk exposure than their peers, indicating structural vulnerabilities even in stable market environments. Others demonstrated large “increase factors,” indicating insufficient protection against stress scenarios. Banks that fell into both categories faced the most exposure and should therefore be closely monitored.

In Figures 1 and 2, the outliers are marked as red dots and the borderline banks as orange dots; the blue dots represent every other institution.

This dynamic approach enables us to pinpoint exactly which parts of a bank's balance sheet — and which aspects of its business strategy — are most likely to cause problems. In short, it offers a level of transparency and detailed analysis that can help us identify why certain banks face increased credit and liquidity risks.

Our research across various outlier banks revealed that, for most banks, increased risk (higher credit and liquidity ratios) is caused by growth in numerators: e.g., a rise in loan delinquencies.

Analyzing Liquidity and Credit Risk Scenarios

Figure 3 depicts scenarios where liquidity risks increase. Interestingly, these scenarios can occur even in a relatively healthy economy. The key factors here are a rapid growth in short-term deposits and short-term interest rates staying higher than long-term rates.

Source: Straterix

Figure 4 illustrates scenarios that lead to increased credit risks. These are typically associated with recessionary conditions, marked by wider credit spreads and higher unemployment rates.

Source: Straterix

In some cases, a prolonged, slightly inverted yield curve can also drive up both credit and liquidity risks. This happens even when interest rates are trending downwards – particularly when combined with around 3% annual inflation, slow or slightly negative GDP growth, and elevated credit spreads and unemployment rates. Such conditions are indicative of a mild recession or a stagflation environment.

By linking bank-specific data to auto-generated scenarios, we can identify not only the most likely and dangerous outcomes for a particular bank but also determine if these scenarios are systemic (affecting the entire financial system) or idiosyncratic (specific to that bank).

Risk managers, supervisors and policymakers who use this hybrid approach can spot early warning signals and devise appropriate contingency plans. It's similar to diagnosing a potentially deadly disease early on, with recommendations that could range from monitoring the situation (keeping an eye on potential risks without taking immediate action) to applying light interventions to prevent risks from escalating. In some cases, it can even lead to a bank taking decisive steps to address severe risks right away.

Parting Thoughts

Future stress tests can be improved by linking historical time series of banks’ segments to macroeconomic variables and by projecting key credit and liquidity risk factors. This can be achieved through a variety of techniques, including regression with regularization and comprehensive, auto-generated scenarios.

Combined, these techniques allow for consistent risk assessment across any bank and its peer group. They can, moreover, help banks’ supervisors identify potential systemic risks in a timely manner.

Adopting a proactive approach to risk management, and to stress testing in particular, enables firms to enhance their resilience, ensure regulatory compliance, make informed strategic decisions, and contribute to the stability of the financial system. 

Alla Gil is co-founder and CEO of Straterix, which provides unique scenario tools for strategic planning and risk management. Prior to forming Straterix, Gil was the global head of Strategic Advisory at Goldman Sachs, Citigroup, and Nomura, where she advised financial institutions and corporations on stress testing, economic capital, ALM, long-term risk projections and optimal capital allocation.