Four Rising Regulatory Expectations Bankers Should Anticipate

If banking executives could count on anything after a year like 2023, it's more regulatory scrutiny for interest rate and liquidity risks. How will that scrutiny translate into questions during an exam? Executives should prepare for these four new expectations.

Interest rate and liquidity risks have been on the minds of most banking executives for more than a year. As elevated rates slow the U.S. economy in 2024, regulators are upping their interest, saying they will “heavily scrutinize” interest rate and liquidity risks in upcoming exams.

Todd Harper, chairman of the National Credit Union Administration (NCUA), for example, informed Congress in November 2023 that credit unions have shown “signs of financial strain” on their balance sheets. Martin Gruenberg, chairman of the Federal Deposit Insurance Corporation (FDIC), observed the same when he said that banks “continue to face significant downside risks from the effects of inflation, rising market interest rates, and geopolitical uncertainty.”

As examiners seek to ensure institutions are prepared for what may come in 2024, banking executives must know: What does “heavy” scrutiny look like? Here are some angles they may not expect, as well as tips to prepare to ace their exam after 15 years of deposit stability in the extreme.

Leave Behind Generic Assumptions

The regulators will be unsatisfied with the use of generic industry assumptions to calculate deposit betas, decays, and loan prepays.

During my 30 years working with institutions on asset and liability management, the common question is: How will you determine the value of non-maturing accounts, such as savings, checking, and money markets? Because these accounts don’t have stated maturity dates like loans and CDs, financial institutions must use a discounted cashflow formula based on the expected future principal and interest payments for those accounts. Examiners want to know how your assumptions were formed for the life of those balances.

It makes sense that examiners will focus on the lifespan of non-maturing accounts — they greatly impact the equity value and volatility — but that does not mean finding good assumptions has been easy. In fact, it’s proven quite tricky.

Before 2022, rates were steady and trivial for so long that historical data had a unanimous outlook for rates: Even if rates rose, they would decline again. That made high-fidelity assumptions difficult to pin down.

Even the period of rate increases before 2006 offered only partial direction because the industry is so different today. Money can move much more quickly through technology; many of those tools and products were not around the last time rates climbed by hundreds of basis points. Deposits also didn’t move immediately in 2022 after the Federal Reserve began raising rates. Each of these new factors limited the usefulness of historical data for liquidity and valuation models.

The industry had not been in a year like 2023 before.

After 18 months of rising rates, banks and credit unions have more information. Review how you’ve moved your deposit and loan rates during the last 18 months. I’d bet that the rates are fairly different than how you previously had your assumptions set up in your model.

Now that we’re through half of a rate cycle, you have data on how deposit balances and loan prepayments have changed. Is the institution still using generic assumptions? Can any be improved based on the firsthand experience of 2022 & 2023? Use that data to help establish and support your modeling assumptions.

Read more: How a Community Bank Helps SMB Borrowers Cope with Rising Rates

Plan for Storytime During the Exam

Given the last two years provided lived experience during rising rates, regulators want to hear about the actual historical behaviors at your institution. They also want to hear how new data from the last year informs institutions’ projections and modeling assumptions.

For example, examiners often want to see four or five scenarios for liquidity testing. We see (and recommend) institutions run at least one “most likely” or “business as usual” scenario in Kinective’s Risk Analytics model, which provides your baseline liquidity. Then, include scenarios reflecting varying degrees of stress: moderate, severe, and extreme. (There’s one other scenario gaining traction in the industry as well, which we’ll discuss in a bit. It’s called reverse stress-testing.)

Testing All the Variables:

In order to have a successful — and thorough — stress test, Kinective advises at least four or five scenarios.

Be prepared to tell the storyline for each of the liquidity scenarios. If you’re talking about the moderate stress scenario, what could trigger that for your institution? What is it in your local economy that could cause your depositors to start withdrawing balances? Is there a major employer in the area that could close a plant or have layoffs? Is there something in the national or global economy that could have an impact on the local market? Storylines for your most severe stress scenario could be a negative news story that triggered a deposit run.

Beyond storylines that can cause the different scenarios, examiners will also want to know the ending in terms of the impact for the institution. For example, what happens to your CD renewal rate? As CDs mature, what percentage of depositors will roll the money into the next CD? What happens to loan originations? Are there more draws on lines of credit?

Be ready to explain how your assumptions are based on the story of 2022 to 2023 and how your scenarios relate to the unfolding narrative in the future.

Back-Test Your Models

Regulators expect to see back-testing in upcoming exams. They’re looking for you to take output from a previous model run and compare that to your actual experience.

For a long time, comparisons between models and recent history were boring because rates weren’t changing. With the rise in interest rates, you’ve got ample opportunity to test your betas and see if the ones you used in your modeling are consistent with what happened.

Back-testing aims to ensure that your model is calibrated properly, that it’s operating off good starting data, and that you have sound assumptions.

Leaders should also be prepared to explain any key differences. When you compare your forecast from a prior report to your actual results, they’re rarely going to match. We recommend a rationalization process.

“With the rise in interest rates, you’ve got ample opportunity to test your betas and see if the ones you used in your modeling are consistent with what actually happened.”

For example, did loan income end up higher than expected? Well, maybe it’s because the portfolio grew. You have good loan demand, which explains the lion’s share of the difference.

But let’s say that your loan income prediction is higher than your actual result quarter after quarter. That’s a good example of a situation where you need additional analysis. What is the source of that differential? Is accurate data on the individual loans being passed into the interest-rate risk model? Is there something wrong with the underlying data? Do you have bad assumptions in the model that are causing this result? Once you find the source, you can take action to remedy the issue.

The back-testing report should give you confidence that the model predicts future results accurately. Likewise, it should give your examiner confidence too.

Read more:

Stress-Test Key Modeling Assumptions

We also recommend stress-testing model assumptions one at a time. Take deposit betas as an example. This is the pricing discretion you practice in terms of the rate you offer to pay on savings and checking accounts. Say you have a base assumption on your savings accounts of 30% sensitivity. That means if interest rates go up 100 basis points, you will raise the savings rate by 30 basis points.

You might increase those deposit betas by an additional 20% for the stress test. So, in our example, instead of savings accounts moving up 30% based on an increase in market interest rates, they move up 50% now. That will result in higher interest expense and, therefore, lower net interest income and a bigger decline in earnings than the model suggests.

Three key modeling assumptions that drive results are:

  • Pricing betas
  • The decay assumptions on your non-maturity accounts
  • Prepayments on your loans and, to some extent, your securities portfolio if you have mortgage-backed securities or collateralized mortgage obligations.

Stress testing demonstrates the impact of being wrong about these key assumptions. It also shows your ALCO members and your board how important these calculations are, driving home the need to allocate adequate resources to establish good assumptions.

While doing these stress-testing scenarios, you should review the information with ALCO. The committee may want to adopt a slightly more conservative set of assumptions in certain areas. Whatever the result of those discussions, adjust your model as needed. Make sure the minutes of the ALCO and board meetings document the activity so you get credit for it with examiners.

How often should you conduct the stress test? Some institutions do so once per regulatory cycle — the bare minimum. But we recommend stress-testing once a year.

Review Less-Discussed Scenarios

The regulators are looking for a series of non-parallel rate scenarios. These scenarios involve moving long-term and short-term rates in different ways — a steepening of the curve, a flattening of the curve and an inversion of the curve — to assess the impact.

A rate ramp — which you typically don’t hear much about — is another rate scenario examiners have asked for lately, along with the standard rate shocks.

Ramps are where rates move gradually over time. You could do the usual stair step but also opt for a “most likely” rate scenario typically driven by the forward curve on U.S. Treasury rates.

How often should you run your rate scenarios? Some institutions do so every quarter, but the more common practice is once a year.

You could run an infinite number of possible rate scenarios, so you have to make it manageable. A good practice is establishing a handful of standard scenarios and running those each time. Then you can not only see the result now, but you can see how it compares to the same type of rate change a year earlier. If you achieve an unexpected result, determine which asset or liability categories are experiencing the big adverse changes that caused your overall exposure to be more severe than what the parallel rate shocks indicate.

Limiting the unexpected should be your goal. That’s what the regulator, your institution’s executive team, and your board want: No surprises. Any time you find a surprise, show that you’ve done extra analysis.

This article is an overview of likely regulatory expectations based on the experiences of banks and credit unions that work with Kinective. Institutions should engage the author directly for consulting tailored to their unique situation.

John Anton is a senior vice president at banking services and technology company Kinective.

This article was originally published on . All content © 2024 by The Financial Brand and may not be reproduced by any means without permission.