“Model risk is the chance that incorrect financial decisions are made based on faulty projections from a financial model. Asset liability management (ALM) models are subject to model risk, especially in interest rate risk (IRR) applications where fundamental balance sheet decisions are made based on indicated sensitivity positions.”
– William McGuire, “Defining Accurate Core Deposit ALM Model Repricing Inputs,” Bank Asset/Liability Management, Vol. 26, No. 9 September 2010, p. 1-5.
From late 2008 through early 2016, there was little need to examine core deposit repricing thoroughly. The target federal funds rate remained unchanged with an upper target of 25 basis points (bps) from December 2008 through December 2015 while the 1-year Treasury rate moved only a few bps from month to month, changing from 49 bps in December 2008 to 53 bps in February 2016. Since then, however, market rates have risen; the upper target funds rate has increased gradually to 2.50% and the 1-year Treasury rate has climbed similarly to 2.55%.
The overall economic landscape for banking has changed dramatically in other ways as well. For example, banks as a whole had $2.0 billion in excess reserves in mid-2008 but held $1.5 trillion in excess reserves as of January 2019, albeit down from $2.7 trillion in mid-2014. These changes make it prudent to re-examine the repricing of core deposits and the implications of those repricing decisions on the amounts and retention of core deposits and ultimately on the earnings implications for a financial institution. Do you want to be as aggressive when increasing deposit rates when there is a tremendous amount of excess reserves in the banking system vs. when virtually no excess reserves are available?
This is the first in a series of articles presenting an integrated methodology that examines both core deposit repricing and the implications of that repricing for core deposit balances and retention. The model outputs can then be employed in ALM model IRR testing applications and should provide insight into the implications for repricing on earnings. This approach is forward looking, forecasting deposit rates and deposit balances. It is institution specific, using the institution’s history in the context of general market conditions to determine its likely future trajectory. And variations of the methodology can be employed by banks and credit unions, large and small. While the model is based on an advanced statistical methodology, it also allows input from management. In coming weeks, we will present factors to consider when designing the model, the data required for a state-of-the-art model, the modeling process itself, and a monitoring and validation process for the model.
The Starting Point
Before examining the repricing process, it is critical to understand the institution’s environment, something that may have changed dramatically from the pre-2009 period when deposit rates last moved substantially. The classic distinction on core deposits is that between zero or low-rate-paid account types vs. premium or high-rate-paid account types. The former would include non-interest rate checking, interest rate checking (negotiable orders of withdrawal or NOW accounts), saving and low-tier money market deposit accounts (MMDAs) while the latter would include upper-tier MMDAs, premier checking and in some cases even CDs. The former are considered traditional core deposits with the perspective that depositors are primarily motivated by factors like liquidity, service and convenience. These deposits are considered relatively insensitive to interest rate considerations, and repricing historically has generally been sluggish. The critical component to retaining these balances is viewed as maintaining service and convenience. In contrast, premium rate core deposits are considered much more sensitive to interest rate considerations, and institutions historically have been much more aggressive in repricing these products in response to market rate changes.
The distinction between repricing of interest checking vs. different tiers of MMDAs vs. CDs is well known, but it should be just the beginning of the institution’s analysis of its environment and depositors and how that impacts the rate sensitivity of its products. Consider four additional factors before conducting any repricing analysis.
First, what is your deposit base and how have you repriced historically? Have you emphasized service and convenience or have you emphasized the competitiveness of your rates? If you emphasized service, your current depositors are going to expect that to continue and are likely to be relatively insensitive to market rate increases that you do not match. In contrast, if you have emphasized maintaining competitive rates and did not raise rates as market rates increased, you should expect much greater interest sensitivity and an outflow of funds when market rates rise. Your depositors will be sensitive to your historical behavior, and their future behavior will be based in part on how consistent your future repricing will be relative to your historical repricing.
Relatedly, there has been substantial conversation in the past 10 years about different vintages of deposits. That is, deposits placed in your institution before the financial crisis may differ in their behavior and rate sensitivity from deposits placed during or after the financial crisis. For example, after the crisis your depositors may be more interested in safety rather than return. In general, market conditions may have led depositors to behave differently at different points in the business cycle. Alternately, deposits placed at different times could have different rate sensitivities if the institution changed strategies, e.g. switched from a service/convenience emphasis to an interest rate emphasis or vice versa. Whether your customers changed their deposit behavior or you changed your pricing behavior, you should consider whether repricing of different cohorts of accounts yields different rate sensitivities.
Second, do your depositors behave differently depending on what their relationship is with you? Does having multiple accounts or direct deposit or automatic bill pay alter depositors’ rate sensitivities? When there are multiple relationships, switching financial institutions becomes more costly, service becomes more important, and the rate sensitivity of deposits may decrease. You should also consider whether accounts using mobile banking differ in their rate sensitivity from accounts that do not use mobile banking. Do your mobile banking users behave differently and have higher or lower balances or a different rate sensitivity? The answer to that question will impact how valuable those accounts are for you.
Third, what is the behavior of your competitors? Are you in a market with many or few competitors? Does the competition focus more on convenience/service or competitive rates? Have your competitors’ behaviors changed since the financial crisis? Institutions like Wells Fargo and Bank of America have greatly increased their national footprint with the potential for even greater nationalization of rates, with markets from Boston to Los Angeles much more likely now to have the same pricing than even 10 years ago. Having a perspective on the importance of these questions and the likely answers to these questions will inform how to structure the question of how to analyze your repricing decisions.
And fourth, have there been any unusual or idiosyncratic events that would have impacted your deposit base? Perhaps the most dramatic example of this type of event would be the financial crisis. At the outset of the crisis, there was a rush to safety with substantial funds leaving more risky assets like stocks and being moved into safe and more liquid assets like MMDAs. Those balances were sometimes labeled surge balances with an expectation that they were temporary and would likely leave once market conditions had returned to normal. The expectation was that surge balances were only temporary funds and that their rate sensitivity might differ from the standard rate sensitivity of funds in that category.
Some financial institutions may be affected by all four of these factors while others may feel the impact of only one or two. It is not necessary to have a definitive perspective on the importance of each of these four factors, but it is critical to ask how each of these factors potentially relates to your institution before undertaking any analysis of deposit repricing. The structure of your institution’s answers will inform how to structure any analysis of repricing and deposit behavior. The key to generating a great model is situational awareness, understanding the factors that potentially impact balances and interest rate sensitivity.
The four factors discussed above are critical components to situational awareness. Other factors that need to be considered are the general state of the economy, including overall growth, unemployment and inflation. Additional factors will depend on the institution – with a bank in Texas possibly heavily dependent on the oil industry, a savings bank in Las Vegas more concerned with tourism, and a credit union affiliated with an employer paying particular attention to the state of the employer’s industry.
Coming up next: What institution-based data do you need to generate a best-in-class model?