By Stephen Brown Klinger
Many community banks develop their asset liability management (ALM) modeling assumptions based on four measures: the bank’s historic experience, current customer loyalty, economic projections and industry benchmark data.
Before discussing best practices, the need to develop accurate modeling assumptions starts with the fact that the banking industry is currently flush with funding. The FDIC’s Statistics on Depository Institutions Report shows that between 2007 and 2012 total domestic deposits grew from $6.9 trillion to $9.4 trillion, of which $2.1 trillion was in money market deposit accounts. With this “bubble” comes additional liquidity and interest rate risk for the industry when rates begin to rise.
FinPro research reveals that total interest expense tripled during the last rate rise (2004 to 2007). It is vital for planning purposes that a bank makes strategic decisions about how to handle interest rate risk and liquidity when rates rise, incorporate those plans into its modeling assumptions, stress test the results in different economic scenarios, and then reassess its plan.
In addition to internal reasons, regulators are scrutinizing the process for developing modeling assumptions with increasing frequency. The FDIC’s FIL-46-2013 “Managing Sensitivity to Market Risk in a Challenging Interest Rate Environment” and the OCC’s “Fall 2013 Semiannual Risk Perspective” provide key pieces of insight as to regulators’ concerns in the face of rising rates:
- Declines in net interest income
- Deposit run-off
- Constrained loan growth
- Extension risk of MBS portfolios
- Reliance on long-duration fixed-income securities for liquidity
- Net unrealized losses on securities
The forward projections of these items are directly impacted by a bank’s modeling assumptions. In fact, the OCC states “The adequacy of interest rate stress scenarios and the appropriate support for key modeling assumptions (non-maturity deposits in particular) will be a particular focal point [of examinations].”
As noted earlier, there are four elements necessary to develop proper ALM assumptions: historic experience, customer loyalty, economic projections, and industry benchmark data.
Historic Experience
It is often said that understanding history is the best way to predict the future, and bank regulators wholeheartedly agree. For establishing decay rate assumptions this means analyzing three to five years of account level experience and then aggregating the flow of funds to the product level at which you model. A beta value study is a little trickier. It involves basic statistics and necessitates product pricing data back to 2006 in order to capture the last rate move (2000 to capture the last three).
These two analyses are the first building blocks in developing assumptions designed to project the future. However, the connection must be drawn between the bank’s experience and the bank’s current customer base based on customer loyalty factors.
Customer Loyalty
The difference between “core deposits” and customer loyalty is one of the common ALCO misunderstandings. The following conversation demonstrates this point best:
Bank: “We do not have hot money deposits. The majority of our deposits are core, our customers are extremely loyal and will not move in rising rates.”
The Regulator: “But your products are paying 45 bps over the competition?”
Bank: “Well, our loyal customers are very price-sophisticated.”
Yes. No. Maybe? Regulators define a core deposit as deposits under $250,000, but in the real world this in only one component of customer behavior. In the example above it is impossible to describe the customers’ price elasticity without further analysis. The only thing the bank has demonstrated is that its customers are acquaintances who are sophisticated enough to shop for the best rate in town, not a good situation. In order to develop a proper beta value strategy for rising rates the bank must stratify its deposit accounts based on six loyalty factors: customer relationship, price, geography, tenure, frequency of use, and size. Scoring these loyalty factors will generate a propensity to renew score which indicates the true loyalty of each account. The key is that this is based on current customer data. When overlaid with the bank’s historic information, this positions banks with best knowledge to develop funding strategies that build value. Some industry leaders and academics still say, “core deposits drive franchise value.” Do not fall into the “core deposit” trap. Customer loyalty builds franchise value.
Economic Projections
Stress testing is an important and evolving component of asset liability management. While still mandated by some regulatory agencies, instantaneous parallel rate shocks on static balance sheets will soon become a thing of the past. Why? Rates have never risen by more than 200 bps in a parallel manner across the entire yield curve, especially overnight. Nor will they ever in the future.
Additionally, not matching future changes to the balance sheet with the bank’s strategic plan (dynamic modeling) creates a larger separation between the most likely actual depiction of the bank’s future interest rate risk and what is captured in ALM modeling. What this means is that banks must match their strategic planning balance sheet with their ALM model and overlay different practical projections of economic scenarios (adjusting for interest rates, new business projections, and subsequent effects on ALM assumptions) as stress tests to the base case. Only then will management, the board, and regulators get a depiction of the bank’s interest rate risk that is actually useful for planning purposes.
Industry Benchmark Data
The concept of benchmarking a bank’s assumptions and results to industry benchmarks is a straightforward concept. However, of the four steps it is the most commonly overlooked. While assumptions must be bank specific based on historic studies and future projections, the ability to benchmark against industry values is a powerful tool to ensure the reasonableness of non-maturity deposit decay rates, price sensitivity (beta values), and loan prepayment speeds.
Different risk thresholds or assumptions may be more appropriate given the institution’s strategic plan, but in those circumstances it makes it even more important to see the forest from the trees in order to prepare a justification for the difference. Additionally, market intelligence on product benchmarks is extremely valuable in supporting ALM assumptions at new institutions or new products where historic information is unavailable.
The increased regulatory focus on banks’ interest rate risk has fueled an increase in matters
requiring board attention (MRBA) regarding ALM assumptions. In fact, the FDIC’s Supervisory Insights publication of summer 2014 notes it as the fastest growing area for MRBAs among bank’s rated “1” and “2.” Common language includes:
- “Develop and document non-maturity deposit assumptions based on a historical analysis of the bank’s customer base and create reports quantifying rollover risk.”
- “Document support for key bank specific assumptions utilized in quarterly IRR modeling.” “Enhance the ALM policy to better define required modeling scenarios and their associated risk limits.”
- “Reasonable EVE sensitivity … Risk limits must be adequately supported.”
A simple way to avoid these criticisms is to make sure your institution has properly addressed each of the four areas: historic experience, customer loyalty, economic projections, and industry benchmark data.
Stephen Brown Klinger is director at FinPro Inc. Contact him at (908)-604-9336 or sbrownklinger@finpro.us.