By Mark H. Smith, Founder


ALM models have come a long way. Over the last three decades they have evolved from simple spreadsheets estimating a gap between the repricing points of assets and liabilities to sophisticated computer-driven software that considers a nearly unlimited number of variables in the process of estimating interest rate risk (IRR). One thing that hasn’t changed is the importance of the role that assumptions play in the estimated modeling outcomes. In almost all cases assumptions drive the outcomes. In this article we will identify the characteristics of sound assumptions and their implementation.

We also understand that there is no perfect set of assumptions. Good assumptions are a moving target. They depend on many variables. Some of those variables are straightforward, such as economic and financial data. Some are not so easily defined because they hinge more on the behavior of counterparties than a rational response to an economic or financial input. In the end, we must make the best assumptions that we can. When we utilize them we must understand the limitations of modeling and be alert to outcomes that may be suspect.

Key assumptions should:

  • Be consistent with historical performance
  • Consider events and circumstances that may alter historical outcomes
  • Be institution specific when practical
  • Be carefully defined and documented
  • Be tested as to their impact on modeled outcomes
  • Be included in the system of internal controls
  • Be understood by the users of the model output
  • Be reviewed at least annually

Consistent With Historical Performance

Assumptions should take into consideration historical performance. It’s true that history will not necessarily repeat itself. But it’s also foolish to ignore history. While assumptions should be formulated with an eye on the past, those factors which may cause future outcomes to differ from the past should be considered. For example, regular shares have long been considered to be a stable source of funds with a low level of rate sensitivity. This was borne out in the 2004/2007 interest rate run-up. It would be easy to consider all of your regular shares to be long-term, stable funds with a low level of sensitivity. However, this assumption may turn out to be at least partially incorrect.

Consider Events Which May Alter Historical Outcomes

For example, several factors which occurred in recent years may cause us to modify our historical outlook as to regular shares. For a period of time after the financial crunch of 2008, credit unions became a destination in the flight to safety for funds departing riskier environments of the equity and some corporate debt and mortgage markets. The result was that many credit unions saw a surge in incoming share deposits in 2009 and 2010. The ultimate disposition of these shares remains uncertain, but it may be unwise to assume they will respond to an increase in rates in the same way as traditional regular shares.

Secondly, over the most recent years, the Fed’s imposition of historically low rates appears to have caused some savers to become ambivalent about savings rates. In other words, why care for a miserable 20 or 30 basis points? Funds from share CDs, which are usually considered to be rate sensitive, appear to have flowed into regular shares and remain, simply because of the ambivalence of their owners at this time. When rates increase, those CD savers may rediscover their rate-sensitive ways.

Be Institution Specific When Practical
In a perfect world, assumptions would be specific to your credit union; that is, they would be derived from the set of circumstances–economic, financial, and behavioral–in which your credit union operates. Some credit unions don’t have the resources to perform this analysis. In that case, indexes are often used in place of institutional-specific assumptions. If an index is utilized, its parameters should match those of the credit union’s as closely as possible. Additionally, the index should be recently updated. Older indexes such as the NCUA NERA Study and the OTC deposit index are extremely outdated and their conclusions are subject to question.

Be Clearly Defined and Documented
A summary of key assumptions should be presented with the outcomes to the management and ALCO of the credit union. A discussion of the assumptions should be included in the ALCO minutes, including highlights of significant changes and their apparent impact on forecasted interest rate risk. The minutes of the ALCO should note changes to key assumptions.

Be Included in the System of Internal Controls
Control over assumptions, their formulation and preservation should be included in the system of internal controls. Conflicts in the formulation and implementation processes should be avoided. The modeling function should be reviewed periodically by the credit union’s internal auditor. For smaller credit unions not staffed with an internal auditor, the fall-back would be to the supervisory committee and/or the outside auditor engaged by the committee.

Be Understood by the Users of the Model Output
The board of directors retains the overall responsibility for asset and liability management (ALM). Given the significant impact that assumptions have on model outcomes, all board members should have a basic understanding of key assumptions.

Be Reviewed Annually
A review of the management process for interest rate risk is required annually by the recently enacted IRR Regulation promulgated by NCUA. Each credit union has the discretion to perform this review internally or to engage an outside firm to perform a Third Party Review or Validation. Inasmuch as the assumptions utilized by the credit union to model interest rate risk (IRR) constitute a key role in the modeling outcomes, their inclusion in the internal review or independent third-party review should be considered mandatory.

Lastly – the Sniff Test
Do your assumptions make sense or are they pulled out of a black box that you don’t understand? There is some complexity to the generation of assumptions. However, if they do not seem realistic or rational to you, we suggest you delay their adoption until you fully understand their makeup.