As we transition out of a record low rate environment, many institutions are looking for answers. Do we have an adequate funding strategy? Are we allocating that funding to the right asset classes? While many questions like this are top-of-mind, one important question involves interest rate sensitivity: how effective is my ALM process?

A valuable ALM process delivers actionable information. Valuable managerial information is timely, accurate, and comprehensive. To improve and gauge the quality of an ALM process, consider the following three tenets: data, cash flow integrity, and analytics. These tenets, as illustrated in Figure 1, are sequential; deficiencies in any one of them flow through the process to the very end result.


As the adage goes, “first things first”. Data are the starting point for any ALM process, and successful ones will have similar attributes. Let’s examine some of the finer points:

Completeness and Accuracy

These two ideas should not be mutually exclusive. As demonstrated in Figure 2 modeled results are subject to the same limitations as the underlying data. As the complexity of assets increases, the required data to ensure accurate modeling increase as well.

Below is an adjustable rate mortgage (ARM) run under three separate conditions: full data, no period caps, and no period or life caps. Powerful cash flow engines and robust analytics cannot make up for poor data quality. Trust in your ALM model is crucial to sound decision making, and that all starts with data.

Cash Flow Integrity

Once we can ensure data quality, the next driver in value creation is an accurate cash flow projection. The following are important characteristics:

Meaningful Cash Flow Projections

Financial institutions typically host complex instruments on the balance sheet, and thus, cash flow modeling deficiencies will drag down front-end analytics and decision making. Cash flows are the basis of all risk analytics and risk management techniques.

Certainty of Principal and Interest Cash Flows

Cash flow reliability becomes more difficult as assets increase in complexity, thus driving the need for a more powerful cash flow engine. Zero coupon bonds and non-option bullets have certainty around principal cash flow, but many assets do not have this luxury of cash flow clairvoyance. More complex assets, like mortgage assets and other instruments with optionality, have more uncertainty surrounding future cash flow.

Although complex, modeling cash flow optionality is possible, and is a must. And while an individual mortgage asset may have uncertain cash flow, there is a much greater degree of certainty around large pools of assets, leading to the importance of cash flow model tuning.

Figure 3 rank orders a set of generic asset categories, starting with least complex at the bottom, to most complex at the top.


Once we ensure data integrity and extract meaningful cash flows, the final step is analyzing the results to determine interest rate, liquidity, and credit risk exposure, highlighted in Figure 4. To properly inform decision makers, ensuring short feedback loops is important not only to decision-making effectiveness, but also to reacting quickly to early warning indicators.

Comparing model estimates to actual behavior is critical. It can provide much needed transparency to model effectiveness. Awareness of drawbacks and deficiencies can help ensure management is thoughtful about modeled results.


There’s no time better than the present to evaluate your ALM process. A deconstruction of the three components – data, cash flow integrity, and analytics – could result in eye-opening realizations that enhance not only risk management functions, but also the performance of your institution.

Michael Oravetz

Associate, ALM & Investment Strategy at ALM First