Increasingly, regulators are pressing institutions to step up their A/L management in terms of market risk modeling, ALM management, and board awareness. Throughout this series of articles, we will point out various perspectives on A/L modeling from a "better practices" and regulatory insight perspective.
While addressing regulatory compliance is important on many levels, management should consider the strategic benefits of utilizing these powerful tools. All too often, we see institutions simply plugging their A/L model with numbers and assumptions just to satisfy regulatory requirements, which lends credence to the familiar phrase "garbage in – garbage out." This is why, as a "better practice," managers should focus on transparency when developing and managing the critical components of the A/L modeling process.
This series of articles will break down the important components of A/L modeling as noted below.
- Input
- Processing
- Assumptions
- Output
- Model Validation or Review
- ALM Policies and Education
First, we will concentrate on input and processing. The second article will discuss model assumptions and output. The third article will discuss the importance of model validation and ALM policies and education.
Input
Data is critical! Having comfort in understanding what is feeding your modeling system is something that is frequently taken for granted. The quality of data drives the capabilities of every aspect of A/L management. As A/L managers, we often assume that information feeding the model from the bank's core systems (in addition to external resources such as Bloomberg) is accurate, but is it?
We frequently identify breakdowns in the data management process during our source data model validations. While information (data) is entering the model and managers are usually comfortable with the process, they become shocked to know that certain balance sheet products have not been captured in their model appropriately. More often than not, A/L characteristics from the source samples (i.e., actual copies of loan notes and time deposit certificates) such as step-up options, balloon dates, teaser terms, period and lifetime caps/floors are not loaded onto the core system accurately or consistently.
This problem inherently begins the "garbage in" phase. If important A/L characteristics, especially optionality, are not flowing through the model, then the model can potentially understate or overstate your institutions income, liquidity, and economic value simulation results through various interest rate scenarios.
We recommend that A/L managers periodically sample various source data from a variety of your balance sheet products. Use this source data to trace the A/L characteristics of the respective product (instrument) through the entire information flow process. This is a better practice that can enhance the accuracy of your ALM reporting, as well as demonstrate proactive management to your regulatory oversight.
Processing
Secondly, we discuss the processing component of ALM modeling. As bankers and ALM practioners, it is difficult, if not impossible, to understand the proprietary engines behind most modeling systems. Whether your institution outsources the ALM process or models risk using an "in-house" model, as a better practice, management should evaluate the capabilities of the model in relation to the size and complexity of the institution's balance sheet.
While cost considerations are always a high priority when evaluating a risk management tool, this factor alone should not be the limiting factor. A well-designed ALM reporting system can, in effect, pay for itself when used strategically, i.e., using the respective model to perform a "what if" scenario for a leverage transaction, which enables management to see how the transaction will affect the entire balance sheet (economic value/durations), income stream, and liquidity.
Evaluating your ALM model or the purchase of a new ALM model is analogous to shopping for a new car or debating whether to trade in your current car – clearly not an easy decision given the amount of vehicle choices and various levels of quality available. Do you want performance, space, economy, or durability? After determining what you want out of the car, checking the specifications, and test driving the options, you can usually reach a clear decision, invariably knowing that you will have to live with some limitations, as most vehicles are hard pressed to answer all of these requirements perfectly. The same thought process is appropriate for evaluating your institution's ALM model.
All models have limitations; however, certain models have severe limitations that prevent the model from being able to accurately capture market risk on a robust and granular level. For example, better models are capable of modeling balance sheet products at the instrument level versus aggregating the product at the line level.
Why would this be important? Let us say, for example, you have a loan portfolio product that has various caps ranging from 8% to 12% and the current interest rate on this product line is 7%. If we aggregated the product line, the average cap would theoretically be 10%. Therefore, the product line on an aggregated basis would not exceed the interest rate cap in the +100, +200, +300bps scenarios (simulations), which would have the effect of understating sensitivity. On an aggregated basis, it would require a +400bps shock before the model captures the interest rate risk exposure of this average cap. Conversely, if this product is modeled at the instrument level, then each individual interest rate cap will be analyzed. This approach inherently creates more robustness in your modeling process, allowing management and the board to understand the institution's market risk at a more granular level. Our outsource reporting solution utilizes instrument level processing for all of our clients.
While this is only one example of a model limitation, there are many limitations and considerations that management must analyze when evaluating an ALM model.
As a better practice and matter of regulatory prudence, executive management should periodically evaluate the abilities of the bank's ALM system to ensure your risk measurement tools are keeping pace with the growth and degree of optionality in your institution's balance sheet. These findings should be well documented and reported to the board regularly. Sometimes, this process may require outside assistance in order to get a "fresh perspective." Certainly, these considerations are left to the discretion of executive management and the internal capabilities of the bank's staff. The Ba/lance team can provide assistance with this service in various forms, including formal model validations and comprehensive consulting.