Don't Stress Over ALM Stress Testing
The regulatory requirements in today’s world can at times seem overwhelming. Don’t let your ALM stress testing stress you out.
While the majority of data provided to your Asset/Liability Management (ALM) model is real data taken directly from your general ledger and subledger core and bond accounting systems, there are several types of key modeling assumptions that have very significant impacts on your ALM modeling results. Among these are your product pricing sensitivities, known as “betas,” non-maturity deposit lives (or decays), and loan prepayment assumptions.
The regulatory agencies understand this, and as a result, frequently require stress testing to be performed on your ALM model assumptions. At first blush, this may sound like a daunting task, but it doesn’t have to be.
Like with most regulatory pronouncements, the general requirement is described, but no real detail is provided on how to meet that expectation. It is important that your stress test be viewed by your examiner as comprehensive and sufficiently impactful. It is also important that the stress testing assumptions be straightforward enough that your Board and ALCO can easily understand what was done and how significantly it impacts the earnings at risk and economic value of equity (EVE) (aka net economic value or NEV) results of your ALM model. One important benefit of this stress testing exercise is to demonstrate to anyone using the ALM Report how significantly the modeling results can be impacted by changes in key modeling assumptions. If the board and ALCO end up taking the establishment of realistic and supportable modeling assumptions more seriously, then the whole exercise becomes well worthwhile. The last thing you want is for the board and management to be making balance sheet management decisions based on ALM results that misrepresent the actual interest rate risk on your balance sheet.
To that end, here are some suggestions to make your ALM model stress testing regime less stressful.
Keep it Simple
A simple, methodical approach is best. Don’t waste time and energy stressing your modeling assumptions in a direction that causes your measured risk to be reduced. Apply a consistent set of modifications across accounts and avoid micro-managing the details.
For most financial institutions, an increase in the non-maturity deposit betas in the face of rising interest rates will have a negative impact on both earnings at risk and EVE. Consider a simple, consistent approach such as doubling the pricing betas on all interest-bearing non-maturity deposit accounts. It is easy to understand what was done and will have a significant impact on the results.
Non-Maturity Deposit Lives
Your non-maturity deposit life assumptions have a very significant impact on your EVE results. The longer your NMD lives, the more they change in value as market rates change and offset more of the change in value of your longest term, fixed-rate assets. To that end, for most institutions that lose economic value when rates rise, a shortening of NMD lives will increase risk. Consider cutting the lives of your non-maturity deposits in half (or double the decay rate). This will demonstrate the impact on EVE if your customer loyalty is far less than the current modeling estimate.
What tends to increase the measured interest rate for most institutions is a slow-down in loan prepayments. This reduces the rate at which loans reprice and increases the degree to which they change in value as rates change. Consider either a halving of loan prepayment speeds or eliminate them altogether.
Additional information is available by digging into your ALM report detail. Even though you changed all your deposit betas in a single model run, you can easily see how much of the overall change in results was due to just the change in your money market betas, for example.
Don’t overwhelm your Board and ALCO with Stress Testing results. Rather than running your Beta, NMD Life and Decay Stress testing all in one reporting period, schedule these out individually throughout the year. These stress runs do not normally need to be run every time you run a report.
If you follow these suggestions, you should be able to meet your testing requirements while lowering your personal stress level. Happy Stressing.
// John Anton, FIMAC Solutions