Last week we covered the first half of the OCC letter requesting information from client banks on their interest rate risk (IRR) model. Along with the letter, you may have run across some forms to be filled in with results from your model. This post and the accompanying webinar are meant to clarify the letter and terminology. Separately, but related, we have produced a video and guide for completing the OCC EV IRR Data Form for Financial Compass Clients (to be released soon).

The remaining items are where many of our clients have questions. Whether you need information on one or all, Plansmitties, fear not. We will do our best. Be sure to share this with friends and request a link to the recording of the webinar that we presented. I had a special guest join me to give his examiner perspective (ok, it’s Dave Wicklund, Advisory Services Director and former FDIC Examiner. We like to tell him he’s special).

5. Does the interest rate risk model’s output under a stress scenario identify which assumptions are the most significant in determining the impact of rising interest rates on the bank’s earnings and capital base?

So many variables, so little guidance. I am not aware of a model that is able to look at all variables simultaneously (prepayment, decay, pricing models, repricing) and determine what portion of the risk is contributed by each. That would be like solving an algebra equation such as x + y + z = 100; what are the values of each variable (for that I would have to consult my 14 year old)? In reading through more carefully, the best we can determine is this relates to sensitivity testing, unquestionably a hot topic among examiners.

Sensitivity Testing: Change one assumption at a time and recalculate – it’s as simple as that. For instance, if your bank-specific decay rate is 60 months on a deposit category, change it to 30 or 12. Are you still within policy for value at risk? Change a few more, one at a time using extremes such as abnormally high or low prepayment. Try to determine how far a variable can change before there is a breach in policy – trial and error method, but not very difficult. Does your model or your model’s service provider allow for such analysis?

6. How and when were the decay rates for non-maturity deposits developed? If based on the bank’s internal or external data, were they drawn from a period similar to the base case or stress scenarios?

You need bank-specific decay rates for your model but this is no longer an option. This means accessing your historical deposit balances, run-off rates, and account lives as well as other information to determine the longevity of your deposit base and the sensitivity the base has to changes in interest rates. National or third party data likely won’t be acceptable to your examiner, if any is available at all. If there is one assumption take-away for you to understand it is this: all assumptions need to be specific to your bank’s current and historical experience.

Given the wide variety of modeled stress scenarios, (-400 through +400, in 100bp increments), bank-specific data is rarely available for each of these stress scenarios.

7. What are the penalties for early withdrawal for certificates of deposits (CDs)?

While it is likely that you can easily answer this question for your institution, the underlying question is, when rates rise, will your customers pay the penalty in order to reinvest the principal in a higher yielding certificate? To get into the stock market? Model early withdrawals just as you would model loan prepayments. In effect, that’s what it is, prepayment on CDs. Regulators are very concerned about the surge of CD deposits taken by banks where the depositor has no loyalty to the organization. These are considered parked for a period of time until something better comes along. If the alternate option is so much better that it offsets penalties, it could negatively impact earnings and stress liquidity. While on the subject, consider this scenario when you develop your stress scenarios in your cash flow model.

8. Are mortgage prepayment speeds altered for IRR stress scenarios? If so, do you include both those mortgages on the bank’s books and those underlying securities in the investment portfolio?

Nothing new here. At a minimum, you need to be calculating earnings and value at risk across 8 parallel rate shock scenarios. Are you using a different prepayment speed for each of the +100, +200, +300 and +400 levels (down rates too)? How do you determine what prepayment rate to use? Analysis of the last 12 months can give you the rate for the base case (0 rate shock), but the last time there was a significant rise in rates was (roughly) March 2004 to March 2006; down rates to consider are March 2007 – 2009. Do you have the historical data to reasonably estimate prepayment in these rate environments? Better yet, will your customer base behave the same way this time? For answers to these questions, you will probably need an advisor with experience in this analysis to develop credible answers.

9. Do you use derivative or wholesale funding products to hedge the bank’s exposure to IRR? If so, please describe the products used and what it is hedging on the bank’s balance sheet?

Most of our clients don’t use bank-wide hedges – at least we don’t see them when setting up their interest rate risk models. Direct hedges (e.g. swaps used to convert a large fixed-rate loan to a variable payment stream) seem to be more what they’re trying to understand –"…what it is hedging" are the key words in the question. Your interest rate risk model isn't necessarily going to link the hedge with the specific loan, so documentation on the specifics will be required.

Wholesale funding (customer wants a $1M 10yr fixed so the bank borrows from the FHLB 10 years) is a match funded event and, like the hedge, the two sides aren’t necessarily linked in the interest rate risk model. For what it’s worth, Compass does allow for hedges and wholesale funding to be included in the interest rate risk analysis.

10. Are non-maturity deposit decay rates and CD breakage rates increased for the IRR model’s stress scenario?

Here we see the continuation of question 8 applied to the deposit side. While the answer is much the same, the fact that it is being asked at all is new. Maybe a bit disconcerting is that, despite the request from examiners, there is likely to be no data to support varying rates by scenario. As we previously noted, given the wide variety of modeled stress scenarios, (-400 through +400, in 100bp increments), bank specific data is rarely available for each of these stress scenarios.

11. List of established IRR limits. Are these limits approved by the Board of Directors?

Your interest rate risk policy should include limits of all types. Perhaps a more meaningful question is when was the last time the board approved or reaffirmed the limits? At a minimum, your limits should include maximum earnings at risk and value at risk (% of earnings and value change from the 0 point). Many are being asked to include the limits at each shock interval -/+ -100 through +/- 400 basis points, and nonparallel shifts as well.

12. Is the investment portfolio separately modeled for interest rate stress scenarios? If so, how does the model incorporate the impact on individual securities whose cash flows and market values are highly sensitive to changes in interest rates, such as Federal Home Loan Bank issued structured notes?

There should be an investment portfolio line or section in every model (both earnings sim and capital valuation). Most models allow users to make assumptions regarding the timing of cash flows (prepayments and call options). Regardless of the interest rate risk model you use, the underlying data needs to support the analysis. That is, assuming the model incorporates the impact of cash flows, does your bond accounting system or source of bond data provide the cash flows?

However, getting the modeled scenarios may be as easy as referring to your bond portfolio report. Most reports include the market value of the portfolio under stress scenarios. The bond dealer typically gets this from sophisticated systems such as Intex, Bloomberg or BondEdge. These systems have access to the data needed in order to generate cash flows for the required analysis. Therefore, answering yes to this question could be as simple as entering the results provided directly into the interest rate risk model. If you cannot override the model calculated results, at least compare them to the results provided by the bond report.

13. Do you factor in the IRR model’s base case results into the bank’s annual budgeting process?

We wonder why the base case results are relevant. If you believe rates are going to spike up 100bps then why wouldn’t we use the results from that level? Referring back to part one of this post, we at Plansmith are excited to see that planning (budgeting and forecasting) is getting attention in the regulatory world. However, a second thought on this. Why would you include ANY rate shock results if you are projecting asset growth or decline since interest rate risk analysis is based on "no–growth" scenarios and no change in composition of the balance sheet per regulatory guidance?

14. Do you factor in the IRR model’s stress scenario results into the bank’s capital plan?

We realize most institutions don’t have a formal capital plan; though, it appears this may be a future area of regulatory emphasis. All risks, including interest rate risk model results, should be considered in the budgeting and strategic planning process. This question seems to imply there is increased regulatory emphasis on capital planning and policies beyond the historical two paragraphs in the policy simply addressing minimum capital ratios and dividend payments.

That covers the remaining questions. For even more information, watch the webinar that we presented.