Our country and our industry continue to sail uncharted waters. We are now in our 17th month of a “zero to 0.25%” Fed Funds target, and the FOMC continues to indicate a prolonged period before upward adjustments begin. In January, regulators issued new guidance on interest rate risk and indicated a concern about the positions some institutions were taking in this low rate environment that could lead to significant IRR once rates rise.
Some level of heightened concern over IRR exposures is definitely warranted. Unfortunately, recent regulatory guidance is akin to building a Rube Goldberg contraption to open a jar of peanut butter.
Of all the risks inherent in banking, Interest Rate Risk is one of the most talked about yet least dangerous. Credit Risk, Liquidity Risk and even Reputational Risk play much larger roles in the success or failure of an institution. While it is possible for IRR to cause an institution’s failure, I have seen no evidence to suggest that IRR was the primary or secondary cause of any of the over 200 failures since September 2007.
Most community banks operate with a minimal level of interest rate risk. I make this statement based on 17 years of analyzing, on a monthly basis, the IRR positions of over 70 institutions. One needs only to look at the stability of these institutions’ net interest margins across interest rate cycles to see that a combination of the banks’ static IRR exposures and management’s responses to market rate changes minimizes true IRR exposures without the use of sophisticated models. NOTE: This statement does not apply to banks that go wild with callable securities, complex derivatives or other nonstandard assets in an attempt to shore up sagging earnings or deploy excess liquidity. Be smart – some of today’s decisions can have long-lasting consequences.
Whether you use a simple method or Rube Goldberg contraption to open your jar of IRR peanut butter, the stuff inside is still rather sticky. The harsh reality of IRR is that none of the models are right. You can’t model IRR without making assumptions, and none of us are omniscient. Our assumptions, whether few or numerous, introduce error into the system. As the old saying goes, Garbage In, Garbage Out.
Perhaps the greatest problem in the IRR equation is that, in an effort to get a more accurate answer, we must increase the number of assumptions. More assumptions = more chances for error. I can think of a few assumptions that, if made incorrectly, can completely reverse a bank’s apparent IRR sensitivity. If a model is hypersensitive to a necessary assumption that cannot be made with certainty, the result of the model is only as good as the assumption itself.
Assumptions about nonmaturity deposit behavior quickly corrupt both Earnings at Risk (EaR) and Economic Value of Equity (EVE) models. Some of the assumptions answer questions like:
Ø How quickly will we have to increase our NOW, MMSA and Savings rates once market rates begin to move?
Ø Does current pricing include a virtual floor (e.g.: Savings rate currently at 0.50%, even though it would normally price below the Fed Funds target rate) that will protect the bank from increasing cost during the first 50-100BP rise in rates?
Ø Once the virtual floor is cleared, will these rates increment in lock step or at some fraction of market rates?
Ø Will there be a repricing lag?
Ø What do we do with Demand Deposit balances? (Answering this question could take several pages and bore all but the most academic among us…)
Similar questions can be asked of other balance sheet sections (cash flows on loans, optionality of securities, etc.). The short version of this long story is that we make educated guesses that build upon each other until the system resembles a house of cards (covered with peanut butter). When our friendly examiners come to visit, they look at our house of cards and suggest that adding some jelly to the mix might make the output more accurate.
Yes, dear friends, I realize that my analogy has become somewhat silly. So have some of the expectations of our regulators in the IRR arena. Please remember: we are in uncharted waters (sailing beyond the edges of the mapped IRR landscape), and most IRR models do not behave well “way out here”. We continue to use our IMSA model for EaR estimations and the RiskExpert system for annual EVE estimations, but recognize, as should you, that these are estimations made with assumptions. Therefore, they are probably wrong, just like everyone else’s models. The good news is that it doesn’t really matter, as IRR is, for most banks, Much Ado About Nothing.
One piece of advice, which I intend to repeat often, is to set the IRR tone carefully with your Examiner in Charge at the outset of your next exam. “We are a noncomplex institution” is a good way to start the conversation, and be prepared to explain why. Complexity = Risk, and we don’t want to look risky if we aren’t.
