Pub. 9 2018 Issue 3

www.wvbankers.org 28 West Virginia Banker T o many, it is a never-ending source of frustration that the exercise of projecting earnings in the context of interest rate risk mode- ling is such a messy process. Wouldn’t it be nice if all the variables involved weren’t so variable? Wouldn’t it be nice if all the uncertainties about rates, markets, competition, and customers’ behavior could be less uncertain? And then, wouldn’t it be nice to be able to know, with unflagging confidence, that the reports produced by your efforts described an outcome from which reali- ty would not deviate? Yes, all these things would be nice, but, unfortunately, none of them are likely to happen. The process is far from neat and tidy, and it is unlikely that “down- to-the-penny” precision will ever be achieved. In fact, about the only thing anyone is close to knowing for sure is that whatever the projections say, they’re probably going to be wrong. Hopefully, not too wrong. Following the Federal Reserve’s near-ze- ro rate policy that began in December 2008, the target rate for Fed funds didn’t change again for seven years, and it was another year after that be- fore it changed again. The same is true for the prime rate. To be sure, credit markets had their ups and downs over that period, but for the big variables that affect banks’ interest income and interest expense, that’s about as close to “neat and tidy” as interest rate risk modeling is ever going to get. The assumptions that govern the repric- ing behavior of assets and liabilities in changing rate environments never really got put to the test because rates never really changed. It’s hardly been a surprise, then, that for most banks, projections over that time period were pretty spot-on, and the results of back-testing exercises confirmed those results. They should have; it was just the past happening over and over. Climate Change The super-low and static post-crisis rate environment has now undergone more than a handful of increases, and the probability is high that there will be more. For the first time in a long time, risk managers this year will be comparing projections of rising-rate outcomes to actual, higher-rate results. How narrow or wide the variances might be will largely be determined by how close one’s modeling assumptions were to real-life behavior. Chances are, pro- jections will be missed, and the misses could potentially be pretty big. No doubt some of these misses will lead to consternation. While there might be times when a panicked response is appropriate, getting back a bad back-test should not be one of them. Remembering that the whole reason behind the concept is to determine the validity and suitability of your assumptions, delving into the reasons why results didn’t line-up with projected estimates is really just a learn- ing opportunity. Did the extrapolation of historical norms ignore demographic changes in one’s customer base? Did the competitive landscape change since that last deposit study was performed? Did Walmart move into, or out of, town? The reasons why things may not play out as expected are many and varied, and it doesn’t necessarily follow that there’s anything “wrong” with your pro- cesses. The problem lies in the assump- tive inputs, and this provides a golden opportunity to see where, and how, adjustments can be made so that reality can better be replicated, and projected, in interest rate risk reports. Inexact Science Redux A common misconception is that the back-testing process itself represents foolproof methodology that then produces a “right” answer for what was missed and how big was the miss. Not so much. One of the trickier aspects of back-testing is backing out, from actual results, the effect of the growth that took place over the period being back-tested. That’s a necessary step so that the growth that occurs over the pe- riod doesn’t misleadingly hide or falsely exacerbate the risk measurements of the model. Undoing a year’s worth of growth and rate changes in order to get back to an apples-to-apples compari- son is a bit dodgy and fraught with op- portunities for error. Neither growth nor rate changes occur in a linear fashion, so apart from volume changes, the tim- ing of the volume changes that are then woven among rate changes represents another significant variable providing yet another opportunity to be wrong. Feel better yet? Well, you should. All the fine-tuning in the world will never guarantee that projections will match outcomes, and the use of one imperfect process to validate another imperfect process should be reminder enough that asset/ liability management is still more art than science.  Lester Murray joined The Baker Group in 1986 and is an Associate Partner within the firm’s Financial Strategies Group. He helps community financial institutions develop and implement investment and interest rate risk management strategies. Before joining The Baker Group, he worked at two broker/dealer banks in Oklahoma City and was also an assistant national bank examiner. A graduate of Oklahoma State University, he holds Bachelor of Science degrees in finance and economics. Contact: 800-937-2257, lester@ GoBaker.com. Asset/Liability Management Is Not a “Perfect” Game By Lester Murray