Originally published 2/28/2007 © 2021 Olson Research Associates, Inc.
Disclaimer: This is part one of a two part post. It is not meant to be a primer on the gap report or decay. Here are two links that will give you some better background on what gap is (Insights for Bank Directors from the St. Louis Fed), and what decay is (“What are core deposit decay assumptions? here's a quick answer...”). I’m assuming you have some working knowledge of both.
One of the most misunderstood areas of asset/liability modeling is the treatment of core deposits. It’s not that the concepts are particularly hard, it’s just that people tend to confuse different techniques, or assume that’s there is a one size fits-all-approach. This is especially true when it comes modeling core deposit behavior for an earnings simulation versus modeling core deposit behavior for an EVE (economic value of equity) simulation. Before I dig into this, let me first restate my dislike for the gap report as an interest rate risk measurement tool. I’ve posted a few times on this:
OK, now that I’ve said that, let’s talk about gap versus decay. What you have to remember is that gap analysis is a tool that attempts to measure earnings sensitivity. A typical gap report shows the maturity and repricing schedules for all of the earning assets and interest-bearing liabilities at a bank. Comparing the value of assets that mature or reprice at each point in time with the value of the liabilities that mature or reprice may reveal the exposure of earnings to changes in interest rates.
The bank's gap measure is constructed by summing the dollar value of assets that come due over a given time interval and comparing it to the liabilities that come due during the same interval. The cash-flows represent potential dollars of earning assets and interest-bearing liabilities that mature, reprice, amortize, prepay, call, and/or (in the case of deposits) withdrawal early. When using gap analysis it is common practice to label all such cash-flows "repricings" because the dollars in a given bucket are assumed to be re-invested or repriced to current market rates.
When a bank has more assets repricing than it does liabilities within one year, it is considered “asset sensitive”. Generally speaking asset sensitive banks will benefit from rising rates and be negatively impacted by falling rates. When a bank has more liabilities repricing than it does assets within one year, it is considered “liability sensitive”. Generally speaking liability sensitive banks will benefit from falling rates and be negatively impacted by rising rates.
One big hurdle in gap analysis is reporting core deposit balances (I’m referring to NOW accounts, Savings accounts, and Money Market accounts). Where do you slot the dollars? In many cases the rates on these accounts are administered rates i.e. established by bank management. In some cases (most likely money market accounts) the rate paid is tied to some market index.
For accounts that are tied to an index, there isn’t typically much of a problem deciding how to gap the balances. Since the rate paid is tied to a market index, the entire balance is typically slotted in the first “less than 3 month” bucket.
For some accounts that have administered rates, it may be easy to decide how to gap the balances. Take for example a statement savings account that has always paid 35 basis points. Since the day the bank opened it’s doors, it has paid 0.35%, and as far as anyone can tell, it will always pay 0.35%. In this case, savings deposits should be placed in a bucket beyond the one year time horizon. Because in gap analysis we’re typically looking at the one-year cumulative gap, we want to make sure these statement savings accounts don’t impact our sensitivity within one year.
For other accounts that have administered rates, the decision about gap is less clear. What about a NOW account where the bank adjusts the rates to follow changes in the market, but doesn’t always change as much as the market changes? What if a bank is currently paying 1.25%, and knows that if the Fed tightens by 25 basis points, the bank will raise it’s NOW account rates, but only up to 1.35%; if the Fed lowers rates by 25 basis points, the bank will adjust NOW account rates down to 1.15%. In either case the bank only changes by a certain percentage of the change in Fed Funds; 10 bp change for every 25 bp change in Fed Funds. 10 bp divided by 25 bp equals a 40% change. In modeling terms this 40% is typically called the “core deposit repricing factor” or “core factor”.
The core factor is a critical part of the earnings simulation, but how is it represented on a gap report? These NOW accounts do reprice up or down when rates change, and let’s say (to keep things simple here) that the entire balance prices up to 1.35% or down to 1.15%. If you follow the traditional rules for a gap report, you would have to say that the whole balance get slotted in the first bucket, just like the money market accounts. But does that really make sense? Probably not.
To understand why, think about a prime-based variable rate loan on the asset-side. Of course it’s likely to get slotted in the first bucket, because if prime rate changes, this loan will reprice immediately. It will reprice to market rates. That’s the classic assumption of any cash-flow on a gap report, it will reprice to market rates.
So here’s the question: Is it reasonable then to slot the entire NOW account balance in the first bucket to match-off with this loan(s)? After all the rate on the NOW accounts are not going to move to market rates, their rate is going to move only partially.
The way we solve this problem in our model is to slot a portion of the balance in the first bucket, and a portion of the balance beyond the one-year bucket. We split the balance using the core factor. In the case of these NOW accounts we take 40% of the balance and put it in the “less than 3 months” bucket, and the remaining 60% we put beyond one-year in the “1 to 3 year” bucket. This is how we represent that the rate on NOW accounts will only move a portion of the change in market rates.
Actually, if you look at the example, we’ve used the core factor to slot the money market and statement savings balances too. The money market accounts that are tied to an index have a 100% core factor; therefore we slot 100% of the balance in the first bucket. The rate on statement savings accounts doesn’t move at all, so it has a 0% core factor; therefore we slot 0% of the balance in the first bucket, and the remaining portion (or 100%) beyond one-year.
So we tie our gap report treatment of core deposits to how these deposits behave in our earnings simulation. We use the account’s core factor to split the balance into two buckets: in the short bucket, “less than 3 months”; and beyond one year in the “1 to 3 year” bucket. We use our earnings simulation behavior here because remember, gap analysis is a tool that attempts to measure earnings sensitivity.
In this first part, when explaining gap report behavior, I think I’ve successfully avoided using the words, “decay-gap”, or “decay buckets”, or “gap decay”. I’ve avoided this on purpose. Decay is a concept used for valuation of core deposits, NOT for income simulation. In my companion post I’ll talk about decay.