In my previous post I discussed the impact your core deposit repricing assumptions (i.e. beta factors) can have on your IRR profile. Often bank management is surprised to learn that small and simple changes to the betas can change their bank from an asset sensitive profile to a liability sensitive profile very quickly. Hopefully your bank’s ALCO and Board have had some discussion about the beta factors used in your model. It’s a critical assumption that all bank management should be familiar with.
Having been part of countless ALCO and Board meetings over the years, I know there’s one common question that’s always asked when the discussion turns to beta factors. “What beta factors are other banks using?” Some modeling purists discount such questions because, technically speaking, it doesn’t matter what beta factors other banks are using. We’re modeling your bank – what matters is your bank’s anticipated behavior – not the simple setting in another bank’s model. The regs related to modeling always contain phrases like, “..will depend on the size and complexity of the bank…”, or “should be commensurate with the size and complexity of the institution…”, etc., suggesting that what other bank’s are doing is not necessarily relevant.
But dismissing such questions as irrelevant ignores a deep seeded human desire to compare. As one writer puts it, “making comparisons is often how we gauge our progress. It’s how we figure out the bar in the first place.” Is the beta factor you’ve chosen “conservative”, “normal”, or “aggressive”? Such judgments are good discussion topics for ALCO, Board, and pricing meetings, but they’re almost impossible to make without peer data.
Here’s beta factor peer data that should help during your discussions. We regularly run an IRR analysis for approximately 200 community banks. Each quarter they individually provide us with beta factors to use in their specific model. This snapshot was taken from analyses run for December 31, 2012.
For interest checking accounts the median beta factor used was 10% (the mean was 16.02%). The typical range (i.e. the middle fifty percent from the 25th percentile to the 75th percentile) was from 3.25% up to 25.00%. Only two banks in the sample set their interest checking beta to 100%. A little less than one-quarter of the banks in the sample set their interest checking beta to 0%.
For savings accounts the median beta used was also 10% (the mean was 17.68%). The typical range starts at a low of 5.00% and ends as high as 25.00%. Again there were two banks in the sample with a savings beta of 100% and a few shy of one-in-four set their savings beta to 0%.
Not surprisingly the data for money market accounts looks a little different. These accounts tend to be more sensitive to changes in rates therefore banks don’t model as much lag (so the betas tend to be higher.) The median money market beta is 50.00% (the mean was 48.19%) with the typical settings ranging from 25.00 to 73.75%. One in every five banks set their money market beta to 100% and only two banks set their money market beta to 0%.
Again, I suppose what other banks have set for their beta factors technically doesn’t have any bearing on your own settings. However I think the value of this peer data is two-fold:
1) At least it tells you if you’re totally out-in-left-field or not.
This is an important benefit. Like it or not there are still many people that don’t understand modeling (especially IRR modeling) terminology. When you start talking beta factors they tune out fast - we’ve all seen the glazed-over looks. Someone like this can still be part of the discussion even without intimate knowledge of the beta factor. Knowing that it’s “high” or “low” has at least some meaning and again it’s a great discussion starter.
2) It demonstrates that banks are not all just using the same parameters
There is indeed variation from bank to bank. This tells me at the very least they are not all just using some default parameter. That doesn’t mean they are using the “correct” beta for their institution (whatever that means), but from bank to bank we do see different settings.