Like clockwork, in times of low interest rates, financial institution executives have questions regarding the changes in their market value results. “What happened to my premium?” they ask. Invariably the answer lies in a little discussed, but oh-so-powerful assumption – the cost of servicing assumption.
Cost adjustments: Wrong assumptions can depress your FI’s market value premium
- Financial institution executives wondering about changes in their market value premiums might need to look at the assumption used for modeling deposit servicing costs.
- Some ALM models rely on cost of servicing assumptions that haven't been updated in years.
- Ideally, have your own cost of servicing data, but if not, here's a basic formula that might generate more meaningful discussion with your board and ALCO.
Faulty assumptions for servicing costs often behind changes
When we embark on a discussion surrounding costs, we inevitably end up back at the question of what costs are reasonable to apply to deposit accounts.
The thought that a simple input could be such a large contributor to value is difficult to believe, as we have been led to believe non-maturity accounts are low cost, long duration funds. How can an account with a long duration and slow repricing beta see market values drop when nothing seemingly has changed?
To understand this phenomenon, we need to understand how these values are calculated.
Calculating the market value of any cash flow requires creating an amortization schedule of projected principle and interest cash flows. For accounts like loans or term instruments, this projection is very simple as the contract dictates the repayment. Of course, there may be some “options” like prepayments or other modifiers that might change the schedule. But fundamentally, the flows are simple. However, non-maturity deposits have no contractual flows to model. That’s the purpose of the decay rates. Decay rates, which estimate how quickly non-maturity deposits will run off, or flow out of the institution, define the speed of principle reduction -- like a loan amortization schedule. With a principle schedule, we can then apply the expected interest rate for the scenario being modeled to get the all-in P&I cash flows.
Therefore, accurate assessments of the decay rates, including any accounting for funds that might be shorter duration (surge deposits), is a critical factor to defining the life -- and interest cost -- of deposit accounts.
Some ALM models use outdated cost estimates.
Speaking of costs, it is common to see an Economic Value of Equity/Net Value of Equity (EVE/NEV) model assign a cost of deposits to each category of accounts/products. Historically these cost assumptions were from sources that are not easy to find or assess how they were arrived at. For many years, and in many databases today, the source was the Federal Reserve Functional Cost Analysis data that ended in 1991. Of course, that doesn’t make much sense given how much has happened since 1991 in terms of banking delivery costs. Internet and mobile banking are commonplace, unit costs on deposit processing have dropped, electronic statements are the norm, more transactions run through debit cards than paper checks, etc. Why not just keep using the costs from a couple decades ago? I remember that last study well because it included a great big disclaimer indicating that the values were estimates and were NOT to be used when pricing products. Why not? Because the estimates were based on a model that fully distributed all operating costs to a product. The allocation rules were very generic and arbitrary. And besides, it assumed that if you cut a product from your offering lineup, the costs assigned would go away as well. In reality, the only costs that would go away for sure are the direct costs associated with the delivery of the product. The CEO still gets paid. The marketing budget just shifts focus. You get the idea. It’s extremely difficult to assign the cost of these accounts only.
To illustrate the point, below you will find the last Federal Reserve Functional Cost Analysis (FFCA) published costs by products. Note: Much like decay and prepayment rates, these costs are the default costs used in ALM firms for years when institution data is not present.
Servicing costs add to the overall cost of the account like interest expense, raising the all-in cost of the deposit in the market value cash flows.
To arrive at the actual value of the account, we discount each cash flow by a market rate, usually the comparable term FHLB advance rate. When the advance rate is greater than the all-in cost, the deposit has “value” to the institution. When the FHLB rate is lower than the all-in cost, the deposit no longer has value, as the cost of borrowing is cheaper than holding deposit accounts.
In today’s market, the FHLB advance curve shows rates below 1% through 10 years, and between 1.0% - 2.2% on the “long-end” of the curve. Even with low overall interest costs on deposits, adding 86 to 257 basis points quickly takes all deposits values down!
So capturing costs involved with deposit accounts is reasonable enough, and using costs in the calculation of market values is fair. But for many financial institutions, the values we have at our disposal are not reliable or able to be documented. Additionally, the cost assumption should also capture the revenue from non-sufficient funds (NSF) fees, service fees, etc., to accurately reflect the net incremental cost of these accounts.
Your own cost-study data or try to estimate
Of course, having your own cost study data based on your institution is the best to use for these costs. Some of our clients have done so and have better and defensible data to use for their analysis. But if an internal study isn’t workable, it’s helpful to use a basic formula that might generate more meaningful discussion with your board and ALCO (Asset/Liability Committee).
Borrowing from “Managing and Pricing Deposit Services, Rose & Hudgins, 9th ed., Chapter 12”, we can use a formula that considers costs based on the average collected balance for each account, where average collected balance = total balance minus funds that have not been collected yet by your institution.
The basic formula is a ratio of
net expenses divided by average collected balance * (1 – reserve requirement for that deposit type)
[Interest Expense + Admin Costs – Service Charges – Other non-interest income + deposit insurance premium]
Interest expense is easy to deal with by assigning the current rate and a beta factor for moving that rate when market interest rates move.
Administrative costs: This list include all kinds of “services” provided to the holders of this account type. Costs for handling withdrawals and deposits (electronic and non-electronic) per item, check cashing costs, opening/closing new accounts, monthly account maintenance costs, etc. Getting to a “per item” cost requires time allocations by staff members and average pay rates. Then, apply that to the overall portfolio.
Service charges & other non-interest income: Listing by source of income the average per unit earnings levels.
Note: Realistically, you may want to estimate the costs and income levels to arrive at a net cost value. Document the thoughts and discuss these with the board and ALCO team for acceptance and acknowledgement. Cost study data is an expensive and time-consuming activity. As an alternative to get started, consider this approach. If you were to grow the balance and accounts by 50%, how much more staff time would be required to manage the cost? How much more non-interest income do you expect to collect? From that estimate we can back into the costs to apply as any new service may artificially have high costs until there is enough scale to lower unit costs.
Deposit insurance premium: This is a real cost to the institution. Using your regulatory requirements for insurance premiums, assign the incremental cost per $1 to the cost equation.
For the denominator, the current level of reserve requirements for banks is 10% of the demand deposit and checking account balances. For all other accounts, the requirement is 0%.
Let’s apply this to an example.
Example: Interest bearing checking
- $2,000 average account balance
- 60 basis points deposit insurance premiums or $12.00 annually
- Interest expense (0.25%) is $5.00 annually
- Administrative costs estimated at $38 per year
- Non-interest income is $50 per year
The numerator formula looks like this: [5.0 + 38.00 - $50.00 + 12.00] = $5.00
The denominator looks like this: [$2,000 * (1 - 0.10)] = $1,800
The result: $5.0 / $1,800 = 0.00278 or .278% or 28 basis points
This result is 85% lower than the FFCA of 1.80% (180 basis points or 0.018), and you see that the internal calculations are usually a LONG way off from the old FFCA values. Based on an $1,800 denominator, the net numerator would have to be $32.40 not our $5.00. There’s a lot of room in between.
What would change the equation to reach that level?
- Same numerator with a $277.78 net average balances (denominator).
- Higher interest rates (numerator)
- Higher operational cost assumptions (numerator)
- Lower fee income per account (numerator)
Let’s be respectful; the old FFCA values were very good in their time. The industry needed something, and the data was sourced from banks directly. Now, if the inputs were accurate or not, we cannot ensure, but at face value, they served as a guide to measure against. But in 1991, the level of non-interest income was much lower than today. A tremendous amount of hand processing, like sending statements with cancelled checks, was the norm. But today’s world has “bent the curve” on costs, allowing for more accounts and growth without the same level of overhead managing these functions.
So take a stab with what you know using the above formula and get to a more meaningful discussion about valuing growth and managing earnings and risk to sustain your future as a community financial institution.