If you are looking for insight on how artificial intelligence can help banking, we give you the Unified Deposit Formula. Prior to using machine learning, we, like most bankers, thought about deposit pricing along a single dimension – price and sensitivity. However, it turns out, that price is not only just part of the equation but often a small part. There are thousands of other factors that play a part. Turn artificial intelligence loose on a data set, and you come up with a formula that allows you to optimize your inputs to achieve your desired level of new deposit balances. In this article, we provide an overview of the equation and highlight some of our lessons learned.
The Unified Deposit Formula
One of the strengths or “unifying” features of the formula is that it ties price, customers, marketing, the asset-liability objectives, deposits, loans and balance sheet profitability together in a nice, neat package. We will start with a simplified version of the formula below.
Each colored input further breaks down into its own separate formula, which we plan to elaborate on in the future. However, it is important to understand the concepts of this formula in order to be effective at deposit gathering.
The other major advantage of this equation outside of its unification properties is that it forces bankers to look at their deposit-gathering effort through a wide-angle lens as opposed to a microscope centered on one product such as certificates of deposits and one input such as price. This is the primary shortcoming of current deposit models in that they treat price as the only input. While national banks are getting more sophisticated, many regional banks still think in terms of price which gives community banks a huge advantage.
As we will soon show, price is often the least important input.
Change in Needed Deposit Balances
The formula starts with asking - given your existing core deposit base, what is the new amount desired? In other words, this is recognition that you can control your deposit growth by controlling your asset growth. Grow loans at an 11% rate, and chances are you will need more deposits and have to expend more effort to raise those deposits.
Later in the week we will expand on this factor and provide bankers with another formula that allows you to size your efforts and marketing budget. For now, let’s break down the rest of the equation.
Predicted Market Penetration at Optimized Offering
The middle portion of the formula is where the magic happens. Here, data and current computer models can lend a material level of assistance to the community bank. However, while the math is important, more important is the understanding of the concepts.
The Predicted Penetration at Optimized Offering answers the question of what amount of deposits you can expect from the any given market. This is a percentage, the level of which depends on your market, your customer base and your products. This calculation can be done for one product such as a money market account or for a complete line up of products.
It is within this function that the Customer Base Elasticity Demand is calculated. Knowing your elasticities for any given product and customer segment allows bankers to optimize value. This is where community banks can excel.
Customer Base Elasticity of Demand
Within the Predicted Market Penetration at Optimized Offering function is an elasticity factor that most bankers are familiar with a notable exception. In banking schools, most bankers were taught to focus on price as your controlling factor to sensitivity. As the data and machine learning has taught us, price is only your most sensitive input in one part of the demand curve. We will break this down in the future, but there are whole sections of the deposit demand curve where price registers the least sensitive response. Product design, fees, flexibility, ATM access and many other factors all often provide bankers with more sensitivity. Thus, this elasticity function looks like the equation below.
This is the part of the equation that captures both product effectiveness and customer segmentation. Choose the right products targeted at the right customer segment, and you produce an elasticity well in excess of “1” which means you will either be more effective at raising deposits or have reduced marketing costs (the next part of the equation).
The heart of this customer base elasticity function is pumped by the perceived value proposition of your offer. Sometimes this is rate, but often it is another factor. Machine learning was instrumental in helping us understand where, when and how to apply various input to elicit the best demand response.
Adding lock box for homeowners associations (HOAs) or trade associations, for example, dramatically increases the probability of winning those deposit rich customers. For a relatively small investment, the percentage change of the deposit volume demanded can go way up. Waiving ATM fees, adding IT support services or hundreds of other ideas can create this type of response depending on your product and customer segment.
The goal here is to get creative and start to try a variety of offers to see what product characteristics resonate with different customer sets. By identifying the customer set first, be it college students, retirees, or breweries, banks can better develop product bundles that resonate. This makes understanding the Market and Amplification part of the equation easier.
Later this week, we will run Part II, where we will quantify the value of marketing and highlight the Market & Amplification part of the Unified Deposit Formula. We will also provide a formula to help banks size their addressable market, gauge the effectiveness of their campaigns and to manage this effort over time.
Next week, look for Part III, which will delve more into the concept of the Predicted Market Penetration at Optimized Offering where we will quantify why rate should be your deposit attribute of last resort.
The Unified Deposit Formula ties a variety of bank functions together and allows a single methodology to be used throughout the bank. It works for a single product or for bundled products. While we are presenting this formula for retail and commercial deposits, it also works for loans, and fee products.
Instead of just price, the key takeaway from this formula is always to be thinking of what attribute, other than price, can you add to a deposit product for a specific customer segment that will get you the most “lift” or new deposit balances.
Until Part II, give some thought about how your bank can increase your marketing penetration by using attributes other than price to attack profitable customer niches with modified deposit products that will garner a better demand response.
Submitted by Chris Nichols on August 27, 2018