The prospect of rising interest rates and more competitive pricing is challenging one of the banking industry’s traditional ways to value deposits.
The longstanding practice of behavioral life measurement, in which banks analyze historical deposit product behavior to help determine how to invest non-maturity deposits, needs to shift to a forward-looking view. That is because observations made during the near-decade long flat interest-rate environment may not be representative of customer behaviors when rates rise.
The growing concern is that behavioral lives will shorten considerably as customers allocate balances to higher-return savings accounts, including those from the growing number of direct bank entrants. This is a reality across the deposit spectrum, including both consumer and commercial deposits.
The issue is important because behavioral-life assumptions are essential components of funds transfer pricing (FTP) methodologies, long used by Treasury functions to assign internal value and cost to sources and uses of funding, respectively. In doing so, FTP affects product and business profitability measurement and also drives pricing decisions, business planning and ultimately compensation.
Leading practitioners are working speedily to revisit the analytics that underpin behavioral life assumptions. Novantas estimates behavioral life assumptions for garden-variety interest-sensitive deposits such as MMDA and Savings are poised to shorten by 1.5 to 2.0 years on average, reducing FTP credit rates by 10bp to 30bp. Rate-sensitive balances without a relationship component are expected to react even more strongly.
Novantas believes banks that fail to respond are, at best, putting themselves at a competitive disadvantage and, at worst, are exposing the balance sheet to considerable interest-rate risk.
The most powerful advancement in behavioral life measurement is the use of statistical modeling and forward-looking scenario analysis to understand potential future behaviors. A common model framework involves using models to predict account attrition (i.e., customers closing their accounts) and average balance evolution (i.e., customers adding or removing funds from their accounts). This framework enables “what-if” scenario testing around the relationship between funding stability and pricing, market interest rates, and other macroeconomic factors. The right models allow banks to quantify the effect and send the correct pricing and profitability signals to the front line. In this way, banks are wisely shifting from backward-looking to forward-looking views of customer behavior, which for many U.S. banks began with CCAR stress testing exercises.
Banks must also decide how to measure the behavioral life of products with contractual characteristics that offer the customer “future optionality.” For example, certificates of deposit (CDs) provide customers the option of renewing into different terms and sometimes allow for rate renegotiation at maturity points. While balances might persist through these changes, some think the renewal should be viewed as a discontinuation of the account’s behavioral life. In this context, CDs with same-tenor renewals and no rate renegotiation may be treated one way, and CDs that renew from short to long tenors and/or renegotiate rates may be viewed differently.
Novantas believes banks should recognize balance persistence as a continuation of behavioral life, that reflects account characteristics changes for different segments of customers. This does lead to some complexity when banks consider term deposits alongside non-maturity deposits, given that there is substantial balance switching between CDs and NMDs that can be more difficult to track. Deciding how to treat these behaviors is critical as rates rise and customers’ product and term preferences change.
A second prerequisite for modeling is to segment deposits in a way that best explains variance in behavior across account and customer characteristics. Best practices expand the view from the industry standard analysis by line of business and product to include pricing level, balance size, depth of customer relationship, region, and industry for non-retail customers. The need for this is obvious in that different segments will react more- or less-strongly to changes in deposit prices in the marketplace.
Having obtained granular behavioral life analyses, it is obvious to use insights to guide business decisions. Longer term, behavioral life scores could be used for targeting customers, for example, but there are additional considerations in doing so. While the front line cannot change the fact that a customer is a retail or small business customer, they may be able to influence the funding balance of the account, or encourage the customer to also open a transaction account, thus steering a customer toward a more favorable FTP treatment for the business.
Such “internal arbitrage” could deteriorate the historical information content in the segment over time, and undermine the original behavioral life signals.
These advancements are part of a broader Treasury transformation which is driven by the need to manage the balance sheet in a forward-looking way. As banks embark on new strategies to fuel growth in a more vibrant period of market interest rates, leading practitioners must revisit historical behavioral life assumptions to send more accurate profitability signals. We believe that winners of the next cycle will be determined in large part by Treasury excellence in analytically-informed profitability signals.
Director, New York
Principal, New York