Past, Present, and the Future with Predictive Analytics
Almost 30 years ago, George Darling published a short book titled, The Business of Banking for Bank Directors. In the opening paragraphs, George compares the business of banking to a manufacturing business, stating the following: “Products that produce revenue for a bank (or credit union) are assets, and the raw materials used to create assets are liabilities.” This statement may seem obvious to most bankers today; however, this was one of the first references in the industry that compared the business of banking to that of a manufacturing business. It also uniquely referenced deposits as a “cost of goods sold” instead of as a product.
Today, the banking industry has experienced the equivalent of five years’ worth of deposit growth in a 16-month time period, coined by some as the “COVID Deposit Surge.” Our “factories” are flush with raw materials in the form of higher levels of DDA, NOW, and Money Market accounts, in a manner that most of us have never experienced. Of course, the cost of these raw materials is also at a historical low given the interest rate levels. Regardless of the “discount,” raw materials must be transformed into a productive earnings stream in the form of loans and investments to maintain higher levels of net interest income.
Unfortunately, most institutions are dealing with competitive pressures in the loan portfolio, as existing prepayments and maturities outstrip loan demand. In fact, as exemplified in Figure 1, loan to deposit ratios are well below their long-term averages at 75%.
At the same time, as shown in Figure 2, cash and short-term investments are almost twice as high as their 10-year average. This dynamic continues to put pressure on net interest income levels as we head into 2022 and demands further focus on these levels in the future.
Analytics to Predict Deposit Behavior
Many institutions are concerned with the stability of the COVID Deposit Surge, which is clearly the predominant driver of the growth in cash and short-term investments.
In response to this unique industry dynamic, DCG has created a powerful predictive deposit forecasting analytics model.
In many ways, these predictive analytics can be regarded as an advanced raw materials (inventory) management system. The deposit forecast goes beyond simple averages or historical patterns–these predictive analytics leverage over two billion individual account records and 20 years of history, covering 5,000 branches in 48 states.
This robust and powerful tool encapsulates a series of linear regressions that utilize this data, coupled with carefully observed patterns, considerable industry knowledge, and specific rate scenario forecasts. These deposit behavior patterns are transformed into key analytics in the form of predictive product average lives (or balance behavior patterns) and effective beta (or potential rate sensitivity).
Why Predictive Analytics?
As management teams begin planning for 2022 and potentially the next economic cycle, most would agree that effective betas or deposit rate sensitivity might differ from one band of market rate movements to another.
For example, the first 100 basis points of rate movements of a money market account will behave differently than the next 100 basis points. In addition, most bankers would agree that the higher the rate movement of a deposit account, the more muted the decay or balance runoff. These are foundational elements of an effective predictive deposit model. In addition, qualitatively, most would agree that a deposit’s type, age, size, and price (as compared to market rates) are fundamental basics in understanding how that deposit may behave now and in the future. With predictive analytics, we can more effectively forecast a deposit balance and rate behavior patterns.
How Can Predictive Analytics Help?
Charles Babbage, who innovated the concept of the digital programmable computer, is known to have said, “Errors using inadequate data are much less than those using no data at all.” This is the essence of the deposit forecasting model as a competitive advantage. The predictive analytics tool enables bankers to articulate answers to the following questions:
How much of the deposit growth that you have experienced is due to the COVID Deposit Surge?
Given market and competitive dynamics, how might your average lives change as market rates fluctuate?
How much do your deposit rates lag as rates increase?
How do you group (and model) your high stability, low rate sensitive accounts from your low stability, high rate sensitive accounts?
How elastic is your deposit base and how much outflow in deposits can you expect from a 10-basis-point move in either direction?
In addition, our model equips and empowers balance sheet managers who seek to triangulate potential risks within their deposit base by examining the vintage of the deposit, the type of deposit, the number of relationships with the customer, and even the rate band.
Predictive Analytics Insights
Recently, we have utilized our predictive analytics model to forecast product average lives and deposit betas for our entire client base. We have uncovered many points of interest from our analysis, including:
Understanding how the shape and slope of the yield curve (and not just the short end of the curve) impacts pricing and competition (especially the five-year point on the curve).
Deposit size can vary–and it has an impact. Institutions can have a wide range of rate sensitivities even in the largest balance accounts. For example, the $100K-$250K tiers do not behave the same way as $250K-$1M tiers. Most banks’ tier structures, however, do not distinguish between these types of large balances (even though they should).
Account age impacts decay more for interest-bearing accounts than for non-interest-bearing accounts, reflecting customer preference for convenience.
As we perform this same analysis for individual institutions, we see additional trends. For example, some institutions can hold deposit rate levels lower for longer compared to competing institutions and the larger industry given their customer base and account structure.
In The Business of Banking for Bank Directors, George also asks a question that remains valid and pivotal to this day: “How is the bank’s liquidity portfolio invested and is it earning a reasonable return?” That question may be harder to answer today than perhaps ever before with the average life of the non-maturity deposit base in question, an ultra-competitive lending environment, and tighter spreads across most sectors in the investment portfolio.
Understanding the stability of the deposit base is paramount for operating effectively in today’s environment given today’s lower asset rate levels and continued margin pressures. The adept combination of data, analytics, and disciplined action can help us answer difficult questions and stay ahead of the competition in what is sure to be a challenging 2022.
Learn more about predictive analytics with DCG's data-driven software solutions.
ABOUT THE AUTHOR
Justin Bakst is a Managing Director at Darling Consulting Group. Justin provides risk management education and strategic consultation to financial institutions leveraging DCG’s analytics solutions. Justin has been a thought leader in risk management including interest risk, credit risk, and liquidity risk. Justin is a frequent speaker and author with the RMA, American Banker, NYTimes, and CIOMagazine.
© 2021 Darling Consulting Group, Inc.