Banks are highly complex to manage. In particular measuring profitability, understanding customer relationships, setting goals and tracking performance, managing consumer and SME delivery networks… The list goes on.
Many often talk about the need for analytics in banking. Recent trends are on deposit pricing and there is an increased focus on operational risk. But which are the core/foundation “must have” building blocks on which management can build strategies and tactics to increase shareholder value: ROE and revenue growth?
1. Profitability model
Calculating profitability in banking is hard. Firstly, because half our products are deposits where the concept of profit is counterintuitive since the bank actually pays interest. Secondly, because profit is not realized nor recognized at the point and time of sale as would be the case with most other industries. Instead profit is accumulated on a daily basis over the often indeterminate duration of a product. Lastly, the inevitable complications of householding—discussed below—makes the choice of the right level of abstraction an oftentimes heated one. There are quite a few pitfalls, most of which fall under the umbrella of “perfect is the enemy of good”. Specifically, inside many banks there are often long debates about the accuracy of the complex calculations required to perform profitability analyses. These debates usually center around Funds Transfer Pricing (FTP) curves, duration of deposit portfolios, cost allocations, as well as householding and domiciling choices which are discussed later on this white paper. Our general advice is to “aim high but start low”. The reality is that for purposes of managing profitability inside your retail network, a basic profitability model will work just fine. Arguably, a basic—and by definition simpler—model is easier to explain and understand among those involved. In many cases it even makes sense to simplify complex components (e.g., a bank’s Treasury might use a highly complex FTP model that might however be suboptimal for use in retail management).
Why is having account level profitability measurement important to manage a retail bank?Because it allows you to make informed management decisions based on fact and not perception. A profitability model will definitively answer questions like the following (partial list). These answers allow for ROA and ROE maximizing strategic and tactical decisions.
For a branch:
How profitable is our branch on 243 Main Street?
Is it above hurdle rate?
What levers should we move to make it profitable?
For an RM or other banker with a customer portfolio:
How profitable is this banker’s portfolio relative to his or her peers?
Which customer relationships are the most profitable that need to be protected the most
Which relationships are less profitable in the portfolio, and what is driving that relative to the most profitable ones (e.g., pricing, product mix, etc.)?
For a customer household:
How profitable is this household?
What product cross-sell, or balance augmentation, or pricing adjustment would “fix” any less profitable or unprofitable situations?
Given the profitability profile, is this household over-served or under-served? E.g., should we relationship-manage a given household based on high profitability even if it has not hit the formulaic triggers on balances?
For a product:
How profitable is a product?
Do we need to adjust pricing or the fee structure?
Do we need to adjust other costs associated with it?
Given its profitability, are we properly calibrating our front-line incentives, our goals, our marketing investment?
Lastly there is the question of platform on which such a model should operate on. There are of course off the shelve models that can be adopted. There are also less robust but possibly acceptable options with easier integration and lower cost to operate. Many banks choose to deploy these at first. Deep down, a bank-wide profitability infrastructure is a management mindset issue and not one related to cost or technology. The latter two can be addressed within a budget cycle; the former is a senior management choice.
2. Householding algorithm
How do you react to the information from a profitability model that shows that an account is unprofitable? Do you try an aggressive “fix or exit” strategy? The answer should be different depending on who owns that account. If it belongs to a single-account household, then maybe; but what if it belongs to a household that has other highly profitable accounts? A “fix or exit” approach could risk losing the entire relationship. It is imperative to know which household each account belongs to. It allows proper management of the relationship. It also enables wallet analyses and the identification of cross-sell opportunities.
Householding choices can be complex. The easiest—and minimum—is to simply household all accounts that are under the same tax ID number, whether consumer SSN or business EIN. But such an approach does not recognize a customer’s spouse, children, etc. It does not recognize subsidiaries of businesses with different EINs. It certainly does not recognize the linkage between a business and the consumer accounts of the owner(s).
Much like the discussion on profitability, the key here is senior management engagement in exploring the trade-offs and establishing the right rules for the bank. These choices can quickly become more complex than they appear. For example, how do you handle a medical practice with 3 doctors who also have individual accounts and who are each married to other business owners who also have personal and business accounts with the bank?
A good development process can help establish the right algorithms to capture all these management choices and automatically create households. Once implemented, good householding programs also encourage the organization to ask the right questions during customer interactions and to fill gaps of the bank’s understanding of its customers.
3. Domiciling algorithm
Because banking is an inventory and not a flow business, the question of where an account (or better yet a household) belongs is critical for attributing profitability to the correct distribution node (e.g., branch, RM, private banker, etc.). There are many approaches, and to use a cliché expression, “there is no right or wrong answer”. What is important is for management to make the right decisions for their bank and to be consistent over the years so the front line can manage their respective portfolios with focus.
4. Goal setting model
Goals (and the related incentives) are often set based on past performance. This is a mistake. It rewards past underperformance with a low easy-to-reach goal and penalizes past strong performance with a high, possibly unattainable, goal. The solution is a goal setting model that is established following a “bottom-up” approach that considers the characteristics of the branch, its market, and the local competition. Once again models can be developed that can incorporate all these parameters and develop goals. For example, the quest for deposits tends to be a localized “zero-sum” game. Studying the deposits in the competitor branches surrounding your branch, and making assumptions on the various dynamics (e.g., expected competitor attrition) allows for the development of a game theory-based model that can help build bottom-up goals.
5. Customer wallet penetration “sniffer”
The holy grail in “growth” marketing has always been cross-sell. A big obstacle is the lack of visibility into the customer’s wallet. Once a bank has established its householding algorithms, this becomes a lot easier. The increase in electronic payments is also making it easier, especially for households where the bank owns the main checking or operating account, for a bank to develop algorithms to develop visibility into the contents of the wallet. A credit card or mortgage payment from a consumer checking account, or a transfer to a payroll company from a business operating account, reveal obvious cross-sell opportunities. Even rudimentary algorithms can present a rich cache of specific opportunities to the front line or to a centralized marketing area.
6. Branch analytics on De Novo, Mature and Steady, Consolidations, etc.
How well is a branch performing? Profitability models (see above) can answer that question in an absolute way. Goal setting models (also see above) can answer relative to the competition and relative to the market. However, these questions become more complex when we are looking at younger branches that have not yet reached maturity and are still unprofitable. Are they on their way to profitability? Is their development curve satisfactory? On the opposite end of the branch lifecycle spectrum, when branches are closed and consolidated to nearby locations, how successful was the retention performance? There are models that have studied these patterns on a national level and over time and which can offer reliable benchmarks for analyzing branch performances even in less usual situations. They can also inform choices to select de novo locations or to make (or refrain from) consolidations decisions.
There are undoubtedly some bankers that after reading this white paper will say “We have all these in place for years and they are working fine”. Others will say “We do not have any of these; where should we start? Should we even bother?” Most of you are probably between those two extremes. The questions you are most likely asking are “Which of the models or algorithms we are missing should we prioritize to build?”, “What is the smart way to introduce them without embarking on a mega-project that might cost money and time without delivering?”, “Could we integrate this analytical infrastructure into our culture? If not, should we build these capabilities?”
These—and many others—are all the right questions for a management team to ask. These are good discussions to have internally and occasionally with outside advisors. At Delos Advisors we obsess about analytics and customer-centricity. We would be delighted to partner with a management team that wishes to further explore the power of analytics in increasing shareholder value.