Are the banking industry's profitability systems profitable?

America is the world's leading apostle of the free market. Americans believe that if businesses don't succeed, it is because they aren't profitable and no one has an incentive to engage in them.

Perhaps it can be said, then, that banks' profitability systems are designed to spread the gospel. As the name implies, these systems automate the profit measurement process, either by product or by customer.

A number of such systems are being implemented in banks today. But conversations with bankers reveal serious doubts about the efficacy of such systems and a history of not relying on the concept of product profitability. The implication is that profitability systems have not proved profitable.

There are three major difficulties. The first is conceptual, revolving around how revenues and costs are defined and allocated. The second is technical. There are significant costs to construct such systems, and there are technical limitations.

The third, and perhaps most important, is in the policy realm. It is not easy to change customers or products that are labeled "unprofitable," and therefore often nothing is done as a result of the profitability analysis.

Profit has always been an elusive concept. Generations of MBAs have had it drilled into them that there is no one right way to measure profit and that many arbitrary assumptions are needed. Moreover, in many cases, such as businesses with heavy cash flow, profit is a largely irrelevant concept.

Banking has a lot of conceptual difficulties that affect the profitability analysis. First is the need for transfer pricing to measure the relative contribution of asset and liability products.

No banking product is complete in and of itself, because assets must have a funding cost and liabilities must have an investment return. The flow of funds mostly uncouples the assets and liabilities, and matching by duration is difficult, subject to change, and never totally achieved.

Next is the subject of credit risk. A bank cannot predict when a customer will default on its obligations - or the word "risk" wouldn't apply. When a specific customer does default, the profitability of that customer presumably changes.

The typical thought process is as follows: In a portfolio of 2,000 wholesale customers, we think each of them will be good, or we wouldn't get into the deal in the first place. Yet, we know that historically, a percentage of them default. Moreover, that percentage is subject to change; the past is no indication of future default rates.

So how can we, in advance, determine the profitability of an individual customer without allocating that risk across all customers - and thereby destroying the concept of customer-specific profitability?

Allocations are yet another problem. Operating costs in particular are notoriously impossible to allocate. Many existing models even ignore them. In banking, the level of shared costs is very high, such as branches, back-office operations, corporate overhead, and the information technology infrastructure itself.

How does one allocate Community Reinvestment Act costs, marketing costs, human resource department costs, mainframe upgrades, and so forth? When costs are triggered by transaction volumes, which are variable, how does one assign this to customers or products without tracking each transaction? The bottom line is that all profitability models must make many allocation assumptions, which may or may not reflect reality, which may or may not be politically based, and which may or may not be believable to the appropriate senior managers.

Profitability systems cannot be viewed as static reports that are essentially giant number-crunching exercises. The conceptual difficulties mentioned above demand a different approach, namely, to turn the profitability systems into more of an ad hoc decision-making aid.

This is done by making use of both relational data bases and client-server technology. The profitability system extracts data from the transaction and core systems and feeds them into a specially designated data base that uses a three-dimensional approach to increase the ability to see different views of the same data. These data are increasingly being placed on a server and the applications to massage the data generally run on a client. The main value is not so much the regularly scheduled reports, but rather the ability to perform "what if' and historical-type calculations.

These systems are generally being set up to measure both product and customer profitability. The two are intrinsically linked, and will become even more so in the future as bank products are increasingly tailored for specialized customer subsegments. In practice, however, the systems are much more likely to be used to analyze segments of the customer base rather than individual customers. And the models and assumptions that go into the retail models are going to be very different from those that go into the wholesale models.

And then there are the policy issues. A large commercial bank recently spent a year and several million dollars calculating the profitability of each of its three million retail customers. They concluded that about one-quarter of them were unprofitable, and were able to draw some obvious inferences about the causes, such as low balances, high transaction costs, or area of residence.

Despite this considerable effort and expense, the bank took no action. The unprofitable customers were not asked to take their business elsewhere, nor were they repriced. No strategic changes in products, channels, or marketing seemed to flow from the analysis. The cross-subsidy from the profitable three-quarters to the unprofitable one-quarter stayed in place.

It is interesting to note some of the reasons why policies or policy action might not automatically follow from profitability systems. First, individual products might be viewed as "strategic," that is, required for market or defensive reasons, even if unprofitable.

They may be viewed as future building blocks that will become profitable when more customers sign up, when that new system is built to reduce operating costs, or when market conditions change. Historical profitability may legitimately be viewed as having only a passing relationship to what the bank should do in the future.

Second, individual relationships might also be viewed as strategic, even if they too show up as unprofitable.

For example, a corporation based in one large bank's hometown stopped corporate borrowing years ago, but still gets cut-rate pricing on trust services, letters of credit, commercial paper backup, and so forth. The relationship is considered too valuable to jeopardize by repricing.

Other reasons for failure to act might include: unwillingness to reduce services - or charge more - in lower-income neighborhoods; a failure by the appropriate senior managers to "buy into" the validity of the analysis; or high costs associated with potential policy actions.

For example, certain branches or support services might be identified as unprofitable through an analysis of channel profitability. Yet, perhaps it would be even more costly to discontinue the branch or service than to leave it in place.

Conversion costs or impact on other products or customers can sometimes be viewed as more unacceptable than the status quo.

All of these issues cloud the long-term impact of profitability systems.

Despite technology in recent years that would seem to allow a more accurate and easier calculation of profitability, it is not clear that there are strong advantages from these investments.

There is an interest, and we can expect future investments by some, but probably not all, banks.

Profitability systems seem like a basic element of managing a bank in a rational way. Yet, the real value of these systems can only be measured by observing what the market does as a result of their use. If enough banks don't make the investment, or go into it half-heartedly, or don't take any actions as a result of the analysis, the conclusion can only be this: The benefits aren't worth the costs. Right now, there are banks on both sides of this issue.

For reprint and licensing requests for this article, click here.
MORE FROM AMERICAN BANKER