IT'S CLEAR that banks' products and services have gotten far better in recent years because of technology investments. But it's also clear that too many banks invest in capacity, often making the most automated lines of business only marginally profitable.
These issues were explored in detail in a 1990 book I co-authored. In "Technology in Banking: Creating Value and Destroying Profits," we argued that technological investment does create value--but for customers, not stockholders.
Nearly half a decade later, it's worth looking at these questions again.
To conceptualize the links between banks' technology expenditures and their business needs, we divided total noninterest expense into three components, which we called M1, M2, and M3. ("M" stands for management level and is unrelated to the monetary aggregate terms.)
M1 represents the bank's core technology infrastructure, including all computers, data centers, telecom facilities, workstations and PCs, local area networks, systems software, spreadsheets, and data feeds. M1 equals about 5% to 10% of noninterest expense. The distinguishing characteristic is that nothing in it is industry specific. By definition, the components of M1 technology are the same in any industry.
Banks, however, do have recognizable patterns in how they use M1 technology. For example, banks use lots of transaction processing technology but no parallel vector supercomputers. Banks employ many classical mainframe technologies, such as IMS, CICS, and COBOL, yet have used comparatively little open systems technology. Patterns may also vary by line of business or geography. Unix is common in trading rooms but not in the back office. The operating system is also more prevalent in Europe than in the United States.
In our book, we argued that no bank could get a competitive advantage solely through the use of M1 technology, simply because the technology is available to everyone. Within a few years of its introduction, any successful and useful M1 technology will be picked up and adopted by most players.
Banks, of course, can misstep on M1 technology by picking the wrong vendor or the wrong platform or waiting too long to invest. Most banks, however, avoid these mistakes.
M2, which we subdivide into two other categories, M2-A and M2-B, involves the creation of bank products and services. M2-A is the technological component, primarily those industry-specific software applications. Examples include DDA and TDA systems, consumer and commercial loan systems, securities processing systems, authorization systems, collection systems, and item processing systems. About 10% of M2-A is specialized hardware such as item processing equipment, check encoders, or ATMs.
The cost of M2-A is about 5% to 10% of noninterest expense. Competitive advantage is possible through proprietary M2-A technology--if there is a unique business strategy. But the vast majority of the time, bank products or services are commodities and don't warrant proprietary investments. Thus, banks have tended in recent years to purchase application packages from third parties and deemphasize the building of new proprietary systems.
M2-B is the manual component of production, e.g., the back office. U.S. banks employ about 1.4 million workers, and most work in M2-B. This includes customer service representatives, tellers, clerks, operations personnel, collectors, researchers, and so forth. M2-B represents about 50% to 60% of a bank's noninterest expense.
Competitive advantage may be gained in M2-B through depth of knowledge and work-force skill. Banking is still very labor intensive. And most products and transactions require a complex mix of M2-A applications software and M2-B manual effort. Exactly what knowledge is required to process bank products is almost never written down but exists in the collective minds of the operations and clerical staffs.
M3 represents the business needs of the bank, reflected in marketing, decision-making, and executive personnel. Costs are generally about 20% to 30% of noninterest expense. Key competitive advantages may be the business strategy, the bank's franchise, location, credit skills, balance sheet, or customer base. Generally, most competitive advantages begin in M3 and should drive specific investments in both M1 and M2.
How do M1, M2, and M3 relate to each other? Think of a bank trader staring at a computer screen. The cost of the trader and what he wants done is M3. The cost of the data on the screen and the applications that put the data there are M2-A. The cost of the deal ticket and the back office work generated by each trade is M2-B. The cost of the screen itself, and all of the cabling and computers behind it, is M1.
Trends
The boundaries of M1, M2, and M3 are not static. Over time, M1 technology has expanded in scope and power. For example, programmers used to put features in their applications programs (M2-A) that are now done by systems software or by middleware (M1).
Historically, banks have built a certain amount of proprietary M1 technology. Mellon Bank, for example, built a telecommunications monitor called Inquiry in the 1960s, before the now-standard CICS was available. Inquiry is still in use, although Mellon will probably ultimately convert off it. Proprietary M1 makes sense where no industry standard is yet available but standards usually evolve over time.
Manual operations in M2-B are constantly being displaced by M2-A investments. Examples include:
* Use of image technology to replace manual encoding of checks.
* Workflow systems to automate the handling of file folders.
* On-line histories replacing manual look-up by CSRs on paper journals or microfiche.
* Electronic data capture replacing paper sales drafts at retailers.
* Electronic check presentment moving to help truncate checks.
Displacing M2-B costs is the primary reason banks invest in technology. Manual or paper processing costs are high and have fewer economies of scale. Once applications systems automate each incremental task or function--a process taking years--productivity increases.
The issue of timing continues to bedevil banks. Advocates for technology usually espouse a rapid build-up of costs in M1 and M2-A to reduce even higher levels of cost in M2-B. This may also enable banks to reduce their overall operating cost ratio. One of the inhibiting factors is the complexity of bank operations and the need to keep the back office running while the investments are made.
Also, M2-B costs as a rule don't decrease in absolute terms following an M2-A investment. Rather, the investments result in increased "value," measured through higher volumes, new feature-functionality, or increased levels of service.
Further, the buy-versus-build issue never seems to go away. As infrastructures have aged, many of the purchase decisions now concern services. Banks are looking to buy systems integration, tailoring, reengineering, and conversion services far more than in years past. Outsourcing and modular application components are also popular.
M3 is also changing under the force of new business needs. Banks must manage market and credit risk better, as derivatives and securities have grown in volume and complexity. Banks must learn to market mutual funds, pick up loan growth again, and develop non-branch retail electronic channels. They must have an acquisition or merger strategy and must move to rationalize both branch networks and back offices. They must find new products that generate fee revenue while coping with new regulatory and compliance initiatives. Many of these developments will create new M2-A investments.
Rules of the Road
The M1-M2-M3 framework suggests some basic rules of the road for banks in the future:
* Don't invest in proprietary M1.
* Don't buy vendor application solutions that use any nonstandard M1 technology--e.g., little-known development tools.
* Avoid the bleeding edge--and the lagging edge--in M1. Don't try to gain a competitive advantage here. Try to be in the middle of the pack of the fast-follower segment.
* Anticipate the expansion of M1 capabilities. Don't build M2-A applications that do things that will later be available through M1 technologies--unless the return is assured before that happens.
* Use industry-standard M2-A technologies for all commodity banking products, processes, functions, and channels unless there's a powerful business strategy that justifies a proprietary solution.
In summary, the business of banking and its large technological infrastructures are heavily interlinked. The effects of each on the other are complex and difficult to discern. They would be hard to manage even if technology weren't changing so rapidly. But because it is, making the right decisions is twice as hard. Vigilance and careful observation are needed just to stay in the game.
