It has been a good year for business lenders. Economic conditions have stimulated loan demand, and commercial loan originations in 1995 are up 10% to 15%. Banks with excess capital and cleaner books are actively seeking quality assets.
Banks are also investing more this year in technology to support this venerable business. Overall spending for commercial lending information technology grew 4.9% this year, to $904 million.
The goals of today's systems investments are twofold: to streamline the loan production pipeline and to strengthen management of the asset pool. A key to both is improving the productivity of the staff professionals who manage credit risks and customer relationships.
There has been limited new development in core accounting systems, the legacy systems comprised of both proprietary and vendor-provided applications that support commercial lending. But two examples stand out:
*LS/2 by Bankers Trust and International Business Machines Corp. has cost more than $22 million to develop over the past five years. The business priority of this mainframe system, which uses a DB/2 data base, is front-end sales of syndicated loans. The application, which has a PC-based front end, has been licensed by First Chicago and Continental Illinois, now a part of BankAmerica Corp.
*ACBS by Advanced Information Resources is an AS/400-based package that has been in development for more than two years. National Bank of Canada was the first of more than 10 institutions to license this relationship and credit management package.
Because of securitization, there are enormous pressures to increase the efficiency of the commercial loan production pipeline. Simple large corporate borrowings have moved to the capital markets, leaving for banks a combination of smaller loans, risky loans, and complex loans. To profitably handle such portfolios, either the costs must be driven down or the productivity driven up. Generally, this has meant:
*New sales and marketing applications to proactively find loan customers - instead of waiting for walk-ins or hanging out at the country club.
*Automated credit risk evaluation tools.
*Automated loan documentation and workflow management tools.
*Specialized pipelines for small, midsize, and large customers.
In small-business lending, loans are generally under $250,000. Given typical spreads, this leaves very little margin for operational expenses. So some of the same techniques used in consumer loan marketing and credit approval are creeping up into these smaller commercial loans.
Credit scoring for these smaller commercial and industrial loans can automate the application process and cut the cost of making the decision. Here, most banks have sought the help of a vendor.
Fair, Isaac & Co., building on its strength in consumer credit scoring, is the leader with more than 100 bank customers for its PC-based product. Other small-business solutions include Appro's 20/20 product and Dun & Bradstreet's Banking Predictive Scoring.
Because scoring techniques for small-business loans are so new, a safe estimate would be that they are in use at less than half the top 100 banks and fewer than 1% of the remainder. Of the larger banks that have implemented such systems, roughly a quarter have built proprietary solutions rather than choose vendor software packages.
The few active smaller banks have all turned to vendors, especially to PC-based systems. But all banks are proceeding gingerly - experimentation, evaluation, and extended rollout is the name of the game. We believe that credit scoring for small, standardized commercial loans will be adopted widely, albeit slowly, by the industry.
Middle-market loans, which range from $250,000 to $10 million, represent the greatest challenge to the industry. Most credit facilities are just part of an overall relationship that also involves noncredit products such as cash management, deposit products, loans sales, trade services, and investment management. Pricing is complicated. Within this customer group, managing the pipeline is important, but portfolio information and improving the relationship yield is more critical.
While sales and marketing professionals addressing the middle market require integrated systems support, they generally don't have it. Existing credit analysis techniques are idiosyncratic, making risk assessments inconsistent and case-by-case, which drives up operational costs. Task assignments and workflows for sales and credit professionals are most often based on administrative thinking, not operational efficiency.
Too many marketing systems and data bases today are replicated and isolated personal lists of names and prospects, not bankwide resources. Individual PC and laptop systems are common, but not effective, tools in supporting the sharing of information. What's needed is one common data base for all commercial relationship employees, with extracts from multiple transaction processing systems to provide meaningful data on existing relationships. Messaging, referral management, and file synchronization are three additional missing elements.
One example of a common-data approach is Twin City Bank's shared development of Gemini - done in partnership with Broadway & Seymour's MRI application. This client/server system incorporates both sales and financial modules to measure and maximize relationship yield. It captures and makes available to all users information on multiple aspects of a customer relationship. With over 100 users, this Microsoft Corp. SQL Server and Windows-based system has supported a 15% growth in fee income at Twin City, a Norwest Corp. subsidiary, and a shift of the revenue mix to a 60/40 fees-to-interest spread.
Automated credit analysis tools incorporate both spreadsheet applications and rating systems. Spreadsheet systems, such as Halycon's Fiscal, analyze borrowers' financial data. Spreadsheet and rating systems, such as Financial Proformas Inc.'s Fast or Crowe Chisek's Famas, add the ability to rate a risk within a consistent framework.
Some sort of spreading and rating technology is used by all of the top 100 commercial lenders. About 10% of these are proprietary systems. The rest of the top 100 banks - and virtually all smaller lenders - rely on packaged applications. Capabilities and costs vary widely, however.
Today, very large customers will take their straightforward funding requirements to the capital markets, and turn to banks only for more specialized or complex borrowing needs. Unlike smaller borrowers, large customers usually transact in volumes sufficient to warrant labor-intensive attention to each function.
Nevertheless, the need to maintain profitable relationships with such firms places the same information demands on banks as are needed to serve the middle market. Multiple members of an "account team" each have expert knowledge on particular bank products and services. They require an infrastructure that supports a common understanding of the relationship, as well as the ability to customize credit facilities, including secondary market and off-balance-sheet products, and the associated risk assessments.
Most large lenders have a mix of loan sizes and therefore face three key needs: creating pipelines that are different for each class of loan; linking the different existing systems that cover the different functions; and automating the management of risk within the overall portfolio.
Portfolio management should go far beyond the traditional focus on simply working nonperforming loans. By using data bases and applications effectively, lenders can work toward the goal of maximizing aggregated, risk-adjusted profits. Techniques like "expected default frequency" help banks look at the various risk characteristics of the current portfolio, and how they have changed over time. We estimate that only 35% of the top 100 banks have built portfolio analysis tools supported by their risk rating system.
Including existing systems, a lender may therefore end up with a dozen or more separate systems that touch on the commercial lending workflow and relationship management, i.e., the pipeline and the asset pool. The very largest banks are therefore seeking an integration that allows some degree of data sharing across these applications. Because of a relatively high share of proprietary applications and/or unique business, the very largest lenders will require systems integration approaches.
What will success look like? Those lenders that can accomplish one-time data entry and minimal labor effort per commercial relationship will keep per-loan costs at acceptable levels. Those that manage their risk better will price new risks better and control losses better. And those that invest in effective marketing systems should be able to increase share. However, none of this can be done without recognizing the costs of the systems investments. Like all technologies, commercial lending systems require a certain amount of scale. The business is already concentrated and will continue so. Smaller lenders must be careful about the effect on them. They should use standard packages and focus their lending on niche areas where they can avoid undue competition.
Diogo Teixeira is president of Tower Group, a consulting firm in Wellesley, Mass. Patricia McGinnis, a technology analyst at Tower, contributed to this article.