EXECUTIVES AT Fidelity Federal Bank, Glendale, Calif. believed they've found a better mousetrap for managing risk.
Using software that can gauge the implications of a variety of scenarios, Fidelity executives say they can better track assets and liabilities, and thus price products and services with a keener eye toward profitability than any financial institution has been able to accomplish in the past.
"It's a win-win situation when you do something like this," said Andre Shih, senior vice president, treasurer, and acting financial officer.
During the 1980s Fidelity Federal, like many other thrift institutions, lent heavily to developers of multifamily housing into the red.
Mr. Shih said Fidelity Federal had neither the tools nor the executive prowess to understand the risks in the loans it was making and to hedge accordingly.
"One has to understand the amount of risk inherent in going into certaibn businesses," observed Mr. Shih.
"And once you understand the risks inherent in each product, you need to qualify and price it."
Mr. Shih, who joined the institution in July 1991, convinced his superiors to invest in a new asset-liability management system known as Radar to help Fidelity Federal officials understand and price product risk.
Developed by Risk Management Technologies and marketed by Software Alliance Corp., both of Berkeley, Calif., Radar relies on a process known as option adjusted cash flow analysis.
That technique theoretically enables a financial institution to better manage interest rate risk by valuing more accurately products or investments with hidden options through a financial modeling process.
For example, a customer could rquest a customized mortgage - one that features an adjustable rate and lifetime interest rate cap that is smaller than normal.
With option adjusted cash flow analysis, Mr. Shih explained, his staff can analyze what it will cost the bank to offer a product incorporating such features and blend that cost into the price it charges the customer.
"It allows us to tailor a customer's requirements and put the institution in a break-even position," said Mr. Shih.
That capability places Fidelity Federal, a unit of Citadel Holding Co., in a better position that many financial institutions.
Many bankers are pricing in an undisciplined manner, observed Richard Herring, professor of finance and director of The Wharton Financial Institutions Center.
"Any kind of technology that will help you look at how a loan will perform over a number of economic scenarios will help you price it accurately," Mr. Herring said. "This would make a lot of sense for something like mortgage loans," which incorporate a variety of options and circumstances that lenders must be able to understand and hedge against.
According to Mr. Shih, the Radar system has proven useful in Fidelity Federal's mortgage lending activities. All mortgage loans originated by the bank since it began using the system in late 1991 are profitable, he said.
In addition to customizing individual mortgages, the bank also uses Radar to reprice its mortgage products on a daily basis, which helps sustain profitability.
"The responsibility we have is to make sure every investment we make from this point on is profitable; that the returns we make will compensate us for the risks we incur, over time. That's why we're using a tool like this," Mr. Shih said.
Fidelity Federal's use of the Radar system, however, is not limited to mortgage origination.
Mr. Shih said the bank uses the system's modeling capabilities to work with customers on restructuring old loans that have gone sour. The system also has contributed greatly to the design of new deposit products, he said.
For example, the bank now offers a deposit account that provides the security of a certificate of deposit combined with the flexibility and liquidity of a money market account. This new "open CD" account, Mr. Shih explained, has no limits on the amount of money a customer deposits or withdraws during its six-month term, and accrues interest based on average daily balances.
"What we do is use the Radar model to price the cost of that [flexibility] option," he said.
Thus, if the going rate on a six-month CD is 3.2%, and the modeling performed by Radar suggests the flexibility of the open CD costs the bank 20% in terms of its loss of guaranteed access to deposited funds, Mr. Shih said he would pay 3% interest on open CD account balances.
Although the concept of modeling is not new of the world of finance, the way Radar functions is something of an anomaly.
In the past, the asset-liability management systems required a certain amount of data pre-processing, according to David LaCross, president of Risk Management and chief architect of Radar. The obvious result was the loss of at least some accuracy in predicting and quantifying risks.
"In a derivatives business that loss of accuracy is a killer," observed Mr. LaCross, a former BankAmerica Corp. executive who developed Radar to fill a risk management void he discovered in his own banking endeavors.
With Radar, however, no preprocessing of data is necessary.
The system, which runs on Hewlett-Packard Co. workstations using a variant of the Unix operating system, can segment and sort through data on what Mr. LaCross describes as a "very granular" level.
That level of detail, he added, allows a banker to examine an investment scenario from a variety of perspectives and to achieve better insights into potential strategies.
The linchpin of Radar is the option adjusted cash flow analysis, which relies on a technique known as Monte Carlo simulation.
Simply stated, the idea behind Monte Carlo simulations is to run any number of "what if" risk scenarios and to chart the probable impact of each on a user's cash flow. With Radar, that process can be handled interactively.
"The key is to run a lot of scenarios - in our case we run 50 to 200 - and to look at the number of cases in which rates go up or go down and by how much," explained Mr. Shih.
That places the bank and its management team in a better position to decide what investment course and what degree of risk it is willing to take, added Mr. LaCross.
"You can distill reams of charts into a very succinct articulation or risk," he said.
"Even the most naive of boards of directors can understand that type of presentation."
Neither Mr. Shih nor Mr. LaCross would reveal precisely how much it cost Fidelity Federal to implement Radar. According to Mr. Shih, the combined cost of the workstations (the thrift previously used International Business Machines Corp. PCs), Unix software, and a graphics package to support the system ran between $60,000 and $100,000.
Radar is priced according to a licensing fee, which Mr. LaCross said depends upon an institution's scale of implementation, but which likely would range from $100,000 to $1 million a year.
Regardless of the cost, the payback on Fidelity Federal's investment in Radar has been almost immediate, according to Mr. Shih.
Nothing that the average bank incurs interest rate risk on the order of 15 to 410 basis points annually, he said that the real cost to a bank arises from not properly understanding and pricing risk.
However, Mr. Shih also cautions that not every bank is currently in a position to benefit from a tool such as Radar, because the system requires good raw data in order to provide valuable results.
"My guess is that a lot of institutions do not have data in the state it needs to be in to be of value in a model like this," he said. It seems that using the better mousetrap also requires the right kind of cheese.
