American Banker  |  Thursday, November 13, 2008

By Chris Costanzo

Print Email Reprints Feedback

Vendors of credit-risk management software are developing a new generation of tools designed to make it easier for lenders to identify borrowers who could get in trouble when economic conditions change, and hopefully prevent a repeat of the subprime mortgage meltdown.

Consumer credit risk management today is largely built around credit scores, those three-digit numbers that represent a borrower's likelihood of repaying debt in a timely manner. But with delinquencies setting records in mortgage, card, and auto lending, the industry is recognizing it needs more than blind faith in scores to make good loans, especially in a souring economy.

The spiraling economy has highlighted a deficit in credit scores: They fall short in incorporating the effects of change in the big-picture economic climate. Observers say the vendors that can move beyond FICO scores and implement a new generation of tools to analyze customers against the backdrop of the economy will be positioned to grow in the years ahead.

The message that credit risk management practices might need review could not have been louder than it was in recent months.

In September alone, Lehman Brothers filed for bankruptcy; Bank of America Corp. agreed to buy Merrill Lynch & Co. Inc.; and the government took over Fannie Mae and Freddie Mac; and in October, the Treasury Department announced a sweeping bailout plan to invest hundreds of billions into the banking industry. All the troubled institutions had exposure to instruments tied to bad mortgage debt.

"If ever there was a wake-up call for the industry, this should be it," said Anthony Hughes, senior director of credit analytics at Moody's Economy.com. "This is the type of event that should be transformative of the practices people use to assess credit risk."

In retrospect, the monumental bust had been brewing for some time. Statistics from the Federal Reserve Board show that U.S. consumers have spent the past 10 years taking on unprecedented amounts of debt. Since 1998, total debt per borrower has doubled, while mortgage debt per borrower has increased 124%. The nation's total mortgage debt has doubled since 2000, to $9 trillion, about $1 trillion of which has an adjustable interest rate.

Providers of credit risk management tools are starting to realize the importance of the dramatic shift in consumer behavior. Credit scorers have always done a good job of identifying which consumers are more likely than others to repay debt, but they have not been so adept at analyzing macroeconomic trends.

Determining how, say, rising unemployment will affect credit risk is widely acknowledged as a difficult but increasingly important task. "There's a complete myopia in terms of assessing credit risk only at the individual level," Mr. Hughes said. "The problem is you miss the big picture."

An individual's credit history has to be viewed through the prism of the macroeconomic picture, or it will be inaccurate, he said. "People lost the forest for the trees. The forest was dying, but the trees looked OK."

The new consumer credit landscape has not been lost on risk management software vendors. Fair Isaac of Minneapolis is rolling out two products that address the need to bring a big-picture perspective to individual credit analysis. Two other vendors — SAS of Cary, N.C., and the much-smaller Strategic Analytics Inc. of Sante Fe, N.M. — are developing new credit assessment methods that would compete with credit scores. And VantageScore Solutions LLC of Stamford, Conn., has entered the market with a scoring methodology of its own.

Providers say lenders are more open to dialogue about new methods of risk management, given the recent catastrophic events in the financial markets.

"I would doubt there isn't anyone who isn't rethinking their models," said Barrett Burns, the president and chief executive officer of VantageScore. "The opportunity is terrific."

The providers also say they started working on their approaches to retail credit risk management well before the crisis unfolded, in response to either Basel II requirements or customer demand. Even so, there is no denying that the convergence of recent financial events and product development has been propitious.

Clark Abrahams, chief financial architect at SAS, was blunt. "If not for the subprime crisis, I don't think we'd be getting any airtime."

The "risk management suite" Fair Isaac introduced in January aims to address some of the new realities in consumer credit.

In response to the increase in consumers' debt, the suite includes a "credit capacity index." The patent-pending technology aims to show how much additional debt borrowers within a given range of credit scores can handle.

Careen Foster, senior product manager at Fair Isaac, said both lender and borrower behavior changes over the course of a credit cycle; when the economy is expanding, lenders tend to extend additional credit, making consumers more inclined to borrow.

"It's important to understand" how credit expansion "changes the way consumers take on additional debt," she said.

Andy Jennings, the chief research officer at Fair Isaac, said the "great irony" of credit scores is that they are used to make lending decisions, even though they do not indicate which borrower within a particular group will not pay. Instead, they rank borrowers against one another in terms of risk, but in reality, "there is a difference in susceptibility in each of these people to an increase in the amount of credit they have."

A credit score can identify people with equal risks when all things are equal, he said, but the credit capacity index can go a step further and identify which people can handle more debt once credit gets more available.

Another component of Fair Isaac's suite is portfolio stress testing, which attempts to incorporate broader economic elements into consumer credit decisions, Mr. Jennings said. "The concept is that we see changes accrue in credit portfolios due to changes in the macroeconomic environment."

Stress testing is standard practice for well-managed portfolios, but the market was asking for a "more formalized" approach, he said. In addition, stress testing is required under Basel II, so the suite has a "Basel II flavor to it."

Fair Isaac says both features were in development well before the credit crisis, though "lenders are much more in tune with the need" for these features, Ms. Foster said.

SAS is also finding a more attentive audience for a new method it is promoting to assess consumer credit risk. The vendor says its "comprehensive credit assessment framework" addresses the fact that scores generally do not incorporate certain pieces of information, such as noncredit payment histories, and do not tend to include borrower capacity as reflected in income, liquidity, total debt, and months of savings in reserve.

Mr. Abrahams said some of these elements may be considered during the application process, but the comprehensive framework improves on that by putting all the information into the proper context at once.

He called the framework a holistic application that marries judgment and scoring.

"Credit scoring averages everyone," he said. "You've got to build context and consider the specifics of the borrower."

Incorporating all this information into a score will help overcome the "villainization of products" in the mortgage business that have let borrowers take out loans they really could not afford, Mr. Abrahams said. "This is good medicine that tastes bad."

Strategic Analytics also is offering a new way to assess credit quality. The nine-year-old company says its software addresses the twin challenges in retail portfolio analysis: Loan volume is high, and information on each one is scarce.

"We have developed a new way of building a credit score," said Joe Breeden, the vendor's president and chief operating officer.

The software, which is used by eight of the top 10 global financial institutions, including Discover Financial Services, HSBC Holdings PLC, and Royal Bank of Canada, seeks to overcome some of the limitations of credit scoring, Mr. Breeden said.

Scores, for example, do not predict how many loans within a portfolio may go bad, and they do not take into account the actual product offered to a consumer, he said; putting a person with a decent score into the wrong product can result in a bad loan.

Strategic Analytics' software, oddly enough, seeks to overcome these shortcomings by using the principles of earth science.

Scientists studying tree rings to understand climate change look at the local conditions, and the environment. Similarly, Strategic Analytics says its software takes into account the life cycle of the product, the quality of the "vintage" (a group of loans made in a given period), and the impact of the environment.

Using this approach when the economy was still booming, Strategic Analytics could see that the relative quality of loans was deteriorating rapidly, Mr. Breeden said. "A credit score doesn't know the difference between an account that goes bad in a good environment and one that goes bad in a bad environment."

The software also is designed to predict the probability of default at the account level, given a certain scenario for the environment, he said. "It's a rating versus a ranking. We're trying to introduce a fundamental technology shift."

VantageScore says it has refrained from reinventing the wheel; it tries to address issues with credit-scoring options. Mr. Burns said it grew out of the demand for scores that were more predictive and applicable to more people.

Companies could build their own customized scores, but those were expensive, he said. "They wanted a better generic score."

VantageScore says it addressed the issue by creating common definitions of the 4.5 billion data points that are generated every month by consumers and stored at the three main credit bureaus. It revalidates the data annually to ensure the model remains predictive even in changing economic conditions.

Ms. Costanzo, American Banker's technology editor from 1998 to 2002, is a freelance writer in Maplewood, N.J.

More articles in Technology