Something old and something new for risk managers

By Dr. Mark Greene, CEO, Fair Isaac
   

Dr. Mark Greene

In its embrace of predictive analytics, neural networks and other analytic technologies, the financial services industry created tremendous value through innovation.

As the credit industry looks for short-term rescue and long-term lessons from the credit crisis, two directions seem clear. These directions seem to take fundamentally different shapes. In fact, however inappropriate a wedding analogy might seem in these dark days, we could call them “something old and something new.”

What’s “old” is the need for sound risk management practices. Over the past 30 years — as an economist of the Federal Reserve Board, as a lender at Citibank and as a service provider at IBM and now Fair Isaac — I have seen the risk management discipline in banking mature. But I have also seen risk management erode in the current credit crisis and the US subprime meltdown that preceded it.

Many lenders rushed to embrace an unproven business model (exotic mortgage products) while ignoring established foundations of credit underwriting. Many lenders took a shortcut by relying solely on credit scores — in fact, my company’s FICO® scores — for credit risk assessment. In doing so, they ignored and in some cases even falsified collateral and capacity to repay, even as consumers falsified their information to get the loan.

Misrepresentation and misunderstanding of the borrower’s risk was certainly part of the problem. A bigger part was that no one clearly assessed the risk inherent in the product itself — which no credit score will capture.

Sound risk management isn’t sexy, it’s just critical. In our experience, the banks that are faring better during these turbulent times share some common characteristics:

  • They understand the tools they use. In particular, they know what a credit score tells them and what it doesn’t tell them. For example, one of the largest US banks changed its small business portfolio model to balance its scoring with greater risk underwriting. While this lowered the approval rate, it also meant better risk exposure for the bank.
  • They exercise caution with risky consumer segments and new lending products. They know that when they target a new consumer segment, those borrowers may have different payment and performance patterns than the rest of their customer base. They also know that new lending products may generate very different customer payment patterns. Put the two together — as in exotic subprime mortgages — and you have a potential time bomb. Banks that were more cautious in sampling the subprime market, and more conservative with a diversified product mix that crosses geographic and cultural boundaries, showed that prudence pays.
  • They remain committed to testing and tracking. This is how you determine the success of a new strategy, customer segment or loan product. With the possibility that we’re seeing both technological and cultural paradigm shifts, significant variation is likely in results, which makes testing and tracking paramount. And while there is no easier way to gain insight into portfolio trends, we can think of few lenders that use tracking and monitoring — let alone champion/challenger strategy testing — to their full potential.

So what’s the “new” thing for lenders struggling to build their health? Analytic innovation.

A recent TowerGroup study* confirmed that the top technology initiative for banks striving to deal with economic conditions is to improve their analytics and modeling capabilities. That’s good, because too many lenders today rely on older models and older versions of systems.

Most lenders are not using the most advanced analytics for collections, fraud and risk assessment. These include innovations such as:

  • Transaction-based risk scores that can do flag likely bankruptcies earlier than the cycle-based scores that are the industry standard
  • Fraud analytics that consider merchant profiles in addition to cardholder profiles, showing a clear ability to identify substantially more frauds in real time at any account false positive rate
  • Optimization and decision modeling that have shown bottom-line increases in marketing, pricing, credit line management, retention and collections

In its embrace of predictive analytics, neural networks and other analytic technologies, the financial services industry created tremendous value through innovation. That’s where lenders should be looking as they seek new ways to grow their business in turbulent times, while protecting their customers, their shareholders and their balance sheets. Sound risk management combined with sharper use of analytics — something old and something new — represents a great union.

* TowerGroup, 2008 Top 10 Business Drivers, Strategic Responses, and IT Initiatives in Retail Banking (Robert Hunt and Kathleen Khirallah, Dec 2007, #v53:55RCN).