|
|
|
||||
|
Return to Article List Return to Article Search Email this page to a friend |
Managing Current Credit Score TechnologyOctober 2006 Ken Williams, KW Financial Solutions Managing loan risk was historically about the relationship between the lender and the borrower. With the advent of empirical technological solutions, a third party, the credit score, was added into the equation. Credit scoring is supposed to make life easier for the lender, but is it resulting in more confusion? With the introduction of the Vantage score further adding yet another lender option, it is worthwhile to see where we are today. This article is a snapshot of today’s scoring environment and will visit four key issues important to retail lenders: 1. What are today’s market place choices for credit score sets? 2. Why are lenders increasingly turning their backs on the marketplace choices? 3. Four key fundamentals to a better understanding of scoring. 4. The six steps necessary to make credit scoring a full partner in the lending process. The Market Today When it comes to scoring models for credit approval, today’s lenders have a confusing array of choices. There are 16 score models with 340 subsets or sub-populations to choose from. There are the three consumer reporting companies (CRCs) and Fair Issac shown in exhibit 1. Fair Issac is included because of its Expansion model that is used to grant credit for potential borrowers with little or no credit. For the most part, the designer of the score set is Fair Issac, although each CRC does have its own model. The recently introduced Vantage score model is not included in this comparison. Each model also has from 24 to 30 segments or subsets such as sub-prime, auto, bankcard, retail revolving and bank installment. These are used for product or industry-specific lending. The Reality of the Marketplace It is no wonder that lenders, when faced with this confusing array of options, turn their backs on them and are reluctant to budge from their historic model of choice. The actual offering of scoring models has been pared down by lender demand to those shown in exhibit 2. Primarily lenders have stayed with the classic line of credit score models, and few have migrated to the Next Generation versions. In a number of cases, lenders were even reluctant to upgrade to the newer versions of the classic models, and only did so when the CRC stopped providing support for the older version. The reality of score set usage is also driven by the asset size of the institution as shown in exhibit 3. This is a dilemma for both the score providers and the lenders they serve. The CRCs would like to see greater usage of the newer models that represent a significant investment for them. Lenders are reluctant to changes in the process, since any change affects the dynamics of risk management. The entire lending process is a tender balance of saying yes to the greatest number of borrowers, while keeping losses at a minimum. This inherently places both groups, who are business partners, at odds when it comes to score set upgrades. Keys to a Better Understanding of Scoring There are four basic facts lenders need to take to heart concerning empirical technology: 1. Credit scoring is a technology and, like other technologies, it evolves over time. Some of the CRCs’ classic models are based on 1995 to 1997 data. Do you use a nine-year-old computer? 2. Precision, precision, precision. The more precise a score set, the better it is at doing its job: predicting loss. Newer score sets and their subsets provide higher levels of precision when properly applied to the underwriting process. 3. The level of risk for a given point score will be different among the three CRCs. Figure 4 shows the point score to loss ratio for the same model offered by the three CRCs. In the critical mid-to-lower 600s, the loss difference can be plus or minus 100 basis points. This does not mean one model is less effective for use than another; it only means the lender must revisit cutoff score assessment if the CRC is changed. 4. The level of risk for a given point score will also be different within a single CRC using the same model, but using different subsets. Figure 5 shows this relationship. Again, this is not a flaw in model design, but recognizes the differing risk parameters within the subsets. It actually shows the subsets can be more precise. What Can a Lender Do? Lenders are not without options for navigating the enhanced scoring environment: 1. The greater the precision, the greater the reliability, resulting in more approvals, lower losses and less processing costs. It is worth the effort to upgrade to the newer models. 2. Newer models are more precise. If you are using one of the classic models, it is time to upgrade. 3. Subset models are more precise. See if your volumes justify the effort to adopt multiple subset models into your process. 4. Consider elevating the weight of a credit score in the overall decision process. This will pay dividends in reduced costs and, with higher precision levels, there is no increase in risk. 5. Understand there can be point differences between models, and revisit risk level cutoffs and rate assessment tiers. 6. Internally track your own results by point score. The best validation is an internal one. The fact most lenders are not taking full advantage of what the technological advances in credit scoring can do for them is good news. This means the migration from last century’s technology offers a significant opportunity to enhance risk management, reduce expense and increase the bottom line. Click here to view exhibits. Hoosier Banker articles are published by the Indiana Bankers Association. With the exception of official announcements, the Indiana Bankers Association disclaims responsibility for opinions expressed and statements made in the articles published in Hoosier Banker and/or appearing on the IBA website. Unless requested otherwise by the author in writing, all material published in Hoosier Banker and/or on the IBA website is the property of the Indiana Bankers Association. |
|