Featured Listings:

Credit Scores

by David Shellenberger
Fair, Isaac and Company

Credit bureau risk scores are one of the many elements reshaping today’s mortgage industry. While credit scoring has existed for more than four decades and is used extensively in such industries as bankcard and auto lending, mortgage lenders have embraced this technique only in the past several years. In 1995 the Federal Home Loan Mortgage Corporation and the Federal National Mortgage Association both recommended that all their loan sellers and servicers use credit bureau risk scores to identify and control credit risk. Today mortgage lenders, including those offering high loan-to-value products (HLTV), commonly use credit bureau risk scores as part of the underwriting decision. To understand the rationale of HLTV lenders approaching this marketplace, it is important to understand how credit bureau risk scores are used to assess a mortgage applicant’s likelihood to repay a mortgage obligation.

Credit bureau scoring is a statistical means of assessing a borrower’s likelihood to repay a loan. The score measures the relative degree of risk that a potential borrower represents to the lender or investor. The models used most widely by mortgage lenders are the Fair, Isaac Credit Bureau Risk Scores or FICO scores. These scores, ranging from approximately 300 to 900 points, are available through the three national credit data repositories (Equifax, Trans Union, and Experian). FICO scores are scaled so that higher scoring borrowers show a relatively lower chance of default than lower scoring borrowers. The scores are derived solely from the data available in a borrower’s credit file. The models do not include information about the borrower’s income, assets, job history, or bank account; such factors lenders and investors consider independently of the score. A borrower’s capacity to repay a credit obligation is particularly relevant in underwriting HLTV mortgage loans and must be considered in addition to the credit bureau risk score.

To develop these empirically based models, Fair, Isaac uses the credit files of millions of consumers and applies complex mathematical methods to perform extensive research into credit data that forecast credit performance. Fair, Isaac identifies distinctive credit patterns that correspond to a likelihood that a consumer will make loan payments. The score is based on all credit-related data in the credit bureau report—not just negative data such as missed mortgage payments or bankruptcies.

A Fair, Isaac credit bureau risk score ranks potential borrowers based on the likelihood of paying their credit obligations. A higher score indicates better credit quality. All other things being equal, borrowers with a score of 660, for example, are less likely to default on a loan than borrowers with a score of 620. Different industries, geographic regions, product types, and individual lenders have illustrated the reliability of the model in ranking borrowers. Table 1 shows the performance of FICO scores in segmenting borrowers in terms of mortgage credit risk; a national sample of mortgagors on new mortgage obligations over approximately twenty-four months provided the data.

Table 1

Credit Score Ranges for New Mortgage
Borrowers from a National Sample

While credit bureau risk scores will consistently rank potential borrowers based on risk, the number of good loans to bad loans denoted by a score is likely to fluctuate because of changes in the economy, regional differences, changes in product offerings, or other reasons. Lenders using credit scores will compare the performance of loans over time to determine the relationship of score and performance for their own business environment.

Generally, mortgage loans with a high loan to value tend to experience a higher default rate than low loan-to-value mortgage loans. While HLTV products may tend to represent a relatively greater credit risk, lenders can use credit bureau risk scores to stratify loan applicants by relative default probability. Credit bureau risk scores enable a lender to acknowledge the risk across an entire spectrum and target applicants with a higher likelihood of repayment. Table 2 illustrates how a lender performs this segmentation by observing borrower performance over time.

Table 2
Risk Segmentation Using Credit Bureau Risk Scores for First Mortgage Loans
(in percent)

Note: This illustration is based on first-mortgage loan performance and reflects borrowers going to severe delinquency over a 24-month period. It does not reflect actual 125 percent LTV-type product performance, although the principle of risk ranking would be the same.

Lenders that use credit bureau risk scores in underwriting strategies must first calibrate the models to their own business environment by comparing the scores at origination with the performance of loans over time. By tracking loan performance according to credit bureau risk score over time, lenders can relate the scores to loss experience and can continuously refine their underwriting strategies and control their exposure to credit losses.

Credit scoring has become an integral part of HLTV lending and enables lenders to make fast, consistent, and accurate underwriting decisions. The data acquired as part of the credit-scoring process are extremely valuable in the continuing management of credit risk as well as portfolio valuation. Through credit scoring, lenders can better acknowledge the risk within borrower populations and price accordingly. As seen in the arena of consumer lending, better assessment of risk allows lenders to create new loan products serving the credit needs of more individuals.

Note

1. Fico is a registered mark of Fair, Isaac and Company.