Credit scoring models assess the ability and willingness of borrowers to pay their debts using data collected by the three consumer credit reporting companies: Equifax, TransUnion, and Experian. Regardless of their income or wealth, borrowers obligate themselves for debt repayment in various ways – many successfully, but some, through poor matching of income and consumption, disorganization, or for other reasons, less successfully. The assessment of debt repayment behavior expressed as a credit score is highly predictive of the probability of repaying current and future debts.
Credit scoring models have been part of Fannie Mae’s automated underwriting since we introduced Desktop Underwriter® (DU®) in 1995. DU created a huge advancement in the precision of mortgage underwriting, which previously had relied largely on the loan-to-value ratio of the proposed loan, with no analytic consideration of credit history. Initially, DU relied heavily on the prospective borrower’s FICO® (Fair Isaac Corporation) credit score as the primary indicator of creditworthiness. In 2000, we replaced the credit score with a proprietary credit risk assessment that is more predictive of performance because it is modeled directly on Fannie Mae loans.
DU’s comprehensive risk assessment considers a number of factors (see Fannie Mae Selling Guide section B3-2-02: Risk Factors Evaluated by DU) such as loan purpose and loan-to-value ratio as well as borrower credit report data. To assess borrower creditworthiness, DU considers many credit report factors (described in the Fannie Mae Selling Guide, section B3-5.3-09: DU Credit Report Analysis).
Trended Credit Data Improves Modeling of Loan Performance
In recent years, expanded information on borrower credit histories has become available. What is called “trended credit data” is historical data at a tradeline (credit line) level on several monthly factors, including: amount owed (balance), minimum payment due, and payment amount made. In 2015, Fannie Mae used 3.7 million credit reports with trended data (dated June 2009 through August 2012) to conduct modeling and analytics to support a comprehensive review and redevelopment of DU’s credit risk assessment.
Including trended data materially improved modeling of loan performance. Based on that finding, Fannie Mae has worked with the credit reporting agencies to have trended data included in the consumer credit reports used for underwriting loans through DU, effective with DU Version 10.0 (scheduled for release the weekend of June 25, 2016). Trended data is not considered in currently available credit scores, so consideration in the DU credit risk assessment will be its first widespread use in the mortgage lending industry.
Trended Data Empowers Creditworthy Borrowers
Including the trended data in DU’s credit risk assessment: 1) improves the accuracy of DU’s overall risk assessment, and 2) will benefit borrowers who regularly pay off revolving debt, increasing the likelihood that they will receive an Approve recommendation from DU. This means that use of trended data in DU’s credit risk assessment can provide more creditworthy borrowers access to mortgage credit. The overall percentage of loans that receive an Approve/Eligible recommendation is expected to remain relatively stable.
Giving weight to how borrowers pay off credit debt puts more power in their hands to control their credit evaluation. Payment delinquencies are a significant factor in credit scores, and borrowers can do nothing but wait for the delinquencies to grow ever farther back in time. But when trended data is considered, by paying credit card balances in full or in large part for a few months, borrowers can demonstrate that a late payment was not deeply reflective of their general debt repayment ability and behavior.
Based on Fannie Mae’s analysis, borrowers can potentially improve their evaluation by the DU credit risk assessment each month by paying off credit card bills in full.
DU Supports Access to Credit with Prudent Risk Management
DU has been the industry leading automated mortgage underwriting system for more than 20 years. DU’s evaluation is fair and objective, applying the same criteria to every mortgage loan application it considers. Fannie Mae is committed to continuous improvement of the DU risk assessment model. We continue to make ongoing investments in our risk management tools, to enable the origination of better performing loans, resulting in reduced costs to service those loans. The addition of trended data to the credit risk assessment is an update that will help to support creditworthy borrowers’ access to mortgage credit while reducing risk.
The author thanks Stacey Shifman and Kristi Heutink for their contributions to this analysis.
VP, Credit Risk Analytics and Modeling