VantageScore paper examines credit score accuracy without some negative elements

STAMFORD, Conn. - 

VantageScore Solutions — developer of the VantageScore credit scoring model — recently released a white paper that found that credit score models can maintain accuracy and predictive power even if certain negative data such as tax liens, civil judgments, and medical and non-medical collections are removed from consumer credit files.

The paper, “Negative Data Suppression and the Impacts on Credit Score Models,” anticipates potential model-development questions raised under the National Consumer Assistance Plan (NCAP). Spearheaded by the three national credit reporting companies (CRCs) — Equifax, Experian and TransUnion — the NCAP is focusing on a variety of measures aimed at making credit-data reporting more accurate and understandable for consumers.

The CRCs are still making a final determination about the measures to be included in the NCAP, which is scheduled for full implementation by March 2018, but so far the recommendations are expected to include removal from consumer credit files of all civil-judgment records, a substantial number of tax liens, and medical-related agency collections less than 180 days old. When included in credit files, these data are predictive of default risk, so most modern credit-scoring models take them into consideration. Such changes could impact a lender's customer acquisition and/or portfolio management processes.

Recognizing this, VantageScore said it embarked on research to discover how the removal of this credit-file data could impact scoring-model accuracy and the ability to develop high-performing models in the future.

Using two million anonymized consumer credit files and a methodology detailed in the white paper, VantageScore developed two scoring models for comparison — one based on credit files containing all the negative data NCAP may exclude and the other based on credit files with all NCAP-related data removed. A comparison of scores obtained by scoring the same population with both models revealed the following findings:

—Despite the predictive value of negative data that would be omitted under NCAP, credit score models can certainly recover predictive performance in their absence.

—In fact, by adopting new variables more closely aligned to current credit-usage behavior — as opposed to the “backward-looking” historical events that NCAP would exclude — models developed in the absence of NCAP data may prove more stable than older models.

—Consumers who scored above 660 using model designs that accommodate NCAP data suppression exhibited better credit-management habits and more stable loan portfolios than their counterparts scored using models built using the negative file data.

—Regulatory focus on eliminating negative data under NCAP, along with the findings outlined above, suggests lenders should evaluate incumbent scoring models to ensure they will perform well in the absence of NCAP-related data.

“In light of the potential for significant changes to these data as a result of the NCAP, lenders must evaluate their scoring models to understand expected outcomes as the CRCs move ahead with their plans," said Sarah Davies, senior vice president of research, analytics and product development at VantageScore Solutions.

“This paper demonstrates that there is an opportunity to provide some relief for consumers whose scores are affected by these data while also preserving model performance,” Davies continued.

For more details on the study, visit to download the white paper.