Are credit scores assigned to conventionally unscoreable consumers considered reliable?

Date: July 01, 2020

The answer is yes, but first a bit of context:

Philosophically, we at VantageScore Solutions approach credit scoring differently than our chief competitor, FICO. We made inclusiveness a priority in our model design from the very beginning. Our competitor only assigns consumers credit scores after six months of credit usage and/or if they’ve had an account update in the past six months. That policy shuts out new entrants and consumers who go six months without using a credit account from automated underwriting systems that require credit scores. The lack of a credit score can cause lenders to ignore these consumers altogether, even though many are highly creditworthy. It can also subject some consumers to unfair loan pricing, in the form of higher fees or interest rates.

Here’s how the VantageScore model scores more people:

  • The VantageScore 3.0 model uses a broader and deeper set of credit file and advanced modeling techniques to capture unique consumer behaviors more accurately.

  • The VantageScore model recognizes the predictive value of tradeline data more than six months old, and also may use data that is more than 24 months old.

  • The VantageScore model provides scores for consumers whose oldest trade is less than six months old and also can provide a score after the first tradeline update.

  • The VantageScore model includes a 13th scorecard designed specifically to generate scores for people with very sparse credit files.

Our model reliably scores 98 percent of consumers with credit files, including 30-35 million more consumers than conventional models. These newly scoreable consumers include, for example, the millions of consumers who use credit infrequently (e.g., not in the past six months).

Scores the VantageScore model assigns to consumers ignored by legacy models are highly accurate:

  • Each year VantageScore Solutions validates its scoring models by conducting a series of tests to ensure that the model performance aligns with expectations. Results are made public each year. In the score range of 600 to 850, VantageScore 3.0 outperformed its benchmark by a double-digit improvement in mortgage performance as measured by the Gini[1] coefficient.

  • The Gini coefficient for the validation sample composed of newly scored consumers was 52.3. A Gini coefficient of 45 or more indicates effective rank ordering by industry standards, and we are pleased with these validation results.

  • Newly scored and conventionally scoreable consumers behave nearly identically when measured with another commonly used industry benchmark: first-payment default rate (which computes the frequency with which new borrowers fail to make their first payment on a new loan or credit card account).

  • We have passed the ultimate test: More than 2,000 heavily regulated lenders have tested and decided to use VantageScore in their own businesses.

Scoring more people is not a replacement for prudent underwriting. In the mortgage market, newly scoreable consumers would still need to meet requirements for down payment, income, employment, and many other factors. But having a credit score enables those consumers to pass through the first gate and, if they ultimately qualify, to pay fair prices for their loans.

In other words, using VantageScore doesn’t lower the bar, it simply widens the window.

[1] Gini is a standard measure of model performance.