Trended credit data brings new accuracy and fairness to credit scoring

Trended credit data brings new accuracy and fairness to credit-scoring

Date: June 22, 2020

Among the many new data sources becoming available, perhaps none hold greater promise for effective usage in the mainstream consumer credit industry than trended credit data. Trended credit data captures the trajectory of borrower behaviors over a period of time, as opposed to a snapshot of consumer behavior from the previous month.

Of course, readers of THE SCORE know that VantageScore 4.0 is the first and only tri-bureau credit scoring model to incorporate this data, and you can read here how it improves the score’s predictive performance. Indeed, among Prime and Superprime consumers the use of these data contributes to predictive lift of up to 20 percent. This lift enables lenders to make even more finely-tuned credit offerings to these consumers.

Industry participants rightly want and need full transparency when it comes to incorporating trended credit data or using models that leverage it. In the spirit of that need, we will be diving deep into this subject in a whitepaper that will come out later this year.

But in the meantime, a little teaser…

Among the topics we will address in this white paper will be the extent to which trended credit data can impact the majority of consumers in the United States – and how. Moreover, what other opportunities exist if more trended credit data were to be reported?

To wit, there are approximately 155 million “conventional consumers” (i.e., consumers who meet a conventional scoring model’s minimum criteria of having at least one account that has been opened for a period of at least six months or has account activity that has been reported in the previous six months) with credit accounts whose trended data might not be fully reported.

If more trended data were available, that would impact approximately 123 million of those consumers.

More specifically, our data scientists found that:

  • 39% (or 48 million) of those consumers would see significant (20-point or more) changes in their current credit score;
  • 20% of those consumers would see an improvement in their credit scores;
  • 19% would see a decline in their credit scores;
  • Credit score improvements generally occur for consumers whose credit scores are less than 650 (representing 1 in 6 of those consumers). And as consumers’ credit scores increase, the improvement of their credit scores also increases;
  • Decreases in credit scores mainly occur in consumers whose scores are 750 and above.

In other words, lower-scoring consumers would likely see score improvements and lenders would be able to segment higher-scoring consumers even more accurately.

These are compelling insights and we’ll socialize this information proactively among key decision-makers to encourage greater reporting across all asset classes. In turn, this will help model builders build better and better models that recognize this data to increase scoring accuracy for the benefit of both lenders and borrowers for greater scoring accuracy!

And by the way, check out a trended credit data overview on American Banker here.

This month’s newsletter has some additional information about the power of trended credit data. We’re also posting the results of our annual model validation, the first of which that digs into how trended credit data continues to deliver predictive performance lift. As I mentioned at the beginning of this article, lenders rightly demand transparency and we are leading the way!


Barrett Burns

CEO and President, VantageScore Solutions