We take a stand on transparency
What’s the biggest complaint both lenders and consumers have about credit scoring models? I’ll spare you the Google search: it’s a lack of transparency. You know, the old black box argument.
When credit scores first came into wide use, perhaps that made sense. Credit scores were used within a lending institution’s “back office” function of credit underwriting. It was an esoteric, three-digit number that only a handful of people needed or wanted to understand. Additionally, model developers did not see the value of educating consumers about how scores actually worked.
Nowadays, obviously, that has changed. Credit scores are widely used for marketing credit products, making credit decisions, understanding month-to-month fluctuations of risk in a lender’s portfolio of loans, making loan purchases by secondary market investors, as well as empowering consumers to improve their financial habits.
Across this lifecycle, understanding credit scores and how they are built — their performance and impact — as well as what’s causing scores to fluctuate, is critical for all stakeholders. As model developers, it is our responsibility to provide more information and shed the practices of the past that led to distrust and misinterpretation.
Beginning with our launch 12 years ago, we’ve aimed to raise the bar on transparency and with the hope that the rest of the industry would catch up. We’ve never shied away from sharing how well our models are performing and we have publicly posted the results of all our validation studies since we launched. Stay tuned for the most recent results.
Another example of this effort of sharing a great deal of information about our models: we publish the tests we perform to ensure the models aren’t unfairly mistreating any particular consumer groups. This is known as “statistical bias” in data speak. Statistical bias concerns about a particular credit score model arise if the probability of default (PD) varies between various groups of consumers even though members of the respective groups receive the same score with the same model. Groups impacted may be minority groups or those with nontraditional credit histories.
This critical test was performed on our most recent model, VantageScore 4.0, and we examined the results to make sure our model isn’t treating these groups of consumers any differently than other groups or the broader population.
Taking that a step further, we’ve published those results in the “Testing Credit Scoring Models for Statistical Bias: Ushering in a New Era of Transparency” whitepaper and posted it on our website to make sure that the marketplace will know that our models show no statistical bias. And I encourage you to read the synopsis that we’ve included in our “Press Room” if you don’t have the time to read the full whitepaper.
Then there’s ReasonCode.org, another example of opening the black box for consumers seeking to understand what their score disclosure notices actually mean. Consumers used to receive cryptically worded codes in these notices that attempted to explain why a consumer’s score wasn’t higher. We wrote deeper explanations and tips for every reason code associated with all four of our models (and also translated that information into Spanish) and created a free search engine for consumers that connects them with this educational content.
These are just a few examples that come to mind. The point is that transparency isn’t a switch you turn on and off. We collaborate with many financial institutions as they are developing or updating their credit score educational content surrounding credit score content. And we’re proud to set the standard for the credit scoring model development in the industry.
Since summer is coming to a close, we thought we’d keep your summer reading list light with a few of our favorite articles from the past six months plus a “5 Questions with” Liz Weston, a financial columnist for NerdWallet. It’s been a busy year for us, and these articles are just the tip of the iceberg.
Looking forward to a productive fall.
CEO and President, VantageScore Solutions