DID YOU KNOW: Credit Score Models Today
Credit score models today: What you need to know
(This article originally appeared in American Banker)
Look up “credit scoring model” and you’ll likely come across a definition that’s along the lines of “a statistically derived algorithm that provides an estimate of the likelihood that a borrower will default on a loan.” That is true, but is that the whole story? Today, credit scores are ubiquitous and impact millions of consumers in getting approved for a loan, the interest rate on that loan and getting a cell phone, among other things.
And thanks to the use of better data and predictive analytics, today’s credit scoring models are changing the game and enabling lenders to fine-tune credit decisions, better target the best borrowers, and bring into the fold millions of “credit invisibles” (i.e., consumers who are creditworthy but cannot be scored using conventional credit score models). In fact, credit scoring models today, and consumers’ access to and awareness of them, are a far cry from when the general-purpose FICO score debuted three decades ago.
Unbeknownst to many of today’s 250 million-plus credit-eligible consumers, people have more than one credit score. Dozens more actually. There are new credit scoring companies and differing methods of calculating credit scores. And lenders will often use more, than one type of credit scoring model in their decision making. Furthermore, lenders also continually test and update their own credit scoring models customized specifically for their own portfolios. All of these factors feed into the complexity of the credit score universe.
In general, credit score models typically fall into one of three buckets — third-party/generic models, industry-specific models and custom models.
Third-party/generic models. A third-party or generic model, which is based primarily on the information in credit reports from the three major consumer credit bureaus (Equifax, Experian, TransUnion), is designed to assess credit risk on a wide range of products. More specifically, it predicts the likelihood that a consumer will default, or go at least 90 days past due, on a credit obligation during the subsequent 24 months.
Industry-specific models. When a credit score model is designed to predict performance on a specific type of credit obligation, such as an automobile loan, they are often known as “industry” models.
Internal/custom models. Some lenders obtain credit reports from consumer reporting agencies and then develop scoring models in-house specifically for their portfolios to better understand the opportunities and risks associated with their customer base. These are commonly known as “custom” models.
To add to the complexity, as noted earlier, some lenders will use more than one type of model. More sophisticated lenders, like large banks, may use a combination of third-party and custom-built models, or perhaps more than one third-party model, in their decision-making.
“There are different ways lenders are leveraging scores generated by generic credit models, even using more than one generic score as ‘overlays’ or as ‘waterfalls.’ And they also are leveraging generic scores as part of the broader custom decision-making engine,” said Dr. Emre Sahingur, senior vice president, predictive analytics, research and product management for VantageScore Solutions.
For example, some lenders incorporate third-party models as an input into their own custom models in addition to many other variables, or as an “overlay” where multiple scores are used together to provide the best prediction. Others may opt for a “waterfall” strategy where a primary and a secondary model are available and the secondary model is used only in the event the borrower does not have a primary score.
On the other hand, a smaller lender may rely almost entirely on third-party scores for their decision making.
And today, lenders have a choice of generic credit score model providers, not just FICO. VantageScore, for instance, came on the scene in 2006, shaking up the industry by introducing competition and innovation. This spurring of innovation has proven beneficial to both the industry and to consumers, driving more predictive, inclusive credit scoring models with the ability to score more than 40 million consumers who are currently overlooked by conventional models without having to engage a supplementary add-on model.
OTHER CREDIT SCORE USES
While the primary function of credit scores is to help assess credit risk in a lending decision, the uses of scores are actually quite varied. Consider the following:
Marketing: Lenders use credit scores as part of their pre-screening process to determine which consumers to target and what products to offer. Consumer communication and the products offered can even be tailored to how a specific group of consumers use credit.
Pricing: A higher credit score may help a consumer qualify for a better interest rate, and vice versa.
Account maintenance: Lenders may use scores in their credit management processes, or to evaluate whether to cross-sell new products to existing customers. Furthermore, lenders may periodically review existing customers’ credit scores and reports to monitor for changes in risk profiles and determine whether any proactive actions may be necessary.
Secondary markets: Lenders disclose the credit scores for the pools of loans they sell to investors to supplement the underlying risk characteristics for secondary market pricing and/or investment purposes.
Credit scoring models have a proven track record of supporting these activities, ultimately helping lenders, investors and consumers alike. But they are not perfect.
“Models are based on the premise that the past will be predictive of the future. As the environment changes, credit scoring models built on historical data may become less reliable. It might be that you’re in a different economic environment. It might be that consumer attitudes toward credit and their usage patterns may have shifted, there may be new types of products [on the market], there may be new rules and regulations,” said Dr. Sahingur. “A lot of things may drive changes in how consumers interact with credit and how they perform, which is why it is important for credit score models to continually be tested and improve with the times, as well as tap into more advanced technology and methodology to score more people, more accurately and predictively. ”
The financial services industry has no doubt seen significant advancements in credit score modeling over the years and the pace of change shows little signs of slowing.
THE LATEST CREDIT SCORE INNOVATIONS
With the availability of vast amounts of data and the emergence of advanced technologies, a new era of credit scoring has emerged and is further transforming the credit modeling landscape.
Among the game-changers is trended credit data. Trended credit data delivers a refined view of a consumer’s financial health over time and enables lenders to better manage risks. This data can be an early indicator should a consumer’s credit worthiness improve or deteriorate in reference to their historical performance versus just “point-in-time snapshot” data, which is what traditional credit score models employ.
Using a model that incorporates trended credit data also has been particularly helpful when trying to “tease out” riskier consumers in the prime and superprime categories.
Also gaining greater attention is alternative credit data, which relates to activity that may not be captured on a traditional credit report — like rent, phone bills, and other forms of debt such as payday loans — and can help lenders gain a more comprehensive view of a consumer’s creditworthiness.
VantageScore paved the way for many innovations to move from theoretical to mainstream, and as a result is enjoying increased adoption among both financial and non-financial institutions.
According to the 2019 Market Study Report, conducted by Oliver Wyman, a global management consultancy firm, 12.3 billion VantageScore credit scores were used from July 2018 – June 2019. Adoption among non-financial institutions (i.e., tenant screening, utilities companies, etc.) was also considerable, and several investment firms also used VantageScore credit scores as part of their investment decision-making process.
Meanwhile, many websites such as CreditKarma, CapitalOne CreditWise, LendingTree and others are partnering with VantageScore to offer consumers the ability to check their credit scores free of charge, as well as provide educational tools and advice. In fact, more than 3 billion VantageScore credit scores were directly provided to consumers. This means that today’s consumers are more equipped than ever before to make informed credit decisions.
In today’s ever-changing market, lenders cannot afford to rest on their laurels. Gaining an understanding of complex scoring models and keeping pace with solutions that set a new standard for predictive performance and modeling innovation is imperative for lenders looking to maintain a competitive edge.