Are credit scores inflated? That’s the take-away from a recent headline based on an analysis from Moody’s Analytics.
Lenders, investors and regulators would be wise to read and internalize the Moody’s report, however they should also consider a number of underlying factors.
Firstly, the notion that average credit scores are rising should come to no one’s surprise. If scores weren’t improving, there would be something wrong with the models.
Credit scoring models are based on the premise that past credit behaviors are indicative of the likelihood of a future default. The models include factors related to a consumer’s use of credit, payment history, level of indebtedness and availability of credit, and credit seeking behavior, among others.
With the Great Recession in the rearview mirror, many of the negative marks such as delinquencies, foreclosures and even bankruptcies, are farther back in consumers’ credit history with a smaller impact on credit scores, or have dropped off the credit files entirely.
Unemployment is at a 50-year low, home prices have, on average, recovered after hitting rock bottom, and cost of borrowing remains low; all positively contributing to a strong credit performance. At the same time, post-crisis, consumers have been more prudent in their use of credit, and credit underwriting standards have improved (the ability to repay requirements with most mortgages is one driving force of this trend). Collectively, these all point to higher credit scores.
In short, credit scores are pro-cyclical in nature. They are derived from historical data and respond to it. Credit scoring models are designed to reward a consumer for recent positive credit behavior. With the passage of time, recent behavior overshadows a past financial misstep or challenge. Conversely, as the credit cycle eventually shifts downward, credit scores will respond and begin to reflect the changes in consumers’ behaviors.
Secondly, it is imperative to remember that the objective of a credit scoring model is to “rank order” the total pool of scoreable consumers based on their risk. The score is only a relative measure of risk. In other words, we are individually compared with the total scoreable population. The meaning of the score, or the actual level of risk, changes in response to the overall level risk in the system.
For example, the default rates experienced for loans with a specific credit score would have been significantly higher in 2008 compared to what it was in 2018, reflective of the very different points in the credit cycle and the level of risk in the system. Well-built models maintain their ability to “rank order”, with lower credit scores resulting in higher levels of risk compared to higher scores. However, these models are not designed to capture the changes in the actual risk levels.
Therefore, lenders must be diligent, anticipate and react to changes in risk levels, adjusting credit score cut-offs as well the broader underwriting criteria (Debt-to-Income, Loan-to-Value, etc.) to manage a portfolio’s expected performance under changing economic conditions. Lenders should validate models regularly and closely monitor performance of their portfolios.
The OCC provides extensive supervisory guidance on model validation best practices. In particular, the OCC notes that validations of credit scoring models should occur at least annually if not more often, and that, “where models and model output have a material impact on business decisions, including decisions related to risk management and capital and liquidity planning, and where model failure would have a particularly harmful impact on a bank’s financial condition, a bank’s model risk management framework should be more extensive and rigorous.” If volatility were to increase rapidly, lenders will consider model validations more frequently than annually.
Importantly, this work must be done for models developed both internally and by third parties, such as models built by VantageScore.
There always will be some bad apples in the bunch when it comes to risk management. A consumer lender who underwrites solely on credit scores will and should be exposed because credit scores are not intended to be a substitute for prudent underwriting criteria. Scores should only be used as an integrated component of sound underwriting standards.
The bottom line is that yes, credit scores are improving alongside the broader economy and that is to be expected. Lenders must refrain from complacency or loosening standards too much, even as volatility remains low, because we all witnessed what can happen should stress be introduced into the economy.
CEO and president, VantageScore Solutions
(1) Supervisory guidance on model risk management, OCC 2011-12, 2011