Off the Lot: Time to reevaluate credit models, or repeat 2022

Pandemic and post-pandemic originations were initially some of the best most companies experienced. Borrowers were flush with stimulus funds and able to save money by staying at home and reducing commuting and other expenses associated with being required to be in an office location. They were able to buy consumer goods, cars and other items with the cash and the funds to power consumption. Although vehicles were in short supply and prices were high, the artificial income of stimulus funds made it more achievable for consumers to continue to buy cars During 2023, most companies looked back on 2022 originations with alarm as they experienced delinquencies and losses from this period. Those originations were some of the worst since the Great Recession, and looking at the reasons gave rise to some important questions. One concern was credit score inflation . Several factors led to credit score migrations. First, significant restrictions were placed on reporting public record data with the National Consumer Assistance Plan (NCAP), according to a February 2018 Consumer Financial Protection Bureau release. Prior to the implementation of NCAP in July 2017, 6% of consumers had civil judgments or tax liens. Subsequently, all civil judgments and about half of the tax liens were removed. The next factor contributing to potential score inflation was the cessation of reporting of medical collections under $500 in April 2023. The CFPB in an April 2023 report said 22.8 million people’s credit scores benefited by approximately 25 points in the first quarter after removal of debts under $500, and by 32 points with the removal of medical debt over $500. In addition to these more regulatory-inspired reporting changes, credit bureaus have individually opened a direct line with consumers, which may change consumers' credit reports. One example is Experian ’s Boost program, which allows consumers to voluntarily self-report certain utility bills to improve their credit scores. Consumers can also now open a checking account, which can also improve a consumer’s credit score. Did the migration lead to higher losses? After extensive review by credit-scoring entities, the performance of current credit scores continues to rank-order performance in line with those scores, with or without the benefits of the credit score migration. They continue to do so as before the pandemic. Jim Bander , Ph.D., a data scientist with Experian, reiterated to Auto Finance News his positions in a January 2024 research paper on credit scores. He reminded me that there is no “magic cutoff” for performance in credit scores. External factors can advantage or disadvantage borrowers, no matter where they sit along the credit spectrum, based on both individual economic and external macro-economic factors. It may seem obvious, but a loss of job, change in marital status, disability or other individual factors will cause changes in payment probabilities just as much as macro factors such as higher overall unemployment making it harder to replace a lost job, higher inflation putting pressure on household budgets, and higher interest rates raising overall costs. Any given credit score may perform differently depending on the service, Bander said. “The better the collections and customer engagement, the better a specific band of credit will perform,” he said. “Operating approach to servicing has an effect on both the willingness and payment priority, but the ability is determined at the time of origination and may change based on economic factors.” In auto lending, there are additional considerations that don’t affect other asset classes. If there is a significant decrease in used-vehicle prices, customers may be less willing to pay for their vehicle when they have significant negative equity and their vehicle fails. Many borrowers whose credit scores were high and may have gained ground during the last few years may be less willing as their vehicle values drop and they are “upside-down” in their vehicle and unable to trade. The buy now, pay later problem Another external factor that is currently not easily identified, is the rise and impact of buy now, pay later — or point-of-sale — loans that has grown in the last few years. The ability to finance items both online and at point of sale, as Walmart is now making accessible with Affirm , may change the ability of a customer to pay. The majority of these forms of credit do not currently report to credit bureaus, which makes it harder to determine actual debt-to-income ratios and the customer’s ability to pay. These loans were even referred to as “phantom debt” by Wells Fargo ’s Senior Economist Tim Quinlan in a CNBC interview earlier this year. The result of these changes makes it even more important that lenders take a very hard look at their models. Models created under prior conditions may need to be updated. Looking outside of the existing risk teams and accessing highly tuned models that may benefit from machine learning and AI are even more important today due to the changes in factors that could affect performance outcomes. Lenders should consider looking at operational and other data points used in underwriting to adapt to the current environment. Greater use and parsing of income data from consumer-permissioned data as well as alternative data analytics that could include rental, utility and payday loan data and gig economy job income may need to be included in more advanced models which weren’t used in prior years. Further, requiring the customer to grant access to this data may allow for greater access to parse actual income and expenses and should allow visibility into the BNPL payments being made by the consumer. The upside is that, with more advanced models and better operational data integrity, a lender may be able to remove friction in the approval process, require less customer-provided documentation and improve the reliability of external data provided and improve both customer and dealer satisfaction. What’s more, a customer could provide access to the data without collecting paperwork that might also not be legitimate. Another interesting and somewhat positive view of the scoring changes is that the existence of civil judgments and medical debt may never have indicated the payment probability of a debtor. There are many sources of model development. Depending on a company’s size and its own risk and data science teams, however, some lenders may not have equal levels of acumen, and larger institutions may need to look outside of the traditional attributes to accommodate the impact on the probability of payment of the current factors. In a time of increasing rates and lower margins, it’s going to be crucial to grow volumes and reduce losses and operating expenses. Fine-tuning models — and including attributes and data not currently analyzed in the decision process — can help lenders avoid repeating 2022 and mitigate against the increasing popularity of BNPL loans. Until next time. Bruce Newmark , formally chief adviser of Vroom and United Auto Credit Corp. , is now retired and a consultant to the industry. Newmark has more than 45 years of experience in automotive retail and finance. Before joining Vroom, Newmark’s tenure spans leadership roles at Fimanco, Inc. , Ford Motor Co. , Hankey Automotive Group and Westlake Financial .

Off the Lot: Time to reevaluate credit models, or repeat 2022

Buy now, pay later is a growing problem for lenders

Credit tightening contributes to subprime market shrink

Troubled EV maker Fisker said to be in talks with Nissan

Related Posts

10 auto finance executives to watch in 2026

RV dealers say rate cuts needed as cash flow tightens

Prestige Financial Services stops originations

Troubled EV maker Fisker said to be in talks with Nissan

Stay Informed with Our Newsletters

The Roadmap Podcast

SPONSORED

Driving Innovation in Automotive Refunds: How Onbe and Allied Solutions Are Modernizing the Payment Experience

Navigating the Margin Squeeze: EVs, Tariffs & Pricing — A Q&A with Earnix

Enhancing credit approaches to boost conversion and dealer loyalty

Welcome Back!

Retrieve your password