Credit card charge-offs reached record levels late last year, and continue to challenge the financial services industry. To identify any factors beyond financial hardship that could be contributing to this rise, TransUnion completed in-depth analyses of charge-off losses and behaviors. Losses from early default charge-offs are a multi-billion-dollar issue and are increasingly plaguing prime and above portfolios. The analysis was based on four questions:
- Have charge-off losses from suspect behavior increased disproportionate to traditional, contractual credit card losses?
- If so, what types of behaviors are driving these suspect losses higher?
- Are these losses impacting specific issuers or consumer profiles more so than others?
- Are there industry responses proving effective in stemming the timed of these losses?
Losses from suspect behavior: Narrowing in on early defaults
We identified three major categories of irregular charge-offs: 1) Early defaults, where charge-offs occur within 12 months after origination 2) Dormant cards that have been open for more than a year and suddenly see activity and 3) Low utilizers, where charge-offs take place on accounts that are not inactive but have low utilization. All three categories of irregular behavior accounted for $776M in Q2 2024 losses, or 5% of industry charge-offs.
Interestingly, the analysis identified early default losses as the vast majority, comprising 73% of all Q2 2024 irregular losses. With the time from delinquency to charge-off at most issuers averaging six to seven months, these losses are particularly suspect and distinct from traditional bankcard charge-offs in which a consumer overextends her or himself, slowly builds a balance and then finds it impossible to make minimum payment due. These early default losses are characterized by rapid balance-building in the first month(s) of a card being opened, coupled with no or minimal payment activity.
Who’s feeling the pain of early defaults?
Losses from these early defaults have grown significantly between 2019 and 2023, up 44% from $350M to $504M, and in just six months to Q2 2024, increased by a further 12% to $564M.
Early default losses have long been endemic in non-prime portfolios: it’s not uncommon for consumers with distressed credit to quickly build balances and find themselves unable to manage a relatively modest credit line. This phenomenon has been—until recently—less common in prime and above credit tiers.
Attempting to understand the impact of these behaviors and losses from the standpoint of an originating risk management decision, the analyzed consumer credit scores dated from two years before the charge-off event or the time of card origination, whichever was earlier.
Knowing that the delinquency behavior preceding a charge-off event decreases a consumer credit score, these older credit scores provide a window into impacts based on consumer behavior well before any negative behaviors on the card in question.
The fact that losses associated to consumers with prime and better credit scores is sounding alarm bells in risk management offices at many major issuers is no surprise based on the data. While losses from these early defaults increased by 44% from 2019 to 2023, examining all cardholders, losses from prime and above consumers increased by 62%. These prime and above losses represented $175 million or 31% of total early default charge-offs in Q2 2024.
Formulating strategies to identify irregular charge-offs
As issuers grapple with compound challenges of heightened charge-off losses coupled with anemic demand for credit from the lowest risk credit tiers, managing these suspect behaviors while providing enough credit lines to attract good balances and top-of-wallet position becomes a more pronounced challenge. While commercially available and proprietary credit risk models continue to perform well in predicting credit risk, they were not built to detect fraudulent or irregular behaviors, which has sent issuers in search of strategies to limit exposure from the growing threat of early defaults.
Simultaneously, commercially available fraud tools abound, aimed at detecting third-party or synthetic fraud. While many are incredibly effective and finely tuned to nascent threats, they do not perform as well at detecting legitimate consumers applying for credit with no intent to repay. In most cases, risk managers need to find a third option—somewhere between a credit risk model and a traditional fraud tool—to detect signs of potential irregular behavior, both at the time of application and in existing portfolios.
These strategies, implemented and being evaluated by many issuers, range from commercially available models to complex decision trees involving scores and attributes. As part of this analysis, multiple treatment strategies were created and applied to portfolios to understand both the ability to reduce loss and how they impact a through-the-door population.
In the application of a commercially available score designed to predict early default risk, setting a score threshold only impacting 4% of applicants, risk managers could isolate cards with a charge-off rate of 17%. Layering on additional attributes allows issuers to refine performance, surpassing these impressive results.
Ultimately, models such as this, along with other tools, including off-the-shelf scores and predictive attributes, can be used by lenders to help mitigate the chances of bringing in accounts that have a high likelihood of displaying unexpected irregular fraud-like behaviors in the first place to reduce losses in the long term.
Carmen Williams is a senior advisor at TransUnion, responsible for researching and reporting on trends important to the company’s 5,000+ card issuer, retail bank and credit union customers.
