The Biggest Consumer Debt Story Lives Between the Data Sources

Four reports landed within weeks of each other this spring: TransUnion's April 2026 Credit Industry Snapshot, the Federal Reserve Bank of New York's Q1 2026 Household Debt and Credit Report, Equifax's April 2026 Consumer Credit Trends Report, and Spinwheel's analysis of $2.9 billion in consumer debt across more than 20,000 borrowers.

They don't contradict each other. All 4 sources show balances rising while traditional delinquency signals remain mostly steady. Instead, they illuminate different parts of the same picture — and the gaps between what each can see may be where the most important questions about consumer financial health are sitting right now.

Understanding Different Data Sources: Bureau Data vs. Consumer Permissioned

Before comparing what these reports show, it's worth being precise about what each is designed to see. While reports from a federal agency rely on broad numbers, the reports from TransUnion, Equifax, and Spinwheel have fundamentally different sources of data.

Bureau Data Is Furnisher-Defined 

TransUnion and Equifax receive data from lenders, servicers, and financial institutions that voluntarily report to them. That model produces broad, consistent, and longitudinal coverage across hundreds of millions of consumers, updated regularly, with decades of historical depth. 

When TransUnion tracks serious delinquency rates across every major lending category, or Equifax reports utilization trends across 593 million bankcard accounts, those signals are only possible at that scale and consistency. No other data source replicates it.

But, furnisher participation is uneven. BNPL obligations — a market the Richmond Fed estimated at $70 billion in transaction value in 2025, growing roughly 20% annually — remain largely absent from credit records. Rent payments, often a household's largest monthly obligation, are reported only when landlords opt into third-party reporting services. Paycheck advance products and some fintech installment loans report inconsistently or not at all. 

For thin-file and near-prime consumers especially, the gap between the bureau record and the actual obligation picture can be substantial.

Consumer-Permissioned Data Is Defined by Authorization 

Consumer-permissioned data works differently. Spinwheel's access is defined not by what furnishers choose to report, but by what consumers authorize — direct pulls from servicers, financial institutions, and account portals, reflecting actual outstanding balances at the moment of consent. 

The consumers in Spinwheel's dataset are people who have actively opted into sharing their financial data — which likely means they skew toward more financially engaged, digitally connected borrowers. 

That’s why I like to look at these reports side by side. They showcase complementary coverage models, not competing ones. Bureau data is broader in population but bounded by furnisher participation. Consumer-permissioned data is bounded by consumer consent but potentially deeper within a given consumer's actual obligation picture.

Where Real-Time Liability Data Adds a Different View

Beyond coverage, Spinwheel's data operates on a different time horizon. Bureau data captures what a consumer owed at statement close, with a typical reporting lag of 30 to 60 days. 

Consumer-permissioned data reflects what a consumer owes now. For a borrower in active paydown, those two figures can diverge meaningfully — and the direction of that divergence carries information that a snapshot doesn't. A few things are visible in Spinwheel's dataset that the macro sources don't easily surface for the engaged, multi-obligation population it captures.

The Debt Complexity Cliff

When consumers move from one debt type to two, median balances increase more than tenfold — from $2,755 to $28,250. Even excluding mortgages, balances are still 7.3 times higher. 

The NY Fed tracks balances by product category with precision but it doesn't track how the simultaneous presence of multiple debt types changes the aggregate risk profile of a given borrower. That requires a different unit of analysis — and it's one that matters most for exactly the kind of prime and near-prime consumer most represented in Spinwheel's data.

The Distribution of Unsecured Debt within the Creditworthy Population 

Spinwheel's data shows that borrowers with "good" credit scores (670-739) carry more credit card debt than any other score tier. TransUnion and Equifax confirm that higher scores correlate with lower delinquency — but where the unsecured burden is most concentrated among consumers who are still paying on time is less visible in the macro data. If heavy unsecured leverage sits in a credit score range of 670 to 799 range rather than below 620, the population most likely to show stress in a downturn may look different than traditional models assume.

The Utilization Paradox

Equifax's April data shows average bankcard utilization falling year-over-year even as absolute balances grew by nearly $40 billion. Spinwheel's borrower data reflects the same dynamic: consumers with larger card portfolios show lower utilization. 

The mechanical explanation is well-understood. The more interesting question is whether a declining utilization rate is increasingly decoupled from actual balance trajectory — and whether real-time balance data, tracked over time, might surface that divergence before it becomes a delinquency signal.

The Signal that No Source Captures Alone

Perhaps the most consequential question across all four reports is one none of them directly addresses: when financial pressure builds for a multi-obligation borrower, which debts get paid first?

The NY Fed's Q1 2026 data offered a relevant signal. Credit card balances fell by $25 billion in Q1 while auto loan balances rose by $18 billion in the same quarter. For borrowers carrying both products simultaneously — which Spinwheel's data suggests is common among the engaged, prime and near-prime population — that divergence raises a question bureau data alone can't answer: is that sequencing, or coincidence? 

The Federal Reserve's research on the 2023 student loan payment resumption documented exactly this kind of substitution behavior, where consumers maintained performance on one obligation by adjusting behavior elsewhere. That dynamic looks like stability at the account level right up until it doesn't.

A lender with visibility into both a borrower's bureau profile and their real-time liability picture across all obligations would be better positioned to ask that question — and potentially to see stress building before it surfaces as a missed payment.

The Federal Reserve Board's debt service ratio adds useful context here. Household debt payments stood at 11.3% of disposable income at the end of 2025, up from 9.1% in early 2021. The aggregate is still well below the pre-financial crisis peak of 15.8%, which is reassuring at the market level. But aggregate ratios smooth over the distribution considerably. Understanding which borrowers are sitting above that figure — and how far above — requires knowing what they actually owe across all obligations, not just what's been reported.

Bureau data tells you the market. Real-time liability data tells you the borrower. The two aren't interchangeable — but used together, they offer something neither can provide alone: a view of consumer financial health that's both broad enough to spot trends and granular enough to act on them.