It's called « paying in four ». With each purchase, one quarter settled at once, the rest over a few weeks, with no apparent interest. Buy-now-pay-later has become the most popular credit of the decade. But it has one quirk : it appears almost nowhere. Not in your credit file, not in household-debt statistics, not before the eyes of the next lender. And when information vanishes, the risk does not : it merely becomes invisible, and therefore mispriced.
1 The painless payment
Before it is a blind spot, BNPL is a payment revolution.
Credit that doesn't say its name
Paying in four, with no apparent pain
« Buy now, pay later » (BNPL) splits a purchase into four instalments : a quarter paid now, the rest over six to eight weeks, with no interest visible to the buyer. The model is funded by the merchant, who pays a commission in exchange for extra sales. Born at the checkout, on the phone, it has spread from sneakers to electronics, then to everyday emergencies, all the way to groceries.
2025 flow (US)
≈ $70bn
growing about 20 % a year since 2021, yet barely 1.1 % of card spending (Richmond Fed, Feb 2026).
Groceries
29 %
of users rely on it for groceries in 2026, double two years earlier (LendingTree).
The first paradox
Huge in flow, BNPL is tiny in stock : because it is repaid in six weeks, the money owed at any moment is only about $3 billion, against $1,230 billion for credit cards, four hundred times less. The danger, then, is not the amount. It is what this debt steals from view.
2 The profit machine
If this debt is everywhere, it is because it enriches every link in the chain.
Three beneficiaries, one engine
①
The providers (Klarna, Affirm, Afterpay). Their revenue comes first from the merchant commission, often 4 to 6 % of the purchase, far more than a card's 2 to 3 %. Add late fees, interest on longer plans, and increasingly advertising : Klarna is turning its app into a shopping window. The model is lucrative per unit ; but credit losses and the race for growth long kept several players in the red — Klarna still lost money in 2025. Profitable in theory, not always in practice.
②
The merchants. They pay more than for a card, but gain handsomely : BNPL lifts conversion (20 to 30 %) and the average basket (20 to 40 %), draws a younger clientele, and above all the merchant is paid upfront while the provider bears the default risk. More sales, less risk : the commission is quickly recouped.
③
The networks (Visa, Mastercard). They first feared being bypassed, then turned instalments into a revenue stream : each instalment is often drawn from a debit card, earning them an interchange fee ; they launched their own instalment offers (Visa Installments, Mastercard Installments) ; and BNPL-branded cards (Klarna, Affirm) run on their rails. Whoever wins the BNPL race, the network takes its toll.
No one to hit the brakes
Here is the quiet engine of the story : everyone gains from pushing instalment credit, and no one in the chain has any interest in shining a light on it. That is precisely what let its blind spot thrive.
3 A debt that shows up nowhere
BNPL grew up in a blind spot of the credit system.
Off radar
No inquiry, no reporting
Taking out a pay-in-four loan almost never triggers a « hard » credit inquiry, and these loans are historically not reported to the bureaus (Equifax, Experian, TransUnion). As a result, the purchase does not appear in your file. Three reasons for this blindness.
Why the invisibility
①
Duration. Six to eight weeks : too short for systems built for the revolving credit of cards, and repaying so brief a loan could even lower the score by shortening the average age of credit.
②
No incentive. Providers had no reason to report ; doing so « could scare customers away », as one industry lawyer puts it. No one was keen to light the room.
③
Format. Each purchase forms a separate line ; a string of small lines that a classic model would misread as instability.
Phantom debt
Regulators gave it a name : phantom debt. A lender extending you credit cannot see that you already carry five instalment loans across three platforms. The debt exists, the commitment is real ; only the system that should see them stays blind.
4 The invisible stack
Invisibility takes on its full meaning when you stack.
Loan stacking
Borrowing at several windows that ignore each other
According to the CFPB (January 2025 report, 2022 data), 63 % of borrowers took out several simultaneous BNPL loans during the year, and 33 % from different providers that cannot see one another. Nearly two-thirds of loans go to subprime profiles ; among 18-24-year-olds, BNPL makes up 28 % of unsecured debt. And it is sliding toward essentials : groceries, sometimes rent.
Simultaneous loans
63 %
of borrowers stack them ; 33 % across several providers (CFPB, 2022).
Below prime
≈ 2/3
of loans go to subprime or deep-subprime borrowers, with a 78 % approval rate (CFPB).
More than half of users say they « couldn't make ends meet » without BNPL. And a troubling signal appears : in 2026, 47 % say they paid late at least once in the year, up from 34 % in 2023, even as measured « hard » defaults fell (1.83 % of losses in 2023). This wide gap between felt lateness and recorded default is not reassuring : it says we are measuring badly.
The symptom in the gap
When lateness climbs while recorded default recedes, it is not a sign that all is well. It is the sign of a fog : the reality of stress is shifting into the zone the official figures do not cover.
5 Lending blind
The real risk is not the hidden amount, it is the decision made without seeing it.
Information asymmetry
Models scoring an incomplete snapshot
Risk models, even the most advanced, price credit from bureau data : yet BNPL is missing from it. A lender therefore underestimates the borrower's true leverage. The blindness rises a notch at the macro level : neither the New York Fed's household-debt dashboard nor the Federal Reserve's G.19 series captures BNPL — the latter says so explicitly. The central bank thus steers household debt with a blind spot.
The nuance that separates analysis from panic
The hidden balance is tiny : about $3 billion, four hundred times less than card debt. BNPL therefore does not distort the level of household leverage. The Richmond Fed (February 2026) indeed judges the systemic risk « limited at present », with no proven evidence of contagion. The problem is not a mountain of concealed debt ; it is a myriad of credit decisions made without seeing, and a measuring instrument gone short-sighted.
Small in volume, large in blindness
The blind spot does not distort the mass ; it distorts judgment, one loan after another. It is a problem of vision, not of weight. And a system that cannot see prices badly : the danger sits there, in the price of risk, not its amount.
6 Where the risk goes
Risk never evaporates. It changes hands.
The plumbing
BNPL doesn't keep its loans, it resells them
Providers fund themselves by packaging their receivables : securitization (ABS) and massive sales to private-credit funds and insurers. Affirm has become a regular issuer of asset-backed bonds, keeping a first-loss slice ; huge funding lines link PayPal to KKR, Klarna to Nelnet, Affirm to managers like Sixth Street or PGIM. 2025 set issuance records. The risk thus migrates from visible balance sheets toward less-watched pockets.
Three signals to watch
①
S&P's warning. Securitizing BNPL is « fraught with difficulty » : the revolving structure means the rated collateral shifts under the investor's feet.
②
The Klarna case. Listed in September 2025, the buy-now-pay-later specialist quickly disappointed : loss provisions up 102 %, shares below their IPO price, a loss-making year. Before its listing, the group had sold $26 billion of loans to Nelnet, tidying up its balance sheet just as investors were examining it closely.
③
Migration into the opaque. The more risk leaves listed balance sheets for private credit and insurance, the less it is observed : the initial blindness extends into the funding.
Works until it doesn't
« If the economy turns, people will stop paying their pay-in-four loans, especially if it isn't reported to the bureaus », an analyst sums up. That is the very definition of a pro-cyclical risk. The first real crack is Klarna.
See also : securitization.
7 The paradox of light
What if making the debt visible triggered the very shock we wanted to avoid ?
The snake biting its tail
To measure is already to act
The fix seems obvious : turn on the light. In late 2025, FICO launched its first BNPL-inclusive scores, and the credit bureaus are beginning to incorporate the data. But buy-now-pay-later users have, on average, more fragile profiles than borrowers as a whole. Bringing them into scoring models could lower millions of ratings, restrict their access to credit, and slow the very demand BNPL had helped create.
FICO is reassuring : by its own work, 85 % of users would see their score move by fewer than ten points. Other analyses, however, suggest an impact tilted more to the downside, and it is precisely among the most fragile profiles that the risk concentrates.
There lies the paradox : making the phenomenon visible can change its trajectory.
The sector splits
①
Affirm reports (to Experian since April 2025) and argues for sector transparency.
②
Klarna and Afterpay refuse, fearing classic models would wrongly penalize « responsible » borrowers for short, frequent use.
③
Visible is not scored. Experian and TransUnion for now keep BNPL in separate sections, outside the « core » score : the light is still partial.
The demand reversal
BNPL inflates the average basket (by 20 to 40 % per studies), and part of those purchases is marginal : it would vanish if it tightened. Making the debt visible could therefore, in one gesture, lower scores
and cool consumption.
See also : aggregate demand.
8 Three scenarios
It remains to be seen how the story closes. Three trajectories, plus a compass.
Sharing the future
From soft normalization to an air pocket
Nothing is settled : the same phantom debt can dissolve smoothly or abruptly reveal its hidden face. It all depends on the cycle, the speed of disclosure, and the moment the regulator chooses.
Three scenarios
①
Gradual normalization. According to the Richmond Fed, the risk remains limited at present and missed payments stay contained. BNPL is gradually folded into credit scores ; risk is better assessed and absorbed without major disruption.
②
The air pocket. A consumer recession lifts arrears on a debt largely absent from traditional risk models. Securitization markets tighten, private credit takes the hit, and the demand BNPL had been supporting contracts in turn. A demand-destruction loop sets in.
③
The regulatory divergence. The United States softens its approach, the CFPB having dropped its rule in May 2025, while the United Kingdom (July 2026) and the European Union (November 2026) impose stricter affordability checks. Two regulatory frameworks, two trajectories.
For the European Union, an interest-free credit is a credit nonetheless. On 20 November 2026, the new Consumer Credit Directive (CCD2, Directive 2023/2225) will bring much of buy-now-pay-later into ordinary credit law. Affordability checks, disclosure duties and transparency requirements : the aim is not to ban BNPL, but to make visible what until now looked more like a means of payment than a debt.
The compass
For the investor : watch not the size of BNPL, but the moment hidden defaults become visible again, and the private credit that carries them starts to doubt. For the citizen : remember that a « painless » payment is still a debt, and that a debt you cannot see is the easiest to forget — until the day it reappears.
Reference : abbreviations & acronyms used (BNPL, ABS, CFPB, FICO, $bn).