Embedded credit meets its cost of funds

Credit offered at the moment of purchase looks like a feature and a balance sheet. That distinction did not matter much when money cost nothing. It is now the whole business.

The category is no longer marginal. Research published by a regional Federal Reserve bank describes buy-now-pay-later volume growing at roughly twenty percent a year since 2021, reaching a scale that is meaningful in absolute terms while remaining small next to card spending.

That framing matters for interpreting the sector. It is large enough to be a real consumer credit channel and small enough that its growth rate, rather than its size, is what draws attention from supervisors and from the institutions funding it.

The economics are a spread. The provider borrows to fund receivables, earns a merchant fee and sometimes interest, and loses money on the loans that do not repay. When borrowing was nearly free the spread survived considerable credit error. It no longer does.

Higher funding costs compress that arithmetic from both directions. Every basis point on the warehouse facility comes off the margin, and the same rate environment that raised the funding cost also raised the pressure on the households borrowing forty dollars at a checkout.

The response has been a narrowing of who gets approved and a shortening of what gets offered, which is a rational underwriting decision with an awkward marketing consequence. A product sold on universal availability becomes a product with a decline rate.

Merchants have their own reassessment underway. A checkout financing option that lifts conversion is worth its fee, but the fee is now negotiated against a measured lift rather than an assumed one, and some categories have discovered the lift was smaller than the pitch.

The competitive picture has also changed shape. Card issuers added instalment features to existing accounts, which removed the novelty advantage, and a consumer who can convert a purchase to payments inside a card they already hold has less reason to open something new.

Business-to-business embedded lending has held up better, for a structural reason. Underwriting a merchant inside a software platform that can see its transaction history and, in some cases, hold its receivables is a genuinely better credit position than underwriting a consumer at a checkout.

That is where the durable version of the model appears to sit. The lender's advantage is data and control rather than distribution, and it survives a rate cycle because it was never dependent on funding being free.

The consumer version will persist at a smaller size and a higher price, which is what a credit product looks like when it has to cover its own cost of money. The growth years were partly a product decision and substantially a monetary one.