Unsecured Consumer Loans

With no vehicle, property or equipment behind the loan, there is no asset to sell after default. That single fact reshapes how a capital provider reads your book: the underwriting decision, the verification evidence behind it and the servicing that follows carry the credit outcome on their own.

Underwriting evidence does the work collateral would

In a secured book, weak underwriting can be partially rescued by the asset. Here it cannot. Post-charge-off recoveries on unsecured consumer paper vary widely by product, borrower segment, jurisdiction and collection strategy, and some books recover materially more than others. Measure your own recovery timing and amounts by cohort rather than importing a general assumption in either direction. Your loan files document the rights you hold and the basis on which you lent; they evidence the claim, they do not stand in for security.

That pushes attention onto three things a provider will test hard: whether you assessed the borrower's ability to repay on evidence rather than assertion, whether you can detect fraud and synthetic identities before funding, and whether your servicing preserves the value of a performing loan when a borrower's circumstances change.

Your own legal obligations depend on your licences, jurisdictions and products, so ask counsel to establish which rules apply to you. The analytical expectation is portable regardless: providers want documented repayment capacity assessment, and they will sample files to see whether the documentation exists.

Two loss numbers that are not comparable

Most confusion in unsecured consumer conversations comes from mixing two measures that share neither denominator nor time period. Work through a fictional cohort in US dollars.

Assume a vintage originated 5m of principal and has now fully run off, so its performance is complete. Gross charge-offs over the life of that cohort totalled 400,000, with 50,000 of recoveries, giving 350,000 of net loss. Against 5m of original principal, that is a 7 percent lifetime net loss rate on a completed vintage. The completion assumption is doing real work here: an unfinished cohort cannot produce a lifetime number.

Now take a separate reporting date. Across a current outstanding balance of 2m, 100,000 is delinquent. That is 5 percent, and it is a point-in-time delinquency ratio on current balance.

Vintage discipline and seasoning

Unsecured consumer losses emerge on a curve. Cohorts typically show low losses in the earliest months, a rising period as delinquency converts to charge-off, then a flattening tail. A static pool view, grouping loans by origination month and tracking each cohort by age, is the clearest way to see whether the curve is shifting, and it pairs well with roll-rate and cure analysis rather than replacing them.

Build the tables so that any reader can compare cohorts at the same age. Month twelve of the 2024 cohorts against month twelve of the 2025 cohorts tells you something. Comparing a mature cohort against a three-month-old one tells you nothing, and presenting that comparison as favourable damages credibility fast.

Seasoning also affects what a provider will finance. Very new originations carry the least performance evidence, and some eligibility criteria require a minimum number of payments made before a loan counts toward the borrowing base. Where that condition applies, it changes your equity need directly, because loans you have funded but cannot yet include sit entirely on your own balance sheet. Check whether it appears in the specific term sheet in front of you rather than assuming it.

Underwriting evidence that gets tested

A credit policy document is the starting point, not the answer. Diligence compares the written policy against a sample of actual files, and the gaps are what get discussed.

Prepare for questions on each of the following, with file-level evidence rather than descriptions.

• Income verification. Which method, applied to which applicant segments, retained in what form. Whether verification is legally required depends on your product, licences and jurisdictions, so ask counsel; commercially, providers will still test what you did. If you verify income for some applications and not others, state the rule and the split.
• Affordability assessment. How you calculate existing obligations and disposable income, what data source you use, and what you do when the applicant's stated position conflicts with the data.
• Fraud and identity controls. Detection at application, response when a flag fires, and measured outcomes such as first-payment default rates by channel.
• Exception approvals. How many loans were approved outside policy, by whom, and how those exceptions performed relative to policy-compliant loans. This last comparison is unusually revealing and many originators have never run it.
• Policy change history. Dated records of every material change to score cut-offs, verification rules or pricing tiers. Without dates, cohort comparisons cannot be interpreted.

Prepayment, selection and the shape of the book

Prepayment can be adverse if the borrowers refinancing away are disproportionately your better credits, leaving a residual pool skewed toward weaker performers and a rising loss rate on a shrinking balance. That is a risk to test in your own data, not a rule. Prepayment can also be concentrated in higher-rate tiers with the most to gain from refinancing, or driven by product features rather than credit quality.

Track prepayment by cohort and by credit tier, not just in aggregate, and check whether the fastest-paying tiers are your strongest or weakest. If composition is shifting month to month, any projection built on origination mix will drift.

Selection risk cuts the other way when you sell. If you sell your strongest loans through a whole-loan sale or deliver them under a forward flow agreement, the retained pool is by definition the residue. A provider financing that retained pool will price the selection, so decide deliberately which loans leave and be able to explain the rule you applied.

Yield is not net return

A headline coupon says nothing about what reaches equity. The path from contractual yield to net return runs through servicing cost, funding cost, fees, dilution from refunds or adjustments, prepayment and credit losses.

Build the bridge explicitly and hold each step separately so that a reader can test any assumption. When you present returns, state whether the figure is a single-period ratio or a dated cash-flow calculation. Those are different measures, and describing a simple ratio as an internal rate of return overstates what the arithmetic supports.

Choosing a financing route

The absence of collateral pushes the structural question onto data quality and servicing. What you can evidence determines which routes are realistically open, and none of them is available to every originator.

Eligibility is where most unsecured books lose capacity. Loans with incomplete verification evidence, undated policy exceptions or unclear consent to transfer tend to fall out of a borrowing base or be excluded from a sale population, and the excluded balance needs another funding source. Where a rating is contemplated, methodology treatment of vintage depth, verification practice, prepayment behaviour and servicer capability is asset and deal specific and has to be reviewed with the agency on your actual data.

The warehouse, whole-loan sale, forward flow and term ABS material covers the mechanics of each.

What to prepare before diligence

Vintage tables and basis labelling come first. Without them your other numbers cannot be interpreted, and with them you will find your own reporting inconsistencies before a provider does. Everything else in this list is easier once those two are in place.