ABF Explained

Originator guide

Auto Loans and Leases

Draft for editorial review. Technical and final editorial approval are pending. Examples with invented numbers are labelled fictional.

Vehicle finance looks like one asset class from outside and behaves like several from inside. Before you put a pool in front of a capital provider, be able to state precisely what payment right you own, what claim you have on the vehicle, and how much of your projected cash flow depends on selling a car rather than collecting from a borrower.

What you actually own

On a retail instalment contract you hold a borrower payment obligation together with a security interest in the vehicle. The contract is the asset. The car is the fallback, and it only becomes cash after a recovery process that costs money and takes time.

Under a lease you hold contracted rentals for the lease term plus an interest in the vehicle at maturity. That residual interest is a separate exposure from the rentals. A lease book can pay every scheduled rental on time and still miss its projections because returned vehicles sold below the residual value carried on your books. Keep the two streams apart in your reporting rather than blending them into a single yield figure.

Loans and leases are not the same exposure

The practical consequence shows up in eligibility criteria and advance rates. A provider financing instalment contracts is lending against scheduled amortising payments with a depreciating fallback. A provider financing leases is lending against rentals plus an assumption about a future sale price that neither of you controls. Expect residual exposure to be treated more conservatively. Some providers exclude residual value from the borrowing base and advance only against the discounted rental stream, though that is a negotiated outcome rather than a fixed convention.

If you originate both, split the tape by product from the start. Mixed files where lease rentals and loan principal sit in the same balance column create avoidable rework and delay. The loan tape discipline of persistent identifiers, explicit balance-versus-flow fields and a documented field dictionary matters more here than in single-product books.

What one default actually costs

Severity, not default frequency, is where auto pools surprise people. Work through a single fictional contract in US dollars.

Exhibit 1

Severity on one defaulted contract

Fictional single-contract illustration in USD. Severity is measured against the balance outstanding at default, not against the pool.
LineBase recovery caseWeak recovery case
Balance at default20,00020,000
Gross sale recovery13,00010,000
Realisation cost2,0002,000
Net recovery11,0008,000
Loss9,00012,000
Severity on defaulted balance45 percent60 percent

The base case gives a 9,000 loss on a 20,000 balance, or 45 percent severity. Hold the recovery process constant and move gross proceeds from 13,000 to 10,000, and severity moves to 60 percent. Nothing about the borrower changed. Only the resale market and the cost of getting the vehicle to it changed.

Realisation cost is the line originators underestimate. It absorbs recovery agent fees, transport, storage, reconditioning and sale costs. Two thousand dollars of cost against a thirteen thousand dollar sale is not a rounding difference, and in the weak case it consumes a quarter of gross proceeds.

Severity is not a pool loss rate

The 45 percent above applies to one contract that defaulted. Multiplying it across a whole pool implies every contract defaults. Pool loss requires a default frequency assumption applied to a defined cohort over a defined period, then multiplied by severity. Keep the two inputs visible and separately sourced.

Where the pool splits

Auto books are rarely homogeneous, and providers will slice them until they find the segments that behave differently. Prepare the cuts yourself.

  • New versus used. Depreciation curves and resale liquidity differ, and borrower mix may differ too, though not in a fixed direction. Treat it as a cut to examine and report separately rather than presenting a blended severity.
  • Origination channel. Direct-to-consumer, dealer-sourced and marketplace-referred volume carry different fraud exposure and different control points. If dealers touch the application, expect dealer-level performance reporting to be requested.
  • Vintage. Cohort tracking through a static pool view shows how each origination month seasoned. Portfolio-level ratios in a growing book hide deterioration because unseasoned contracts keep enlarging the denominator.
  • Term and loan-to-value at origination. Longer terms extend the window during which the balance exceeds vehicle value. Providers will ask how much of the book sits in that position and for how long.

Fraud is an origination control problem

Synthetic identities, misstated income, straw purchasers and vehicle misrepresentation can present later as early payment defaults. Clustered defaults in the first few payments are a signal worth investigating, but they can equally reflect affordability failures, credit-model drift or operational errors such as mis-set first payment dates. Diagnose before concluding.

Document what you verify, how, and with what evidence retained. Identity verification method, income verification method and source, vehicle identification number validation against the collateral record, and dealer or channel monitoring all belong in a written policy that matches what your operations team actually does. A provider comparing the policy to a sample of files will find the gap quickly, and the diligence process is where that comparison happens.

Title, insurance and custody evidence

Whether your lien is properly recorded and enforceable is a question for counsel in each relevant jurisdiction, and it varies by state and by titling system. What you can control is the evidence: a complete record of title status per contract, an exception report showing contracts where title is pending or missing, an ageing of those exceptions, and a named owner for clearing them.

The same applies to physical damage insurance. A lapse does not change your lien; it removes the insurance proceeds you would otherwise rely on if the vehicle is damaged, destroyed or stolen. The security interest survives, but the collateral behind it may not. Track coverage status, document your force-placed process if you operate one, and be ready to show how many contracts currently sit without verified coverage.

Servicing and recovery operations

Repossession cost and timing depend on jurisdiction, procedure and the servicer you use, so do not model a single national assumption. Model by jurisdiction where your volume concentrates and hold the assumption clearly labelled.

The servicer relationship also determines whether your severity data is trustworthy. Payment allocation between principal, interest and fees must be consistent, deferral and extension policies must be written and enforced, and any change in those policies must be dated so that a reader can tell whether performance improved or was simply re-labelled. Extensions granted freely will suppress delinquency without improving cash. Providers know this and will ask for extension volume alongside delinquency. More on operational expectations sits in the servicing requirements guide.

Choosing a financing route

Auto originators generally choose between three routes, and the choice turns on the evidence you can produce rather than on preference. Availability depends on your scale, track record and counterparties, and none of these structures is open to every originator.

Exhibit 2

Financing routes for an auto book

Comparison of structural routes. Availability and terms depend on the specific originator, pool and counterparties.
RouteWhat it fundsWhat the evidence has to support
Warehouse facilityRevolving advances against eligible contracts you retainContract-level eligibility testing, title and insurance status per contract, reliable borrowing-base reporting
Whole-loan sale or forward flowSale of originated contracts, retaining or releasing servicingConsistent origination to agreed criteria, clean transfer documentation, repurchase and servicing terms you can perform
Term ABSRefinance of a seasoned pool through issued notesMulti-vintage loss and severity history, granular static-pool data, documentation and disclosure work scoped with counsel

Eligibility drives the split more than anything else. Contracts with pending title, unverified insurance or missing channel data tend to fall out of a borrowing base and sit on your own balance sheet, which pushes some originators toward sale structures where the buyer's criteria are applied at delivery instead. Where a rating is contemplated, methodology treatment of residual exposure, used-vehicle severity and servicer capability is asset and deal specific and requires review with the agency; do not assume a treatment from another issuer's transaction carries across.

Mechanics for each route sit in the warehouse, whole-loan sale, forward flow and term ABS material.

What to have ready before diligence

Exhibit 3

Auto pool readiness checklist

Prepare these before the first provider conversation. Each item needs a named owner and a stated as-of date.
Evidence itemWhat good looks likeCommon gap
Product splitLoans and leases on separate tapes with residual assumptions statedRentals and principal combined in one balance field
Vintage performanceCohort tables by origination month showing loss development by ageOnly portfolio-level ratios that mask growth effects
Severity buildGross recovery, realisation cost and net recovery held as separate fieldsNet figures only, with cost buried in an adjustment
Title statusPer-contract status with an aged exception reportAggregate percentage with no exception ageing
Insurance trackingCoverage status per contract and documented force-placed processPolicy described but no current status file
Extension policyWritten policy, dated changes and monthly extension volumePractice differs from policy with no volume reporting
Channel reportingPerformance by dealer or referral sourceChannel field absent from the tape entirely

Start with the product split and the vintage tables. Building them tends to surface the data problems you would otherwise discover mid-review, and until the tape reconciles and the cohorts are clean, no discussion of structure or pricing rests on anything solid.