Amlok
The idea in brief.
Auto finance contains two connected exposures: the borrower’s ability to pay and the lender’s recovery if payments stop. Strong collateral does not create monthly cash flow, and strong income does not eliminate depreciation or negative equity. This guide shows how an Amlok-supported workflow can connect the application, retail installment contract, income evidence, vehicle identity and lender repayment records while keeping each risk measure interpretable.
- Assess repayment capacity and collateral exposure separately, then review them together.
- Trace financed fees and rolled-in negative equity into the actual principal.
- Validate new and used vehicles, channels and origination vintages separately.
A practical implementation guide grounded in Amlok’s documented product scope and segment research. Market figures retain their original population and period. Worked examples are invented, reproducible scenarios. They do not represent customer results, underwriting cutoffs or claims of validated model performance.
The market and the decision.
Statistics Canada reported that median new-vehicle prices rose 61.5% and used-vehicle prices 82.2% between December 2019 and December 2024. These historical market measures are not the depreciation curve of a specific vehicle. They provide context for why a current, sourced collateral value matters. [1]
An auto lender needs more than a contract amount. Separate purchase price, taxes, optional products, down payment, trade-in allowance and remaining trade-in debt. A longer term can reduce the installment while leaving the borrower exposed for longer. Keep loan and lease structures, new and used vehicles, and dealer and direct channels distinct.
Vehicle prices changed materially
View figure data
| Measure | Value |
|---|---|
| Median new-vehicle price change | +61.5% |
| Median used-vehicle price change | +82.2% |
What Amlok brings to the workflow.
Insights can assemble the supported borrower and facility evidence into a frozen assessment: identity and income verification results, recurring commitments, contract terms and supplied collateral information. The agreed workflow should reconcile the financed amount and repayment schedule before presenting affordability or loan-to-value measures. [2] [3]
For the credit team, this means one evidence trail from the amount requested to the installment assessed and the vehicle valuation used. Amlok does not make an absent valuation appear complete. Asset-specific calculations require the relevant authorized source. Signal can then compare payment and financial changes against the approved baseline while the lender retains servicing and recovery decisions.
Connect the evidence that matters.
Agree the evidence perimeter before asking for an assessment. Each record needs a subject, source, observation date, currency where applicable, and coverage window. Resolve conflicting statements explicitly; the most recently uploaded file is not necessarily the most recent financial position. [3]
| Evidence | What it supports | Review requirement |
|---|---|---|
| Application and borrower evidence | Applicant identity, supported income and existing commitments | Separate verified employment from deposits that merely resemble payroll. |
| Vehicle identity and invoice | VIN, new/used status, price and financed components | Resolve identity mismatches; do not substitute a similar vehicle. |
| Trade-in and payoff statement | Allowance, remaining debt and rolled-in negative equity | Reconcile payoff date and the actual amount added to the new facility. |
| Retail installment contract | Principal, rate, term, installment and due dates | Recompute the schedule using the contract’s conventions. |
| Dated valuation and repayment ledger | Collateral exposure, arrears and realized recovery | Retain valuation basis, source, timing and net recovery costs. |
Measures your team can review.
Use the definitions below as an assessment specification. Your institution selects policy thresholds and permitted adjustments. These measures are useful only when their inputs, time periods and contractual meanings align. [4]
- Payment-to-income: proposed auto installment divided by supported recurring income using a declared gross or net convention.
- Total debt burden: include other scheduled debt, not just the vehicle payment.
- Loan-to-value: financed principal divided by a dated vehicle value; show rolled-in negative equity separately.
- Arrears migration: distinguish first-payment default, missed installments, cures and repossession by months on book.
- Net loss severity: unrecovered principal plus defined eligible costs, less realized recoveries, divided by exposure at default.
Worked example: affordable payment, thinner collateral.
An illustrative CAD 40,000 purchase includes CAD 2,000 financed fees, CAD 4,000 rolled-in negative equity and a CAD 5,000 down payment. Principal is CAD 41,000. At a supplied CAD 38,000 vehicle value, origination loan-to-value is 107.9%.
Later, suppose exposure at default is CAD 34,000, gross sale proceeds CAD 27,000 and recovery costs CAD 2,000. Net recovery is CAD 25,000 and loss CAD 9,000, or 26.5% of exposure. These assumed figures illustrate arithmetic only; they do not estimate a market recovery rate or authorize a repossession.
| Component | Amount | Interpretation |
|---|---|---|
| Purchase price | 40,000 | Vehicle purchase |
| Fees + prior negative equity | 6,000 | Additional financed exposure |
| Down payment | −5,000 | Reduces principal |
| Financed principal | 41,000 | 40,000 + 6,000 − 5,000 |
| Principal / supplied value | 107.9% | 41,000 / 38,000 |
| Net recovery | 25,000 | 27,000 sale − 2,000 costs |
| Illustrative loss severity | 26.5% | (34,000 − 25,000) / 34,000 |
Keep the original decision in view.
Set the baseline at the approved facility terms and supported borrower position. Monitor payment records and permitted financial evidence without assuming the original valuation remains current. If a new valuation is supplied, retain both its date and methodology so a price-source change is not confused with a real value movement.
A missed installment should be reconciled to the lender ledger, including reversals, deferrals and corrections. A change in cash flow can inform review before delinquency, but the institution chooses contact, accommodation and recovery actions. An alert is a supported observation, not a repossession instruction. [3]
From evidence to a reviewable change
- 01Match the vehicle
Resolve borrower, facility and VIN evidence.
- 02Reconcile financing
Explain price, fees, trade-in debt and principal.
- 03Track two exposures
Follow payment capacity and supplied collateral evidence.
- 04Review a change
Compare the new position with the approved snapshot.
Build it into your operating model.
Start with one defined use case, your current decision policy and the output your reviewers need. Confirm which source connections, document types, calculations and verification stages are available within the agreed workflow. Create the subject and application, obtain the required authorization, synchronize permitted records and inspect the ready input manifest. Freeze that evidence window before launching an Insights underwriting run. [3]
Store the completed risk snapshot with the lender’s actual decision and its reasons. For ongoing monitoring, use a ready approved baseline from the completed assessment or the agreed existing-portfolio onboarding process. Signal compares newly available evidence with that baseline. Signed webhooks notify your systems; source cadence and consent determine freshness. A requested daily check does not guarantee daily source data. [3]
Design separate routes for a supported risk change, a source outage and an unresolved discrepancy. Retain source references and reviewer corrections. Borrower outreach, payments, limit changes, collections and insurance administration remain in your authorized systems. Verification and analytical investigation are included stages within an entitled Insights or Signal workflow, not independent Amlok products. [2]
Prove the fit before expanding.
Use separate new/used, prime/non-prime and origination-channel cohorts. Score amount, rate, term, VIN and payoff extraction independently. Evaluate later vintages and hold out the borrower and identifiable vehicle to avoid duplicate exposure in development and test sets. [4]
Measure calculation error, unsupported valuations and analyst corrections before evaluating loss prediction. Any outcome study needs mature arrears and recovery histories; repossession occurrence and net loss are different targets. Report uncertainty for small recovery cohorts rather than presenting a universal recovery benefit.
Agree success criteria before reviewing the test results. Compare the current process and the Amlok-supported process on the same permitted evidence, recording analyst time, corrections, unresolved cases and completed outputs. Report coverage alongside speed so incomplete files cannot make the workflow appear faster. Monitor quality by source availability and relevant customer cohorts, with uncertainty where samples are small.
Amlok’s public results currently document specific structured-loan and financial-evidence retrieval experiments. Those results are useful evidence about those tasks; they do not validate every product, institution or industry in this guide. A deployment claim needs a representative evaluation of your use case. [5]
- Scope: choose one product, one decision moment, required evidence and a clearly accountable reviewer.
- Replay: use dated historical snapshots with later outcomes kept out of the input evidence.
- Shadow: compare outputs alongside your existing process before relying on them operationally.
- Expand: review quality, data continuity and monitoring relevance before adding further cohorts or sources.
Sources & methodology.
Figures identify their population, period, and units. Scenarios marked illustrative are Amlok calculations using stated assumptions, not observed customer outcomes.
- 1Impacts of rising costs and claims on personal automobile insuranceStatistics Canada
Released 2 April 2025; December 2019–December 2024 median vehicle prices. Price changes describe market composition as well as prices, not depreciation of a given financed vehicle.
- 2Products and endpoint scopeAmlok
Documented ownership of Insights, Signal and included processing stages. Workflow availability and provider coverage depend on the agreed implementation.
- 3Integration workflowsAmlok
Frozen input manifests, immutable risk snapshots, approved monitoring baselines, source-dependent refresh and institution-owned intervention.
- 4Consumer Auto benchmark and evaluation definitionsAmlok
Segment-specific research definitions and proposed evaluation tasks. This is a research framework, not a claim of completed segment validation.
- 5Measured research resultsAmlok
Published SBA structured-loan and FinQA retrieval experiments have limited populations and tasks. They do not establish performance for this segment.
