Proportionate Creditworthiness Assessments

What proportionality changes in underwriting

Not every loan requires identical forms, bank statements or income-verification steps. A modest instalment with a short term may present a different risk from a large revolving line or a repayment profile with substantial variable interest. Proportionality concerns how deeply a firm must investigate relevant factors, not whether it must make a sound decision where an assessment is legally required. Product amount, repayment pattern, existing financial strain, inconsistent statements and data reliability should all inform the process. A policy must explain which checks are mandatory and what conditions trigger additional evidence. Avoid an automatic rule that a loan below a marketing price point can never cause difficulty. The creditor should identify the actual repayment commitment and relevant customer circumstances before selecting a lighter verification pathway.

The UK standard in CONC 5.2A.20

FCA CONC 5.2A.20 explicitly states that the extent and scope of a reasonable creditworthiness assessment, and the steps taken to satisfy it, depend on and must be proportionate to each case. It calls for consideration of information types, sufficiency, further inquiries, credit reference agency information, other sources and verification. CONC 5.2A links risk of missing repayments and affordability risk. A firm must respond to known information suggesting unsustainable borrowing, even where standard policy normally permits declarations. Loan size is relevant but not the only determinant. A large instalment, long term or unusual expenditure pattern can justify deeper inquiry. A written approval policy should document how the firm's chosen checks meet the current FCA requirements and why a particular exception was reasonable.

Evidence escalation for riskier applications

A risk-tiered process might begin with accurate existing customer records and application declarations, then require payslips, permitted account information or other confirmation when specified triggers arise. Those triggers could include income volatility, unexplained debts, large payments relative to available income, material account changes or data that contradicts the applicant. The creditor should not assume that more personal data always improves assessment: each request needs an appropriate purpose and lawful basis. When a source cannot be used, define an alternative route that does not automatically exclude a person for an impermissible reason. Reviewers should record why they accepted an estimate or sought corroboration. Controls should detect successive small credit increases, duplicate expenditure and stale data that might bypass a risk threshold or materially misstate repayment capacity.

EU CCD2 requirements and implementation

Directive (EU) 2023/2225 Article 18 establishes a revised assessment based on adequate, relevant and accurate financial information and seeks to prevent irresponsible lending. The Directive should not be described as adopting the text of FCA CONC 5.2A. Member States had to transpose it by November 2025, and implementing provisions generally apply from 20 November 2026. Existing national law, applicable transitions and national requirements matter to decisions in October 2026. A lender operating in multiple Member States should map what level of inquiry is lawful and adequate for its products and customers. National supervisors may have expectations about verification, information quality and the handling of negative assessment results. Record which jurisdictional rule supports an assessment tier rather than advertising a single EU-wide exemption for small or digital loans.

US limits on direct comparisons

US federal rules do not include a single proportionality clause matching FCA CONC 5.2A for every consumer product. Covered mortgage transactions have specific ability-to-repay requirements in Regulation Z, while Regulation B addresses permissible evaluation, prohibited discrimination and adverse-action notices. A creditor may use operational risk tiers but still must meet the rules applicable to each loan and state. A low-dollar loan cannot be assumed free of consumer protection or fair-lending obligations. For automated models, explainability matters when denials trigger notices under section 1002.9. A less burdensome data process is permissible only where the actual law allows it. Document why information is collected or excluded and preserve the real factors used when a covered application is denied.

Governance of thresholds and exceptions

Policy owners should define thresholds for loan size, term, payment burden, data confidence and material contradictions, with separate treatment for product categories and legal regimes. Run scenario tests for low income, irregular wages, shared household expenses, recent bureau discrepancies and repeat borrowing. Ensure that reviewers have authority to request more evidence or stop an approval when warning signs arise. Overrides should require reasons tied to documented facts, not commercial pressure. Underwriting, legal, data protection and technology teams should jointly approve changes to assessment depth or vendor sources. Maintain version histories so the firm's treatment of a historic application can be explained. Do not label a policy proportionate simply because its average processing time is short or its approval rate is high.

Reviewing whether the policy works

Independent assurance should examine files from every assessment tier, including the easiest automated approval routes. Check whether assessments reflect the terms offered, whether estimated expenditure was reasonable, and whether contradictions triggered appropriate review. Monitor repayment difficulty and complaints among borrowers assessed with lighter checks, while recognising that poor outcomes alone do not prove every assessment was unlawful. Compare decisions for similarly situated applicants and examine evidence of discriminatory effects under the applicable law. Update thresholds when the product or customer population changes. Preserve remediation evidence if rules, models or data classifications are revised. A defensible policy explains the chosen amount of inquiry and shows that its exceptions work in practice, rather than treating proportionality as a cost-saving label.

Frequently asked questions

Related research

Creditworthiness and Affordability Assessment Compliance ↗

Creditworthiness Assessment Recordkeeping and Audit Evidence ↗