An affordability assessment is a lender's check that a borrower can meet the repayments on a credit agreement from their income, after essential spending and existing debts, without falling into hardship, over the whole life of the agreement. It asks whether the customer can afford the loan, not only whether the lender will be repaid.
Why does it matter once AI agents act?
Affordability work is heavy on evidence: bank statements, payslips, credit files, declared spending. Agents are good at gathering and sorting that evidence. The judgement on top of it is where errors cause harm. Deciding whether an outgoing is essential, whether irregular income will continue, or whether a pattern of spending suggests the customer is struggling is not arithmetic.
The cost of a wrong yes falls on the customer, who takes on debt they cannot carry. In the UK, the FCA Consumer Duty expects firms to deliver good outcomes for retail customers, and lending someone cannot afford is a plain example of a poor one. The wider limits on agents in lending are in AI agents in banking.
How does an affordability assessment work in practice?
- Income: establish what comes in and how reliable it is.
- Essential outgoings: housing, utilities, food, travel to work, dependants.
- Existing commitments: other credit, arrears, payment plans.
- Resilience: what happens if income falls or rates rise over the term.
- Comparison: does what is left cover the new repayment with a margin?
- Record: the decision and the reasons, with the evidence behind each figure.
Where the facts come from matters. Income shown on statements retrieved from the customer's bank is stronger evidence than income the customer types into a form. An agent should label each figure by its source so the reviewer can see which numbers are evidenced and which are only declared. How to check income and affordability evidence with an AI agent sets out the steps.
What does that look like in an example?
A car finance lender receives an application declaring monthly income of 3,000. The agent reads three months of statements: 2,400 a month in salary and irregular payments from delivery work that vary widely. It calculates affordability twice, once on salary alone and once including the average extra income, and the loan only works on the second. It also notices frequent payments to betting sites. It drafts a note setting out both calculations and flags the spending pattern. An underwriter decides, and the betting pattern prompts a conversation rather than a silent decline.
How is it different from a creditworthiness assessment?
A creditworthiness assessment looks at credit risk: how likely the borrower is to repay, which protects the lender. An affordability assessment looks at the borrower: whether repaying will cause them harm. A customer can be a good credit risk and still be unable to afford a new loan without cutting essentials. Some regimes treat affordability as part of creditworthiness; the two questions remain different, and a lender should be able to answer each separately.
Which related terms should you know?
- Income verification: confirming declared income against evidence.
- Vulnerability: circumstances that make a customer more likely to suffer harm, which an affordability review often surfaces.
For working tools, see the affordability assessment checklist and can an AI agent check affordability for a loan?
Questions people ask
Is an affordability assessment the same as a credit check?
No. A credit check reads the customer's credit file. An affordability assessment uses that file as one input alongside income, spending and circumstances.
Can an agent decline an application on affordability grounds?
It can recommend a decline and set out why. The decision should sit with a person or with decision rules the lender has approved and tested, and the customer should be told the reasons where the rules require it.