Only as good as the data

When it comes to regulation in the UK mortgage market, there is often value in read-across of general principles.  

In July, the Financial Conduct Authority (FCA) published a market view relating to firms’ implementation of the Consumer Duty – ‘Outcomes monitoring: good practice and areas for improvement’.  

Beyond surface-level metrics

The regulator warned what we all know, that management information alone does not demonstrate good customer outcomes. A valuation dashboard may show a figure, a confidence score and a rapid turnaround time, but those metrics do not prove that the property information behind a lending decision is appropriate for the risk being taken.  

A mortgage valuation is most often discussed as though it were a discrete event, when a property is assessed, a value assigned and the lending process moves on. In practice, it is the output of information that can include transaction records, property attributes, local market evidence and modelled assumptions.  

Each input has a source, date, degree of coverage and potential limitation, meaning the quality of the conclusion depends on how well those factors are understood and governed.  

This is increasingly important as lenders seek timely decisions. HM Land Registry’s Price Paid Data, updated monthly, provides an official record of sales for value in England and Wales that have been lodged for registration. It is a valuable component of property intelligence, yet no individual dataset can explain a property in full.  

Lenders must understand how recently information was recorded, whether it is comparable to the security being assessed and how it relates to other evidence.  

Reframing valuation governance

The FCA’s July review usefully reframes the standard. The regulator found that some firms relied on lagging indicators, lacked clear thresholds or did not maintain audit trails from identifying an issue through to action and outcome. It also highlighted gaps and inconsistencies in key data that limited firms’ ability to evidence customer outcomes.  

The implications for valuation governance are clear: lenders should be able to trace material property information, establish when it was refreshed and understand how a conflict, exception or model override was resolved.  

That is more demanding than retaining a final value. Valuation data should be treated as a governed decision record, and if an automated process identifies a property as materially different from its comparables, that should be visible.  

If performance changes in a particular geography or property type, the lender needs a threshold for intervention. If a surveyor’s view departs from automated evidence, the difference should be capable of informed challenge. Good governance gives expert judgement a clear evidential foundation.  

Beyond origination

Property risk changes over the life of a mortgage. Local transactions, physical condition, leasehold arrangements, energy performance and planning decisions can alter the lens through which a lender views a security.  

A portfolio assessed sensibly at completion can still develop concentrations of risk if property information is not refreshed, interpreted and connected to the risk framework. Lenders need to identify cases requiring attention before a refinancing event or change in risk appetite exposes the weakness.  

Security, resilience, and AI threats

There is also a security and resilience dimension. Mortgage valuation workflows draw on data suppliers, surveyors, models, geospatial tools and cloud-based platforms. HM Treasury reported in July that 82% of surveyed UK banks, insurers and asset managers regarded cyber attacks as a top-five systemic risk.  

Only 10% of organisations reported preparedness for AI-augmented cyber threats. A lender that cannot establish a data feed’s provenance, identify where it is used or recover it after disruption may be relying on an incomplete view of the asset supporting the loan.  

The Bank of England has reinforced that concern, warning that frontier artificial intelligence (AI) can accelerate the identification and exploitation of software vulnerabilities. It has urged firms to reconsider whether recovery arrangements and key technology providers remain sufficiently resilient.  

Future-proofing mortgage decisions

For lenders using more automated property intelligence, this is a reason to strengthen governance before scaling usage. They need clear visibility of ownership of data sources and an understanding of which lending decisions would be affected if a provider were unavailable or compromised.  

The future of mortgage valuation will depend on how effectively lenders make sense of growing volumes of property information. The lenders best placed to use that information well will be those that can demonstrate its lineage, protect its integrity and apply it with appropriate judgement.  

‍