Mortgage lenders need more than AI to understand value and risk – Miller

AI integration is playing a role in driving automation, reducing costs, answering knowledge gaps and minimising human error. 

Whether it’s data analysis, advanced underwriting, vulnerability detection, fraud prevention or customer service and support, there are so many use cases across financial services – particularly for mortgage lenders. 

In a market and economy where firms are being squeezed, it’s not a surprise that many lenders are questioning whether AI could be that silver bullet – the answer to all their challenges. The common outcome seems to be that while AI is certainly impressive and its capabilities are ever-growing, it’s not ready to eat the whole elephant in one bite. In my view, it is a powerful toolkit firms can leverage to support and facilitate their work, rather than a magic wand. 

I think one of the best examples to demonstrate this is in understanding the value of property portfolios and, more importantly, identifying any potential risk. While AI can certainly complement, it cannot yet replace the insight gained by having boots on the ground and eyes on the asset. 

Sponsored

One Year On: Helping You Add Value with Halifax’s Green Living Reward

Sponsored by Halifax Intermediaries

Increasing automation 

Given the hefty mortgage books most lenders are now working with, we have already seen increasing use of automated valuation models (AVMs) to process large volumes of data and generate bulk valuations. By leveraging real-time market data and algorithms, lenders are able to establish a macro view of their portfolio in an efficient and cost-effective way. 

Many modern AVMs now use AI and machine learning to analyse these large data sets – including historical sales figures, property attributes and market trends – to provide valuations for mortgages and lending purposes. While this is hugely valuable for lenders and a complementary use case for AI integration, it doesn’t give lenders the complete picture they need. 

While efficient, AVMs and AI are still reliant on data that may not fully capture regional or even local discrepancies in value, full market sentiment or, most importantly, the physical condition of the property. For lenders trying to balance risk, governance and competitiveness, an AI-first approach would, for now, be short-sighted. 

Condition is critical 

AI is certainly suited to analysing data and crunching the numbers, but is unable to spot a property in poor condition – a situation where any discrepancy in value is only exacerbated. ChatGPT or Copilot cannot see signs of neglect on a property, nor can it assess the impact of neighbouring properties or the broader environment on desirability. The best way to do this remains with drive-by valuations – allowing lenders to get eyes on the asset to understand its condition and any potential risks. 

These can be completed quickly and discreetly without the occupier’s knowledge or involvement, and can be undertaken as little or as often as required in line with the lender’s appetite to risk. Alongside key insights about the property, an asset manager’s inspection will also consider the surrounding properties and the local area to provide any further intelligence to support the valuation. 

With everyone keeping a close eye on the future path of arrears, there’s no doubt that lenders need to be vigilant. Early intervention is only possible, though, with the right information and intelligence. This includes the valuation, a projected market value – what an asset manager would expect to achieve within a 90-day marketing period – and a physical assessment of the property.

With this information, lenders can segment their portfolios, price competitively in the right markets and exercise good caution and governance. 

Good outcomes 

Most important of all, early intervention provides the biggest opportunity to ensure a good outcome for all parties involved – not least the borrower. Lenders can act early and, with the help of an asset manager, use an assisted voluntary sale scheme to achieve the best possible price in the shortest amount of time. Using this route, it is still possible for borrowers to secure some equity and start fresh.

This is a real win all round and a great way to demonstrate compliance with Consumer Duty. 

Managing valuations and understanding risk across a large mortgage book is no small task and can certainly be resource-intensive – as seen in the adoption of AVMs. While lenders may not have the capacity in-house to implement drive-by valuations at scale, many rely on outsourcing to manage this process and capture this critical information. 

Partnering with an expert asset manager allows lenders to keep focus on their core business, while still gaining the necessary insights to not just service, but protect and grow their mortgage books. Meanwhile, lenders can continue to use AI wisely to leverage its efficiency in areas where it is already making a sizeable impact.