INSIGHTS / Blog

Alongside Business Leader, SETsquared, recently hosted a virtual roundtable discussion on the topic of ‘Future Transaction Technologies’. Our aim was to bring together business leaders and academic experts to look at how AI and blockchain are disrupting transactions in the finance, construction, property and consumer services sectors.

A theme that emerged strongly is the role of AI in the credit analysis and lending market. Keen to explore this further, I followed up with OakNorth, a leading fintech, to really understand the role of AI in this sector and where and how the human element comes into play.

“A big part of credit analysis is tackling data. How to get data, how to make sense of the data you have and how to compare one company with another company and that’s where AI comes in.”

Valentina Kristensen is Director of Growth & Communications at OakNorth, a  fintech that specialises in creating a better borrowing experience for the ‘missing middle’- scale-up companies that drive GDP growth and job creation. It does this via its credit analysis and monitoring platform which it licenses to 18 banks globally, and leverages in the UK to do its own balance sheet lending via OakNorth Bank.

“Our founders met at LSE whilst doing their master’s degree and went on to start their own company. Keen to scale their business without losing control of it, they looked for debt finance but quickly found this was difficult to access. Using off the shelf solutions, banks would only lend to them if they had property to provide as security and weren’t willing to take the time to understand the business and its unique needs.”

“Their second venture sought to tackle this problem. There are many players in the retail and small business lending space (i.e. loans of tens of thousands up to a few hundred thousand dollars), but no one else is focused on the Missing Middle (i.e. loans of £1m – £25m) that we’re aware of.”

“It’s the application of AI that’s disrupting the industry rather than AI itself. The traditional approach to creating a credit paper where a credit analyst manually sources the data is time-intensive, costly and doesn’t scale. We leverage big data and machine learning to pull together this credit paper in a fraction of the time and with a much broader spectrum of data sources – including unconventional and previously unavailable data.”

“The best way to explain it is through an example. Let’s say we were approached by a vegan restaurant that wants a loan to open another branch. A decade ago, vegan restaurants were pretty rare, but now veganism is much more common and rising in popularity. So, if you were to take a historical look back at the how vegan restaurants perform, you’d have pretty limited data, so wouldn’t be able to paint a full picture of how that business could perform in the current market. We can use AI to undertake sentiment analysis, to get a sense of the momentum in the vegan restaurant trend or to understand the consumer sentiment around their existing restaurant with online reviews etc, leading to a much fairer credit assessment for the client.”

“AI also has a huge role to play in portfolio monitoring – our Platform proactively monitors the portfolio and flags up any potential issues before a negative event arises. If for example, there’s a covenant that says that at any given time, a business must have £100k of liquid cash available, our Platform can see in advance whether they are likely to break that covenant and we can therefore address it with the client and potentially  change terms or they can re-finance elsewhere. At a traditional bank, they typically only review the portfolio on an annual basis and will only review it outside of this period if there’s a reason to such as a late payment – at which point, a default may be inevitable within a month or so. It is the Platform’s monitoring piece which has enabled OakNorth Bank in the UK, the first bank to use the Platform, to lend £4bn to date with only two defaults and no credit loses.”

Well that seems like a better result for both the lender and the client. At the end of the day people still want to borrow from people, so as long as AI drives the back end, with a human at the front end this seems like the future of lending.

Dr Tiejun Ma, Associate Professor in Risk and Decision Analytics at the University of Southampton also took part on the roundtable discussion. He’s been working with industry sectors to develop prediction algorithms for user behaviour and for financial risks.

“One example of a huge change that’s happened over the last few years is using mobile phones to actually borrow money and make transactions. We can use various wisdom that can predict, for example, individuals credit risk, not having the traditional bank transaction data, but based on the mobile data available and assessing the possibilities of how much money you can lend to individuals and the technology allows this decision to be made in minutes. So rather than traditionally taking days or even weeks, we have been working with Big Data and mobile finance companies where we’ve developed intelligent algorithms where AI plays an important role in improving the prediction accuracies in terms of credit risks, but it also brings quite a number of challenges.”

“One of the challenges is that many of the AI algorithms are not explainable. So, you can’t really give a full view to the Financial Conduct Authority regulation. And also, you don’t really know where the prediction power is coming from.”

“So what we saw was AI needs to be combined with explainable prediction algorithms to get the benefit of being able to understand the underlying driver of improving the prediction power and explain what you’re really trying to predict, together with a state of the art AI algorithms to really get the edge  and fulfil regulations including ethics.”

“Fraud prevention is another major challenge of using mobile data for lending. How can you prevent fraudsters trying to use fake ID to borrow money? Well risk assessment through latest AI models plays an important role in trying to identify such malicious activities through multi-dimensional behaviour analysis and smart risk assessment models, which could substantially reduce operational costs for finance firms.”

What this discussion has made clear is that AI has an important role to play in the lending market. It has the potential to and is improving the customer experience, helping banks make more accurate decisions, driving profits and preventing fraud.

The big banks are here to stay. So, the fintech companies using AI to provide agile solutions will only thrive if they help the larger incumbents by partnering and collaborating.

If you’re currently scaling your business and are looking for support to raise investment for your R&D – our Scale-Up Programme could hold the key to unlocking your next phase of growth.

Find out more about the Scale-Up Programme