Exclusive: ‘Debt and AI: a question of ethics’ – Raman Bhatia, HSBC in “The Fintech Magazine”

Open banking and digital technology have already helped HSBC dramatically extend access to credit as it begins to reimagine the lending model. But those tools must be used thoughtfully to protect banks and customers, says  Raman Bhatia, former Head of HSBC’s Digital Banking arm in the UK and Europe.

The world of lending is on the cusp of a revolution. The new decade could see the emergence of lending widgets that sit wherever customers are, taking decisions on loan approvals based on the data accessed via application programming interfaces (APIs) from decentralised identity stores. Consumers would be pulled closer to lenders, with credit becoming seamless and topping up accounts in an instant.

But what are safe levels of indebtedness? And how far should technology that can take greater control of customers’ personal finances go?

Those are the kinds of questions that have been occupying minds at HSBC, as the bank tests technology that can create opportunities to expand credit to a larger population while holding the bank’s risk appetite constant.

“In that sort of world, you can have lending everywhere, but it needs to be managed very thoughtfully because the conduct issues become far more complicated,” says Raman Bhatia, former head of HSBC’s digital banking arm in the UK and Europe. “We could be talking about a world where identity has been abstracted out and exists as a service, which means know your customer (KYC) checks can be done at the point of engagement, wherever the customer is. They take their identity with them and ending decision models exist as a service, as APIs.”

The reform that made such a future possible in Europe and is now being adopted in various forms across the world, is open banking, which liberates data to be shared between independent parties with customer consent. HSBC identified the opportunities early on. Even before the UK Open Banking implementation and the European revised Payment Services Directive arrived in 2018, it offered its customers the ability to aggregate their accounts in one place using its Connected Money feature. Now it leverages open banking to reduce friction for customers, such as pulling in data to part-populate forms during onboarding. But where it’s had the biggest impact on both the bank’s business and its customers’ financial lives is in improving HSBC’s affordability decision-making process. This has resulted in a 30 per cent increase in loan approvals. 

The breakthrough, says Bhatia, was being given access to customers’ transaction data from non-HSBC accounts. It means the bank has enough information, even on thin file customers – those who don’t use HSBC as their primary bank but apply to it for a loan – to build up a full picture of their creditworthiness.  

“With open banking we have far better visibility,” says Bhatia. “HSBC uses a consent solution, digitally, and pulls in customers’ non-HSBC current account data, which goes through an automated categorisation and underwriting process.”

Now, HSBC wants to further reimagine the lending model.

New, real-time data signals 

Lending traditionally employs three main data signals: identity, affordability and creditworthiness, the latter two being backward-looking indicators. 

“But HSBC is beginning to frame and test some very innovative hypotheses,” says Bhatia. “For example, if there is a customer who is mobile, digitally engaged, and has a habit of checking their balance repeatedly, could that be a signal of that customer’s creditworthiness? Would the models around default incorporate that?”

While open banking has created an infrastructure that allows lenders to ingest such new sources of data, the challenge now is in developing the technology to interpret more dynamic, real-time information to inform lending decisions.

“There are a few innovative players that have emerged, whose raison d’être is to find better models using machine learning. Machine learning creates a new set of opportunities around decision-making, and we are working to apply machine learning models,” says Bhatia.

There is one customer segment he believes could benefit from a more imaginative approach. Bhatia admits SMEs  (small and medium-sized enterprises) have been chronically underserved by the UK’s established banks due to difficulties around assessing creditworthiness. But he is hopeful new digital signals can be found to improve that. He says: “It’s one area where we’re seeing a lot of innovation. On the challenger front, there are quite a few players that gaining scale in SME lending. But HSBC is doing the same. It is working on a new proposition for SMEs, which would have a new lending engine as part of it. It is also asking whether it can find new signals of data for effective lending decisions for SMEs.”

HSBC is beta testing Kinetic, a new SME bank due to launch in 2020, which integrates the Xero accounting package among other third-party apps.

 “I remain optimistic around HSBC’s ability to serve small businesses,” says Bhatia, “but it remains a challenging space, hence the open banking competition around it and grants for challengers.”

A collaborative approach

The biggest bank in the UK has been keen to collaborate to improve its decisionmaking process. Both Equifax and AccountScore work in partnership with HSBC to provide and analyse loan applicants’ current account transaction data. And Glasgow-based Castlight Financial provides an engine that categorises income and expenditure into 155 types to reveal an applicant’s monthly disposable income.

Bhatia adds: “HSBC is also working with a company specialising in machine learning models to find pockets of lending opportunity around point of sale or buy now, pay later – although with that, the industry has to tread carefully because it could create a new debt trap.”

Big banks, having been dragged through the debt crisis, are particularly sensitised to the dangers of expanding credit, given the personal lending market is still dominated by them.

How long that continues as new entrants find dynamic ways of using data for know your customer (KYC), creditworthiness and affordability checks tailored to individuals, not borrower categories, remains to be seen. But while paper has almost disappeared from lending processes (70 per cent of HSBC’s loans are now processed digitally from end to end), and comparison websites have reformed how consumers select who they apply to, fundamentally, little has changed for borrowers, says Bhatia. 

Challenger banks have improved customer experience but he believes internet-only players ‘have shown no appetite, no capability, to do lending at scale’. Open banking, he believes, could change that – but customers will need to be both educated and protected.

“The flip side of this new era of lending is that there could be issues around persistent debt,” warns Bhatia. 

“Digital, if not managed well, can create systemic persistent debt, which means we must balance lending with providing advice around borrowing.

“HSBC has done that through push notifications, which tell customers when they are likely to hit an overdraft, for instance,” says Bhatia.

“Left to their own devices, many customers are likely to borrow beyond their means. The bank obviously has a credit risk policy, which is based on prudent conservatism around where it can lend. But it needs to use the power of digital and data to guide people around their credit score, engagement with money and saving enough, so that they don’t need to borrow, in effect.”

A principled debate

That conservatism has helped HSBC build a trusted brand, which extends to its cautious approach to using AI. It’s a member of the Veritas consortium working on an ethical AI framework, drawn together by the Monetary Authority of Singapore.

“The bank wants to have a principles-based discussion around when and how AI should be applied in the world of credit and lending,” says Bhatia. “It’s having a very robust debate, internally, around its AI principles, and it wants to move cautiously. But where data can move seamlessly, identity is decentralised and decision-making can happen outside of your own estate – that is a future HSBC is preparing for.”

This article was first published in The Fintech Magazine: Issue #15, Page 46 & 47.

Featured in this Article:

HSBC Raman Bhatia

Author: Laimis Bilys