Antony Jenkins
Thursday, May 31, 2018

Ten-minute mortgages are closer than you think

Sarah has a new job in Sheffield, and she and her boyfriend Tim are looking for a mortgage for a house they have fallen in love with.


They have instructed an artificially intelligent agent provided by their favourite social media company to gather all the data about them from across the digital world – their finances, education, lifestyle, employment prospects, credit history and risk tolerance – to identify the best mortgage for their circumstances.


That mortgage is provided by a pension fund that uses blockchain technology to make cheap loans directly to borrowers. The fund’s automated agent has already exchanged all the data that both parties need to advance the loan, and a smart contract has been created in minutes.


Once Sarah and Tim are happy with the deal, they trigger the contract, disbursing the loan proceeds and finalising the purchase of the house. And because theirs is a smart contract, their monthly payments adjust automatically according to their expenses, so they can overpay when they’re able, or reduce a payment when they need extra cash for a holiday.


The march of automation


This will sound far-fetched to most – it’s hard to imagine getting a mortgage without having to endure an arduous process of phone calls, in-branch meetings and long waits for correspondence. Yet more and more FinTechs are investing in the artificial intelligence technologies – smart algorithms that act autonomously according to their context – that could eventually make these digital ten-minute mortgages possible. LinkedIn’s survey of finance professionals, for example, shows that 63% of FinTechs are interested in AI-based investing as a core future technology, and I believe that number will only grow over the coming years.


That’s because banking is rapidly digitising, faster than I had previously thought. Bank branch footfall in the UK was down 30% over five years in July 2016. The trend is global: in Asia, over 95% of transactions made with Citibank occurred outside a branch as early as 2012, and the company’s Korean division recently announced plans to shut down 80% of its branch network, calling branches an “outdated model”. And as the popularity of digital transactions like contactless payments grows, so customers offer their financial providers ever sharper clues about their habits that can be used to offer more personalised services.


You would think retail banks were in a prime position to capitalise on this trend towards the digital, since they hold all the information about their customers that could allow them to offer smart products like dynamic mortgages. Yet LinkedIn’s survey shows that retail banks are behind the curve, with bankers much less interested in AI technologies than their counterparts in FinTech. Consequently, they are underinvesting – PwC’s annual FinTech report for 2017 showed that nearly half of large FinTechs are investing in AI compared to only 30% of large incumbent institutions.


Changing mentalities


So, why aren’t banks doing more? Firstly, it’s a function of their current systems. Machine learning – the most promising example of AI today – requires a centralised, clean dataset to be trained on, and before automated agents can start arranging contracts, banks need an open, standardised architecture over which data can flow. But the banks’ current creaky patchwork of IT systems developed over many years severely hampers this essential integration, making it very difficult for them to pull all this information together in the first place.


And secondly, the issue is cultural. Since the incumbents’ mindset involves risk minimisation, only tinkering with existing, successful business models when required, it is hard to step back and make significant adjustments that carry greater perceived risk, even when the technological pressures are becoming more obvious. That is even truer when it comes to the critical hardware and software infrastructures that underpin all financial transactions – making adjustments to that system while it is live would be like performing heart surgery on a marathon runner.


That is why it is important that banks set aside resources to set up greenfield operations outside their current architecture, providing the sandbox in which they can contemplate radical moves before migrating services from their existing systems. As I recently argued at Money 20/20 with Oliver Bussmann, partnerships between FinTechs and incumbents may provide the best model to achieve this, allowing banks to hedge their risks while providing FinTechs with profitable revenue streams.


It would be foolish to deny the march of automation. I believe the cocktail of fast mobile data, cloud, AI and distributed ledgers could take the disruption from bank apps to the next level. That means there’s not a huge amount of time for CEOs to wait for their organisation to “catch up” to the need for this technology. If they do not, it will be their FinTech rivals who offer the ten-minute digital mortgages that are so superior to what banks currently offer. Without swift action, the Kodak moment of banks’ obsolescence may be approaching.

Return to Insights