Powered by MOMENTUM MEDIA
SUBSCRIBE TO OUR NEWSLETTER SIGN UP
Powered by MOMENTUM MEDIA

Website Notifications

Get notifications in real-time for staying up to date with content that matters to you.

Predictive machine learning could come to mortgages

digital web digital web
Reporter 4 minute read

Big data and machine learning could soon help lenders offer tailored mortgages based on the real-time financial picture of a borrower, according to the CEO of a data aggregation firm.

Anil Arora, CEO of Envestnet Yodlee, a leading data aggregation and analytics platform for digital financial services, explained that by monitoring and tracking the historical transaction data of customers, lenders could see a real-time risk analysis of them.

It’s this type of machine learning and technology that could be applied to the mortgage industry to provide better tailored products to users, Mr Arora said.

Advertisement
Advertisement

Speaking at The Adviser’s US Study Tour in San Francisco last week, the data specialist said: “The average US citizen is sitting on about 15 financial accounts across an average of seven institutions... You’ve got various parts of your financial life fragmented across different accounts and different institutions. But we can look at all that.”

The company currently offers its risk insight solutions as a white label product to banks — and it’s here that Mr Arora suggests the tailored mortgage solutions could be adopted.

“We can use the data from Bank of America, for example, to say that [this customer] has this kind of behaviour… Perhaps, you want to offer him a targeted mortgage offer. Or we can say, they already have a mortgage and their rates are not comparable. They can save money. And you should offer them this kind of rate.”

Envestnet Yodlee collates “hundreds of millions” of transactions across different groups to provide more visibility to lenders. For example, they have “real-time visibility into assets, into income flows, into spending patterns” and then uses that to make predictions.

PROMOTED CONTENT


“You’re using more holistic data, so you can actually come up with some really scientific levels of prediction on whether this person really can repay their loan or mortgage.”

Mr Arora went on to highlight that this type of predictive and machine learning technology can also provide real-time offers without relying on credit scores.

“It's great for those who have no credit history… In the US, we have a completely broken credit system. If you go and look at the data set that fuels the credit scores, it makes no sense anymore. About half of the data is obsolete.”

The Envestnet Yodlee CEO gave the example of using a FICO mortgage score, which is heavily weighted towards timely payments, rather than net worth, of a borrower. With this model, according to the CEO, a borrower may be very cash-rich, but if they made one late payment, their FICO score would drop, despite the fact that they have “millions in the bank”.

He added that as mortgage offers are traditionally made with data that is six months old, and that borrowers have to wait for several weeks before a loan is approved, their financial position could have changed in that time and therefore unfairly penalise them for their past financial position.

“We live in a digital world and they’re making you do all these paper copies… It makes no sense. The borrow experience is awful. So, how do we improve that? You can automate gathering of data, but more important than that is you can make better risk decisions.

“If you had real-time information and the ability to then take that information to give a more insightful assessment of your client, you’re going to the next level. Because you can see what they are spending their money on — and if they have a propensity to buy a lot of expensive cars regularly, or they buy a lot of expensive wine, or they use gambling sites, you can add that data intelligence and machine learning to the offer.”

He concluded: “We have to figure out ways of how can we improve the systems to make better credit decisions. Part of it is a lot of customers don’t have credit cards, so this is another way to address that data gap.

“This is the service that we just launched recently in the US, and we will be at some point also going live in Australia... The one last geography that we have to crack over there is the big four banks.”

The Adviser’s US Study Tour 2017 was held in San Francisco between 1 and 3 November. Designed exclusively for Australia’s elite mortgage brokers and third-party industry leaders, it focused on innovation, technology, data and how the best US mortgage brokers dominate their markets.

Speakers at the conference included, among others: Jonathan Miranda — director of strategy, technology at ‎Salesforce — who spoke about the future of AI; StreamLoan CEO and co-founder Stephen Bulfer, who outlined how the adoption and progression of artificial intelligence could enable brokers to provide a more customised and valuable offering to their customers.; and Todd Duncan, who suggested brokers brokers move from being a broker toward having “a professional mortgage practice".

[Related: Brokers should ‘move away from being a broker’]

Predictive machine learning could come to mortgages
digital web
TheAdviser logo
digital web

 

more from the adviser
uptick Broker market share surpasses 60%

More than 60 per cent of residential mortgages were written by br...

anna bligh ABA calls for best interests duty clarification

The banking association is calling on government to provide great...

KAte Carnell ASBFEO New business funding guide released

An updated Business Funding Guide for brokers, accountants and o...

FROM THE WEB