Mortgage lending is expected to increase slightly in 2019 and the Bank of England may raise the base rate which means that competition on rates between lenders will continue. But there are other ways for borrowers to differentiate between lenders – apart from on price – such as customer experience, speed of service, an efficient service and making the customer journey easier.
Teleperformance, which provides digital integrated business services, has carried out some research and found that it takes 18 to 40 to days to go from mortgage application to final mortgage offer with the bulk of that being taken up by the time it takes to carry out a valuation.
The mortgage application requires original copies of utilities bills and bank statements but that slows down the process. Some lenders now accept digital versions, which is good because 69% of consumers receive paperless bank statements, and that goes up to 75% amongst 18 to 24 year olds, according to a research by ID verification software firm GBG.
Can identity data be used to speed up the mortgage application process and make it more accurate and less susceptible to fraud? Is it FCA rules or lender procedures that dictate borrowers having to produce hard documents?
There is continued growth in the use of artificial intelligence and Robotic Process Automation (RPA), which relies on data to perform repetitive and automated tasks, thus eliminating human error and freeing up humans to do other things.
Teleperformance has found that within the mortgage market, the five key metrics that influence customer decisions are brand, price, accessibility, customer experience and speed of service. The big lenders are strong in the first three but lose out in the customer journey and responsiveness offered by small lenders and fintech players.
Digitisation covers a host of other applications such as digital signatures; and does open banking, now a year old, hold the most potential for change in the mortgage market?
How far down the line is the mortgage industry with accepting electronic documentation instead of customers having to produce paper copies?
Kris Brewster, head of products, proposition & corporate communications, Skipton Building Society: We currently ask for payment documents and will accept scans but we are moving away from that and towards getting as much automation as possible into our back office processes. We use a combination of technology – automated valuations, income verification, access to data through open banking. All parts of the mortgage chain are starting to link up, even into conveyancing and Land Registry, so for us all the information is there, it’s about linking that up. I expect everyone across the industry is working on exactly the same thing.
Are consumers demanding the move from the conventional paper-based application to a more digitised process? People are used to home shopping, ordering goods online with delivery the next day. Is that shaping the way to build a lending process?
Matt Ward, head of mortgage service delivery, Santander: Very much so. I think there’s a consumer expectation that they should be able to communicate with their bank in the same way they would in any other area of their life. There’s a drive from consumers and intermediaries towards electronically submitting documentation and the challenge for the industry is to ensure we’re keeping this secure.
How much impact does regulation have on digital documentation?
Paul Clampin, chief lending officer, Landbay: The challenge is to work within the regulatory framework particularly anti-money laundering. The test for lenders is proving that the customer is who they say they are and that the security is adequate. Landbay is a fintech lender so we’re able to take applications to completion in 18 days but that’s where we have been provided with the information very early on in the process.
We’ve been able to automate large parts of the buy-to-let market because 97% is still intermediary based and they tend to do things in a formulaic process. The biggest change to the buy-to-let market and lenders generally will be the development of a single source of application data but that has not been developed fully.
How can sourcing systems and APIs move the application process on?
Paul Clampin: If all brokers use the same sourcing system for quotes and then send the application to lenders, they can immediately respond with an indicative quote on price and what information they’ll need. I think we’re a long way away from that, but something like that was piloted in 2003. I think that’s more likely to happen in the residential side or maybe the complex prime side of the market than it is for buy-to-let which has more complexities.
Andrew Asaam, director of mortgages, Virgin Money: I think that’s going to happen in 2019 and it’s going to be a game changer. It’s not for all intermediaries but scale intermediaries will get operational benefits from having a digitalised process.
Kris Brewster: We’re working on direct submission via APIs with various sourcing systems. I would expect that to be in place soon and that will be the long-term future for submission using data, cutting out the inefficiency, cutting out the rekeying and the manual work that broker admin teams have to do. There would be a strong push for that and it will take different lenders, different times to join that party but I suspect in the long run everyone will have to play in that way.
Andrew Asaam: I think we’re all digitising our processes, we’re just on different parts of the journey depending on funding and legacy systems. Digitisation of the ultimate customer journey involves signatures, documents, appointments, omni-channel with data consistency, APIs and all this will happen over the next 12 months.
Kris Brewster: One dilemma is around building systems yourself or using a partner. Thinking about the approach to APIs, it’s about getting that balance between what you own and what is with your partner. It is quite a change for a lot of lenders if we’re used to controlling all of that process ourselves.
Paul Clampin: It’s also about legacy systems. How quickly you want to move and your ability to move which determines whether you use a third party or do it yourself.
Alex Maddox, product & capital markets director, Kensington Mortgages: Risk is a key issue with third parties. How do we deal with prequalification from price comparison websites? If a customer has been prequalified with a credit check, and it’s not the credit reference agency we use internally, how do we deal with that, do we check it again? If it gets sent out to multiple lenders does everyone do their own credit check?
How advanced are lenders in respect of Robotic Process Automation (RPA) and artificial intelligence?
Matt Ward: I think quite a lot of lenders use RPAs particularly in larger organisations. We certainly use RPAs where we believe we’ve got repetitive processes, particularly those of a legacy nature where you’re moving data from one system to another. It’s a highly accurate manner in which to manage this.
Mike Sloman, SVP business development, Teleperformance: It’s quite ironic that one of the beauties of RPA is that it works best with legacy systems. When you have multiple legacy stable systems that are easy to automate it actually takes away a lot of the business case for replacing systems or putting new platforms in.
Kris Brewster: Our robotics at Skipton is in the pilot phase and we expect to be using that properly across real life processes in early 2019.
Puneet Taneja, EVP operations, BFSI at Teleperformance: The progress made on the use of robotics on the servicing side is much higher than in originations. If you look at appointment bookings today in most branches you still have a long, drawn-out process and we see opportunities for robotics to take a much more active role.
Alex Maddox: We used robotics in servicing first but we’ve now rolled it out to underwriting and even integrating with external parties to pull in data as an alternative to APIs. So it can be used everywhere but I think servicing is a good starting point, it’s a lower focus from a risk point of view.
Can RPAs help with underwriting?
Paul Clampin: I think it works more in residential because data sets are more established and more predictable than other markets such as complex buy-to-let, although standard buy-to-let might be easier.
Alex Maddox: I think it’s a journey. We’ve all been using rules-based engines to speed things up but that’s the first step on the customer journey. You need certain information to feed into those rules so how do you pull that in through robotics or APIs?
As we are a specialist lender, there will always be a lot of human interaction in every single case because our customers have some kind of complexity to their situation and it’s complexity that’s difficult to automate. But the complex part of the market is either in the top end of the buy-to-let space or is a relatively small percentage of the owner-occupied market. So there’s a lot of standard business out there that could be automated very efficiently and provide a great customer experience.
Andrew Assam: There are two challenges here. It might prove payment of rental amount but it does not factor in SVR plus 3% for stress testing requirements and the deposit raising is probably an even bigger constraint. So I think taking rental payments into account helps but I’m not sure it’s enough.
Matt Ward: Rental information will show consumers’ ability to repay, or commitment to repay, so clearly you will improve your view of them from a risk perspective.
Alex Maddox: It’s similar to the mortgage prisoner analysis where someone has exhibited an ability to pay off a certain amount but is that suitable evidence that they will be able to pay the same amount going forward? On all of our models historical payment is one of the best predictors of future payment ability, but unfortunately it doesn’t tick the regulatory box, because of the 3% interest rate stress test.
What future does robo advice have?
Kris Brewster: I would expect to see more development of robo advice or guided execution only solutions for customers particularly as remortgaging increases. There is technology to help support the advice process and there are cases that don’t necessarily need all the skill of an adviser.
Matt Ward: It’s an interesting challenge because we’ve been finding there are lots of digitally savvy customers that are capable of using online resources. If they wish to interact in this way, that’s absolutely fine, but those that are ’digitally savvy’ may not be ’mortgage savvy’. How do we know when it may be best to just put that customer straight into an advice process to make sure they have help from the outset? This is a market that everybody’s going to start to explore I would imagine.
Alex Maddox: One of the areas where AI is being used is chatbots – and that’s across finance not just in mortgages with different levels of success.
Matt Ward: It could be hybrid solutions that customers will be looking for in the future – a live chat, then a video conference with an adviser while they’re filling in their forms so they can talk things through.
Paul Clampin: The challenge with robo advice is ensuring it’s fit and proper for each individual circumstance, that’s very complicated. I don’t imagine lenders are going to use robo advice, their professional indemnity insurance would be quite interesting too.
Andrew Asaam: The regulator does want robo advice and there is a lender actively talking to the FCA about it and could be ready to launch in around three to six months. We’re being disrupted in a huge way with aggregators, fintechs, digitisation of unified services and in three to five years the industry will look very different.
How can customer retention be improved?
Kris Brewster: There’s potential to support customers post mortgage completion apart from an annual statement. We know people are not moving as often, they are home improving and borrowing more, particularly as they get into later life. I think there’s scope to improve the customer experience, and this is a new area that could be better developed.
Matt Ward: I agree. We could be more visible to the customer so they have an understanding of how they can interact with us, for example, a notification of the next deals that are available for them or reminders that they can overpay.
Alex Maddox: There’s a bunch of fintechs out there who are trying to engage with customers during the mortgage life so if lenders don’t start to engage, someone else will and you will lose your customer.
Kris Brewster: The challenge of retention is huge and I think one of the other changes is that aggregators will enter the mortgage market over the next 12 to 18 months acting somewhat like brokers but also with a lot of technology. They could do to the mortgage market what they have done to the home insurance market and disrupt it.
When interest rates start to go up some people may move into arrears. Can AI help to predict which borrowers are going to be affected and who might struggle in the future?
Alex Maddox: Our outcomes, which are based on machine learning, are the same ones that we use to predict arrears. They’re the same ones that we use to work out what headcount we need to cover arrears, based on the level of arrears that we expect. So it’s a single application that runs across the whole business.
What’s more interesting with digitalisation is identifying vulnerable customers. You have to train your staff to identify vulnerable customers early and if you can overlay technology on top of that then we should all do that.
If we can use open banking to help us really understand the customer’s true situation then we should be able to put together a better forbearance strategy working with the customer to help them out in that situation. But customers will have to be ready to open up their bank accounts to us and let us work with them or let their tax adviser work with them. The more information that is shared between the different parties the better and more successful the forbearance strategy will be.
Other points
It was also noted that speech analytics can detect people who have rung more than once to ask trigger questions such as what would be the impact on their mortgage if the rate goes up by 0.5%. You can build an algorithm that pulls up a list of people asking similar questions and red flag them.
There are systems now which alert the borrower to a better mortgage deal, even though they’re not specifically looking. It was pointed out that the deal would have to be appropriate to the individual’s circumstances which could be a challenge.
Post offer process
There was a discussion around the post offer process which can be slow with customer feedback showing frustration around valuations and particularly conveyancing. There are technology firms working to improve the visibility of the process and show the customer where they are in the journey. All well and good as far as communication is concerned but that doesn’t help the process move on if there is a weak link. If you are in a chain of five properties and one takes three months to complete, the fact that somebody else can do it in a week is irrelevant because you’ll still be waiting.
ID verification
A customer has to provide ID documentation to the estate agent, intermediary, lender and conveyancer. Could there be one ID hub instead that each part of the process can access? Regulation demands that each supplier is satisfied that their ID verification is robust but some lenders will accept the broker’s ID checks. It boils down to reliance on third parties and is that a regulatory issue or a risk appetite issue?
It was suggested that there could be some kind of block chain or ledger type technology that can allow information to be passed along with the case as it transfers from the first interaction right through to completion. We are quite a long way from that but it could be an interesting solution.
Risk based pricing
There was a discussion on the move from big data to small data. Organisations may be rich with data, but do they actually know their customers? If you have the right data about people this can result in risk based personalised pricing but we’re quite a way from that at the moment. Retention is the obvious place where you could start using your data to offer bespoke pricing based on risk and capital usage. It’s harder in the new business market because there’s a more competitive force.
So to conclude, technology is evolving all the time and 2019 looks like a year where even more technological development will enhance the mortgage and house buying process. From mortgage application through to back office systems and processes as well as post mortgage completion, there is a great deal of work going on behind the scenes.
Source: Mortgage Finance Gazette