While virtually all lenders in the UK rely on traditional credit bureaus to assess a consumer’s risk, a new generation of companies are rising which are looking to utilise new technologies and techniques to supplement the traditional credit scoring method. Jonathan Minter investigates

---

For decades now the world of consumer credit has been underpinned by some basic underlying principles. One such principle is that looking at someone’s financial history and their current financial position will give you a decent-to-good idea of the risk an individual poses to a lender. While the system has evolved and changed over time, this basic principle is as true today as it was in the past.

For those who fit into the system, this works well enough. Mr and Mrs X have paid and never defaulted on their credit card, they’ve kept up with their mortgage and car loan, and have a history of generally being reliable. It makes sense that they are able to borrow at ‘prime’ rates, assuming they can afford it. Meanwhile Mr and Mrs Y have a slightly more chequered history for whatever reason. This increases the risk in lending to them, meaning finance houses lending to them need to charge a bit more interest.

However in today’s globalised world there are people who don’t always fulfil the criteria to receive a healthy credit rating for reasons that don’t necessarily reflect badly on their ability to pay back money lent to them.

This was one of the reasons alternative credit bureau Aire came about. Matt Davies, head of partnerships at Aire explains: "One of the reasons why the company came about is that one of the co-founders came over to the UK from New York eight years ago.

"He was working for JP Morgan, so had a really good FICO rating in the US, but because he was new to this country he had real difficulty getting accepted by any of the banks, credits cards, mobile phones or anything. There are more and more people coming into the country."

Aire isn’t alone in having spotted a gap in the market. Gideon Valkin, chief operating officer and co-founder of Friendly Score, notes: "The problem is: how does the world figure out a way to find creditworthy individuals outside of traditional credit bureaus? There are a lot of very interesting businesses solving this problem."

According to Valkin, this isn’t an entirely new phenomenon. He points to KrediTech in Germany, Capital One in the US, and Capitek in South Africa as companies that have looked at alternative ways of assessing someone’s creditworthiness, but notes in these cases, the companies were lending businesses.

What has changed is that companies such as Friendly Score and Aire are looking at different types of data from a credit bureau point of view, as opposed to a lender’s. Valkin says: "We believe there’s a need to have a new credit bureau for this type of data, tapping into lots of lenders and aggregating this type of alternative data – in order to aggregate and distribute this new understanding of creditworthiness."

These companies come in a wide variety of forms and guises, but all seek to use new technologies to fill in gaps which traditional credit bureaus do not currently fill.

Hello Soda

One such company is Hello Soda. Speaking to Motor Finance, James Blake, chief executive officer of the company, says Hello Soda is not a credit agency per se, but acts as more of a data and analytics business which supplements the traditional credit reference agencies with a different set of data.

A former employee of Experian, Blake says: "Credit data is still very valuable, but it only provides you with 50% of the story. There will be people who aren’t on it. So people new to the country, people who are a different demographic who don’t have a credit file. It doesn’t mean they’re a bad credit risk, it just means they’ve never had credit before. This is what we can help with.

"One of my frustrations, coming from the risk industry, was the information out there wasn’t enough to make decisions on. Businesses want to make good decisions and service their customers. But if the information is not there, how do they do that?"

Hello Soda offers a way for lenders to verify who they’re talking to using what they call ‘unstructured data’. Blake says: "This is information about people which can be gleaned from their use of social media, from what they are talking about, and the things they are going to do. And what they are saying about what they like or don’t like to do. However, all this information out there is very unstructured. What we can do is pull out the information and put it into a structure."

One example of how Hello Soda uses this data is to verify if someone works where they say they do. One initial premise was to look for clues such as the use of the word ‘work’. This proved to be problematic, however, due to the nebulous nature of language – someone may use the term to mean doing one’s day job, but also use it in the context of working out in the gym, for example. Instead, Hello Soda now looks at patterns and conversations and who people have as contacts on social media.

How it works

According to Blake, the average day of peak spending for consumers is about two days after their payday. Using data such as this, Hello Soda is able to look for evidence of facts such as consumers’ paydays. On this topic, Blake says that people tend to be quite organised and work paycheck to paycheck. He adds: "There are billions of records and pieces of information which are being filtered and returned as a mathematical algorithm that makes predictions."

Blake says Hello Soda and its clients are able to use examples such as this to help verify someone is who they say they are.
Using this data, Hello Soda also provides clients with information about customers which can be used to tailor customer contact – for example, the extent of their brand loyalty, or how often they want to be contacted by email.

This type of data can be updated more quickly than financial data, and Blake points to a friend of his who, for historical reasons, has relatively poor credit. This person has in the past two years found a stable, well-paid job and can now afford to go on holiday and buy a car. The risk in lending to him is lower than in the past, and his digital footprint will reflect this relatively quickly. However his traditional credit score will take time to reflect this.

In no way, though, does Blake suggest this new type of data will replace traditional credit scoring: "My view is you will always need the financial information. We’re looking to tackle very different problems. The credit information won’t be able to tackle a problem around personalisation, and it won’t be able to tackle a problem around life events and triggers."

Hello Soda isn’t alone in looking at people’s digital footprint to assess potential lenders. Friendly Score uses a mixture of social media and transactional data combined with algorithms in order to assess the likelihood that someone will pay back credit.

Explaining the basic principle behind the company, Valkin says: "What we’re looking at is someone’s online existence – their online profiles, the data they’ve put on those profiles, and information their friends have put on. We then see, from all these possible data points, how we can extract information to help us to predict if this person is going to be a good or bad borrower."

Like Hello Soda, Friendly Score is charting relatively new territory – social media is a relatively new field, and is even newer in terms of how it might be used for lending – and as such both Blake and Valkin say many companies need to go through a learning curve about how to best use the data accessible through people’s online profiles.

Valkin draws a parallel with other industries which are in some ways ahead in this respect. He says: "It would make sense in the same way that marketing businesses have realised that profiling someone according to the information that’s on Facebook or LinkedIn is really useful in predicting what they’re probably going to buy or not buy. We think the same thing is going to happen in this industry."

Friendly Score started in Poland, where an online lender started testing Facebook data against the outcomes of customer loans. According to Valkin, this resulted in lots of interesting observations, which were tested against other lenders. "The scores continued to be interesting; they then built a score card based on those initial findings. Basically anything that was statistically significant was incorporated into the score card," Valkin says.

The next step was adding LinkedIn, followed by Twitter and Google, and more recently transactional data – such as Paypal. New data sources take approximately two weeks to add, he said.

On a high level, Friendly Score is looking at three different categories. According to Valkin, the first is personal data. He says: "This is stuff that could be learned from filling in a form. So on LinkedIn it could be where they went to university, where they work, how long have they’ve worked there, how big is the company they work for, where they live, and so on."

The second element they look at is social network data. Valkin says: "This is quite new, but it’s clear there’s a lot of value in understanding someone’s social network in trying to predict whether or not they’re going to be a good borrower."

The third type of data Friendly Score looks at is behavioural data, that is, looking at how people behave online. A lot of this has to do with communication habits and how people use language.

Both Friendly Score and Hello Soda require access to someone’s social media data in order to assess a customer, and, as such, the customer has to agree to ‘opt in’, as opposed to traditional bureaus like Experian, which collect data on people, often with the consumer having little or no knowledge of this at the time.

Both companies face the problem of having to persuade consumers it’s in their interest to allow them to scrape their social media and online accounts.

According to Blake, being clear on consent is key to getting customers to agree to the access. He says: "We have 80% connection rates once consumers know what we’re doing and what we’re not doing with that data. We’re not looking at it and going ‘Oh, you were in the pub last night’. We deal in algorithms and scores. We don’t pass that information onto anybody and we don’t hold that information. We’re just enabling you to access the things you need for specific reasons. So if it’s insurance, for example, it’s to make sure you’ve got the most suitable insurance product for you."

Valkin adds that people don’t use Friendly Score for fun, and where there’s no incentive for the consumer to give access, conversion is very low.

Friendly Score isn’t being used by consumers who already have good credit with a mortgage and a credit card and so on. Instead, Valkin says: "We’re going to people who can’t get loans because they’re too young, they’ve never had a bank account or they’re new to the country and we’re saying: ‘If you do this, you may be able to get a loan.’ We’re also going to people more like myself and saying ‘If you do this you could get quite a good reduction on your interest rate because you may become a better credit risk’."

Aire

While Friendly Score and Hello Soda look to use existing digital material as a means of helping companies learn more about their customers, Aire takes a totally different approach by instead conducting virtual online interviews lasting approximately three to four minutes.

Describing the tests, Davies at Aire says: "The test is made up of a range of questions that could be drop-down, or ‘click on your favourite’, and so on. What we’re trying to understand is the basics about the person: where they live, what’s their profession, what they do, what they spend their income on, what they studied, and their expenses etc.

"And then an important part for us is the cognitive behaviour and lifestyle section, where we try and get an understanding of their attitudes towards risks, how they live their lives on a day-to-day basis, how they manage their finances, what attracts them to a particular brand of financial product, if they keep records of their finances, and so on."

A common question Aire gets asked, is how do they stop people from putting what they think is the ‘correct’ answer, as opposed to how they really feel. Davies says there are 50 questions which are cross-referenced to each other and correlated against each other to form a bigger picture. In most cases, these questions don’t have an objective ‘right’ or ‘wrong’ answer, and the system adapts to data on responses.

In addition, Davies says: "So if you do the test once, and then try again, it’ll change the questions and dig for more answers. We’ve also got a strong security layer where we try to verify the truth and look at fraud detection. This is a big layer for us because it’s all user-submitted."

Like Friendly Score, Aire seeks to supplement traditional scoring rather than replace it. Davies says: "The reason for that is the existing credit bureaus have a huge amount of data on you, me and everyone. It’s data that is collected in the background – we haven’t necessarily agreed to this information about us being taken It’s also historical data."

While many of the consumers using Aire have ‘thin’ credit files, this is not the only reason to look beyond traditional methods. Davies says it can also be used to assess the future potential of people who already have a credit history: "Usually if you’re not a thin credit file type of person, which I’m not, there’s no real forward-looking capability to traditional scores. It’s purely a score based on past performance. So it doesn’t necessarily take into account the possibility of my job improving. I work in mathematics in London, for example, which at the moment is doing well. So my prospects are comparatively good. So we can give this forward-looking view."

While in this example the use of Aire might cause the score to go up, Davies is keen to add: "While we can improve or enhance the score we don’t just qualify people for finance. It’s possible for people to get a lower score."

The use of interviews in order to establish creditworthiness is relatively new, and in order to ensure the results are as accurate as possible, Aire has been involved in a number of research projects looking at behavioural models.

Davies says the company operates with machine learning at the heart of its algorithms, and the behavioural layer of this was based on research undertaken with a professor from the University of Cambridge on how the behaviour one portrays is going to continue in the future.

According to Davies, customers will typically be brought to Aire when they apply for a loan or finance, but the normal credit provider doesn’t have enough information for the lender to be able to accept. In these case the consumer will be given the option to use Aire in order to re-score. If the consumer consents, they are connected to Aire, where they are tested and re-scored. Depending on the results, the consumer may then be able to borrow, or get a better rate.

Aire has only been operating commercially for approximately six months, and is in the process of increasing the value of the loans it will provide scores for. Currently Aire is being used for loans of up to £3,000. However as the algorithm improves, and lenders become more used to the idea of using psychometric tests to supplement traditional data on edge-case scenarios, or people with thin credit files, Aire will look to work on larger loans.

As with Friendly Score and Hello Soda, Aire is still improving its offerings. One way this will naturally occur is in feedback from lenders, who under the FCA will need to inform Aire of defaults. The company can then use this information to improve its algorithms.
When asked specifically about motor finance, Davies says: "We’re interested in looking at motor finance. It’s suitable, but we’re a small company, so we can’t target everybody just yet. It’s about ramping up, and building trust."

One factor repeatedly brought up by people Motor Finance interviewed is that these new types of risk assessment are convenient for those who’ve recently moved across borders. Davies, for example, says it can take up to a year to get a proper rating with a traditional ratings bureau in another country, adding: "And then you’re starting from the bottom up."

This can be especially true for people coming from the developing world, where Davies notes: "Some countries don’t have credit bureaus, some only have negative credit bureaus where they’ll only record if you’ve done something wrong. These countries are opportunities for us."

As with Friendly Score and Hello Soda, Aire sees itself operating in areas that traditional bureaus do not. Aside from international travellers, these companies can be used for the rapidly growing self-employed sector, which often has difficulty buying credit.

Lifestyle changes

Davies says these new companies are suited to modern lifestyles: "Lifestyles are changing. How many people will end up with a mortgage now? That’s reducing. How many people are now paying rent? How many people are paying for Spotify or Netflix? These sort of things are very much a lifestyle choice, but won’t necessarily get filtered into your credit score."

Valkin suggests something similar, when he says Friendly Score isn’t looking to replace the traditional bureaus. Instead, he says: "We’re interested in using new and alternative data to assess people that can’t be assessed by them. If you do that, we become a great complement to them, either as separate from them, or as one combined product."

In addition, he predicts that companies such as his will prove that these new data sources have predictive abilities for borrowers. This might lead to consolidation in the future, he says.

The past decades have seen an explosion of technological solutions to age-old problems, driven by a rise in the internet, computing and processing power and big data to name a few examples, and the pace continues to accelerate.

This is increasingly impacting the world of motor finance, where customers are conducting large parts of the car buying journey online, becoming less patient for lending decisions, and increasingly using mobile. Only the future will show whether the psychometric tests used by Aire or the use of social media by companies like Hello Soda and Friendly Score end up becoming as mainstream as the likes of Experian. However for now it is safe to say that the rise in digital technology is helping explore new and inventive ways of assessing credit risk and verifying identities.