Given the number of startups dedicated to its development, social credit scoring looks like it’s the next big thing in the financial technology industry. But what is social credit scoring – and why does it matter so much?
“The model construct that we label ‘social credit score’ captures a customer’s attractiveness or type as perceived by a firm based on social network information, in which the firm bestows some benefits that are monotonically increasing with type.” – Credit Scoring with Social Network Data, 2014
“Running with the wrong crowd will never help you.” – Ryan Cabrera, musician
Most people have heard of credit scoring, the process by which credit companies determine a loan-seeker’s financial stability. It hinges on their financial history, including credit card payments and previous loans, using them to show whether the applicant might be likely to default, or if there’s limited information on them – in either case, as a higher risk to the company, they’re likely to get a worse offer on their loan or credit card, if they’re offered one at all.
What you might not have heard of is ‘social’ credit scoring. An increasing number of companies – especially start-ups in the past five years – are adding to their client’s information banks directly from their phone records, internet history and even their choice of friends.
Major data analytics company FICO (based in the US, and originally named Fair, Isaac and Company) opened its doors in 1956 and sold their first credit scoring system two years later. Today, the FICO score is the basis of many major lending agency policies. It’s determined by previous payments and loans, as well as consumer micro-models and financial tests.
But now FICO are also looking into social credit scoring, using social media details and smartphone data. They’re not alone in looking for non-traditional data sources. TransUnion, a US credit bureau founded in 1968, claims they aren’t using data from social media, but are looking into alternative sources like payday lending information and even club memberships.
These companies share two things – information, and a competitive market. FICO bases their score on data from TransUnion, but also from Experian and Equifax, who are responsible for determining credit score in the UK.
As FICO and TransUnion make private developments in social credit scoring, Equifax and Experian will have to create their own approach or lose out on US business – and once they have their approach in the US, no doubt the practices will carry over and take over the UK market too. When the US credit market advances, the UK won’t be far behind.
It’s not a Western-only pursuit, either. The Chinese government have plans to install a nationalised social credit system – and to have the entire population of China individually ranked by 2020.
Different companies are using different metrics, but ‘social credit scoring’ as a process is becoming a widespread part of the loan industry. These aren’t scattered individual cases around the world – this is a global development in the finance sector. The consumer’s internet footprint and their personal data are more important than ever.
“If you look at how many times a person says ‘wasted’ in their profile, it has some value in predicting whether they’re going to repay their debt. It’s not much, but it’s more than zero.” - Will Lansing, FICO CEO
While financial records show whether someone has defaulted on a loan in the past, some companies are also taking to analysing their personality through Facebook posts and social interactions.
Many of them, including Singapore-based Lenddo and US-based Moven, both established in 2011, and Score Assured, founded in 2015, are looking for keywords – on the basis that if you say ‘drunk’ or ‘party’ too often then you’re less likely to be reliable, negatively affecting your score – but others are going a step further.
Believing that social profiles are easy to fake, but how you act online is not, TrustingSocial look for any anomalies in the way you use the internet and your mobile phone in the hopes of digging out fraudulent behaviour.
For now, they’re testing their system in emerging markets, as these places have fewer financial records available, hoping to tap into an ‘invisible’ market. If successful, however, it probably won’t be long before their system is used more widely.
Working along similar lines, Hello Soda have developed a technology called PROFILE, which they’re aiming to provide to other lenders. It analyses a customer’s complete social presence, using language processing and psycholinguistics in order to identify life events, dates of pay, location, age and employment, among others, and generate their social credit score.
What does this mean for credit scores? As a social approach rather than a financial one, it’s representative of a shift away from numbers on a page and a move towards analysing consumers as people.
“Character, for hundreds of years, was really the foundation of lending… [lenders] would talk to people in the community. What we figured out how to do algorithmically is quantify and understand someone’s trustworthiness.” – Jeffrey Stewart, Lenddo CEO
On the other side of the spectrum, InVenture and Friendly Score skip reading applicants’ posts and instead look at their browsing history and what sites are visited.
Searches for ‘payday loans’ or information on debt management stand out as their first ports of call in these analyses, but they’ll even check out your mobile phone call history, scanning for premium rate numbers and other financial companies.
Mobile-centric startup Branch (based out of San Francisco and Nairobi) specialises in using these phone records, through their app. Working along the same lines as InVenture, it checks SMS history, call log, and contact lists to establish financial security.
Africa is an important target for companies like these. Micro-loan company Tala Mobile, also targeting East Africa, uses over 10,000 data points including social connections, phone contacts, and even geographic locations in order to build a profile.
With their specialised software, Branch, Tala Mobile and others can even look at the transactional behaviour of the loan-seeker. While this is much more similar to standard financial checks than other social credit scoring methods, it has the benefit of taking into account the personal economics of the customer, as well as their overall risk profile. This means that even spending patterns factor in – your Amazon and eBay purchase history might be more important than you think.
“Your digital trail can establish your financial track record.” – Matt Flannery, Branch founder
Social credit scoring isn’t just about analysing who you are as a person – it’s also about making sure you are who you say you are. Fraud is a major concern for finance companies. One of their most important processes is ensuring that they’re not offering credit to someone who doesn’t exist and who will disappear as soon as they’ve parted with the money.
To this end, several companies use social verification services to ensure details like name, date of birth, email address, employer, university and even mobile phone number are all legitimate and correct.
Checking details isn’t the only way to catch out frauds. Form filling, for example, isn’t a popular activity, whatever you’re doing it for.
But cutting and pasting entries for financial applications can be an indicator of fraudulent behaviour, according to custom financial technology company Kreditech, who have offered programs for all these verification processes since 2012, adapting their methods to the developing technology market. One of the social checks they offer is checking whether information has been cut and pasted across multiple forms and applications – it’s tests like these that are increasingly playing a part in determining an applicant’s financial status.
Along with Lenddo and Moven, Kreditech also use social media as an enhancement for a more traditional credit check. They look for posts about paydays and employment status, to ensure that there’s a continuous and steady income with which to pay off the loan – that is to say, that the client has financial stability.
“As people graduate from college, they may have to put a deposit down on an apartment. They’ll need a car or a real bed. These are all credit decisions.” – Max Levchin, PayPal co-founder and Affirm founder
It’s not just credit companies that find social credit history useful. UK-based startup Score Assured provide the same sort of social media tracking service to potential landlords as well as to recruiters, with similar purpose – checking that the prospective tenant or employee is not only suitable, but that their information is all valid and up-to-date.
Because it requires the applicant’s permission, it has access to private data, as well as public – chat messages are entered into the algorithm for scoring, so it’s not a simple shallow public profile check.
One of the more unusual checks involved in the world of social credit scoring is that pushed by Affirm, a financial technology company set up by Paypal co-founder Max Levchin in 2012. At the forefront of the social credit scoring wave with a combination of the other tests and technologies, the company includes unusual, outlying data to build a complete picture of the people to whom they might choose to lend.
This includes, strangely enough, digital code library GitHub – Affirm checks to see if a loan applicant has contributed software code to their database. It’s not their only check, but it’s an example of the wide-ranging and obscure data they use to build a social profile for the credit check.
“At the moment Twitter, Facebook, Google and LinkedIn are the primary social platforms, although we also integrate data from services like Klout and PeerIndex. In the future, we’re also looking at integrating data from peer-to-peer networks and businesses like Kiva, to get to understand your charitable side.” – Brett King, Movenbank CEO
These techniques seem a logical progression into the digital age, but where could the technology go next? The true nature of social credit scoring is that it’s an evolving practice, barely in its infancy – even after several years of development.
While it seems a central part of modern life, for example, Facebook (which is necessary for many of these algorithms and scoring services) has only been prevalent since about 2007, and online for only three years before that.
It’s in a process of constant updates with new looks and new features. Accordingly, any system which relies on it has to be updated constantly as well, and with new facets of information added – one of the more obvious recent changes has included international location data to show where ‘friends’ are currently online around the world – there’s more data to be converted into credit scoring.
There’s another lesson to be learned from the history of Facebook: it went live at Harvard on 4 February 2004. On 1 October 2005 it expanded into 21 universities in the UK, having taking America by storm. Information services that succeed in the US quickly expand, including to the UK. Successful technology that seems a long way off geographically is rarely far off in terms of how it’ll impact our own lives.
Social credit scoring is part of the move into the digital age. Our online presence will dictate what can be tracked, and what can be tracked will tell credit companies whether we’re a good investment. Watch this space.