When assessing applicants’ financial reliability, lenders globally have traditionally relied on a limited range of data sources. In the U.S., Canada, the U.K. and Germany, creditworthiness is determined primarily based on credit scores provided by large
credit reporting agencies (e.g. Experian, Equifax, SCHUFA). These scores are typically based on the applicant’s recent bank account and payment history, and his/her borrowing and repayment activity, with approval and the terms of the loan, including the loan
amount and interest rate, closely tied to the applicant’s credit score. Some other countries, including Brazil and Australia, are also transitioning towards this system based on positive credit scores. In other countries such as France and Japan, lenders focus
on employment history and corresponding regular income when determining applicants’ creditworthiness, while outstanding debt and featuring on negative lists tracking unpaid / missed payments often serve as detracting factors.
While these established credit scoring systems typically work well for the more financially active and well-off consumers, large segments of the population globally are unable to prove their creditworthiness through these metrics. For instance, in the U.S.
system, those with a limited or no credit history, such as first-time borrowers and non-citizens / non-residents typically fail to build a sufficient credit history or reach a high-enough credit score. Similarly, in France and Japan, those who are not employed
full time or do not rely on a single employer for their income (e.g. the self-employed, gig economy workers) might struggle to obtain a loan from their bank. Moreover, these credit scoring systems become significantly less inclusive when applied to developing
economies, where large segments of the society remain un- or underbanked, or without official employment income – producing very limited traditional data for lenders to base their credit scoring on.
The accelerating digitalization of transactions is producing an increasing variety and volume of consumer data, extending the pool of information that lenders can potentially use to determine applicants’ creditworthiness. Moreover, the emergence of open
banking infrastructures globally has helped facilitate data sharing between industry players, increasing the availability of consumer data. Utilizing machine learning technology, the increasing amount of data that is not necessarily related to applicants’
financial history is now increasingly used to build predictive models assessing creditworthiness.
Traditional lenders and credit bureaus have recently started to extend the pool of data their credit scoring systems rely on, with data such as utility and rent payment history increasingly included in credit reports. On the other hand, some FinTechs have
developed fundamentally new approaches to assessing creditworthiness, incorporating a wider variety of data sources.
Improving credit scores through alternative data
One area where innovation has focused on is helping lenders improve loan approval rates by offering supplemental information about their applicants that lack sufficient traditional data.
For instance, U.K. PropTech firm CreditLadder has built a tool enabling customers to use their rent payment history to improve their credit score at Experian and Equifax. Using TrueLayer’s open banking APIs, CreditLadder connects with the applicant’s
bank to access their rent payment activity, which is then incorporated into their overall score at the credit bureau.
Aire, another U.K. FinTech, connects with lenders via a real-time API integration, stepping into the online loan application process in case the applicant lacks sufficient data to prove his/her creditworthiness towards the lender. Through a virtual
interview, Aire’s machine learning technology assesses the applicant’s financial situation, spending habits, professional background and lifestyle to produce a behavioural profile to support the lender’s decision making. By increasing approval rates without
affecting the lender’s risk appetite, Aire has helped its partners distribute over US$10BN worth of credit.
Building alternative credit profiles
Other innovators are focusing on building a more complete profile of applicants that often serve as the sole source of their partnering lenders’ decision making process.
Singapore-based Lenddo uses non-traditional data to help consumers across South-East Asia, Africa and South America without a credit history build a credit profile. Lenddo uses thousands of data points from consumers’ digital footprints, including
their social media activity, browsing behaviour, geolocation and other smartphone data to assess their creditworthiness. Since its launch 4 years ago, Lenddo has helped over 5M applicants in 15 countries to access credit from partnering lenders.
Credit Kudos, a U.K. based ‘challenger credit bureau’ enables consumers to utilize their open banking data to build an independent credit report incorporating a wide range of financial data, including the user’s day-to-day banking and payment activity.
Following its recent partnership with AI technology firm Cybertonica, Credit Kudos will also be able to incorporate biometrics and behavioural analytics into its algorithms, making its credit scoring systems more robust.
Another limitation of traditional credit scoring systems is that they can typically only be applied locally. One FinTech which is tackling this obstacle is U.S.-based
Nova Credit, which uses data from international credit bureaus to help international students and professionals from 8 countries globally to build a ‘credit passport’. Customers are then able to utilise Nova Credit’s partnership network in the U.S. (incl.
American Express and IntelliRent) to apply for credit cards, student loans and other lending products.
Lending to the underbanked
While most credit scoring innovators use their algorithms to help other lenders better assess creditworthiness, some FinTechs are using their technology to provide loans directly to consumers.
For instance, U.S.-based start-up Tala offers short term microloans in Kenya, Mexico, India and the Philippines. Given the lack of traditional data in these countries, Tala’s credit scoring algorithms rely largely on the applicant’s phone usage patterns
and online activity to decide whether and at what rate to offer a loan. Similarly, mobile app based microlender
Branch relies primarily on data from its applicants’ smartphones, analysing their contact lists, GPS information, text and call logs, as well as their interaction with the Branch platform and customer service.
Although alternative lenders using non-traditional data are most widespread in developing countries, access to credit is far from universal in developed countries.
Deserve, for instance, offers credit cards to consumers in the U.S. without a credit history or social security number. Instead, Deserve assesses creditworthiness based on the applicant’s bank account activity, with regular income (from any source) and
regular on-time payment of bills/rent the main qualifying requirements.
Whether by supplementing traditional credit scoring data, building alternative credit profiles or providing credit directly, these innovative solutions are making credit more accessible and more affordable to previously underserved segments. At the same
time, alternative credit scoring systems are also improving risk modelling for existing lenders, making their algorithms that previously relied on traditional data more robust. As machine learning technologies improve by processing more and more data, these
predictive models will become increasingly reliable methods of assessing creditworthiness and are likely to be increasingly adopted by both incumbent and alternative lenders.