We’ve spoken at length about the advantages open banking will bring. How it will enable us to support our customers better, and improve credit access for those who have previously been excluded. How it will help drive competitiveness. But with more data in circulation than ever before, it’s easy to feel overwhelmed at the sheer volume of the information available.
Our recent business report, compiled using research we commissioned with Forrester, showed that more than a third of businesses are expressing concern over how they can navigate and innovate within this new complex data landscape.
Open banking, as we know, opens up a rich new source of bank account data, providing the income and expenditure information that’s invaluable in assessing affordability. It will enable you to better serve your customers by offering them more personalised and targeted services and propositions. It will also provide more accurate data specific to the individual rather than household, in line with the recommendations made in the Financial Conduct Authority’s (FCA’s) consultation paper on creditworthiness.
But, as with any type of data, to use it you must be able to extract the right insights. You have to do that consistently and fairly, on a huge scale. And you have to do it quickly, with little or no disruption to the customer journey. How?
Preparing for an open-data world
To draw meaningful insights from open-banking data, organisations need to be able to aggregate and categorise it. We believe advanced analytics, including machine learning, will enable lenders to achieve just that, at scale. We can do both parts.
Analytics will continue to play an integral role in scoring and extracting insight from data to inform decisions. It moves us fully into a world where data drives decisions, but with much more clarity and certainty than before. There is scope for everyone to benefit from the opportunities this brings.
Machine learning means teaching computers to analyse large data sets and find patterns and trends in the data to make predictions. By training a machine to process repetitive validation without the in-built assumptions of a human, you can focus your attention on the most complex lending models responsible for the biggest exposures.
What’s happening at Experian
At Experian, we’ve invested heavily in an open-banking platform. Our machine-based learning system can achieve the highest possible theoretical accuracy when analysing transaction data for the first time – data it has never seen before. We can now categorise twelve months’ worth of bank transactional data in one second and feed it into automated decision software, which better informs credit decisions and reduces the time to offer.
This brings a significant opportunity for you to amend your lending rules and create new rules based on a more intelligent and personalised view of your customers. And the data isn’t limited to lending decisions, either – it can personalise the customer’s entire experience.
The machine could identify an opportunity where you may need to have a conversation with the customer. For example, to tailor an APR due to a change in their circumstances, offer more suitable products, or identify pre-delinquency.
Perhaps even more excitingly, the engine is picking up on deeper trends in lender data that point to potential inefficiencies in today’s affordability calculations. They suggest an unnecessarily high rate of lender referrals and declines due to low confidence in the data, rather than the customer being a higher risk. The implication here is that automation can significantly reduce the cost of manual referrals and ensure data is being used to drive decisions.
The regulators’ stance on automation
In their Consultation Paper CP17/27 on Assessing Creditworthiness in Consumer Credit, published in July 2017, the FCA allows for flexibility, stating: “We do not want to discourage the use and development of automated systems that may provide more reliable results than asking the customer for large amounts of information […] At the same time, our proposed rules are non-prescriptive about the processes which firms may adopt.”
Many of the benefits of automation and advanced analytics align closely with the FCA’s objectives. It can be used to help protect against financial distress by picking up on early warning signs, or identifying spending patterns that are associated with certain types of vulnerability. It would add more sophisticated insight into what is genuinely affordable for each individual customer, and would bring more consistency to the way affordability is assessed. It would also improve competitiveness by opening up the market to new entrants, such as FinTechs, who are already using new technology to create better customer experiences.
Understandably, there is some nervousness around the role of analytics in use cases such as affordability. We appreciate some lenders may continue with traditional models; bureau-based assessments, and some will combine both bureau and open banking. A combination of both will likely bring a much more granular level of detail that helps in making decision, particularly for marginal decisions or consumers with complex financial lives.
Digitising traditional process
Machine learning enables bank transactions to be categorised and aggregated to offer a single view of an individual’s spending traits and patterns.
By merging credit information with statement information, we can provide fresh insight on a customer’s capacity to afford financial products and services in an affordable way. Categorisation of a person’s financial circumstances, such as combining income and expenditure with savings and pension and investment information, will provide a much more holistic approach to finance.
Information sharing has the power to revolutionise current practice. In the future, lenders will be able to run an affordability test across their whole portfolio to give an early indication of financial distress. Processes which are currently manual, such as a bank statement check for a mortgage application, can now be automated or put online to reduce errors and cost.
We believe machine learning has huge benefits to the integration of open data into financial assessments. But broader than that, we believe advanced analytics such as this is the future of lending.
Miles Cheetham, Head of Customer Insight at the Open Banking Implementation Entity (OBIE), says that: “Above all, we’re looking for better outcomes for the customer. We’re making sure banking is fit for purpose in a 21st century digital world, underpinning our digital economy and leading to greater growth.” Let’s make sure our technology is fit for that world too.