In this blog, we explore how the lending industry can make credit more inclusive.
More than 5.8 million people in the UK are excluded from large parts of the credit market, either because they have limited credit history or because they have no credit file at all.
On top of that, an estimated 2.5 million people have been marginally declined – meaning that, despite having a full credit history, they’ve been narrowly rejected for a credit product through automated decision management that is based on scoring policies set against segments, as opposed to people. While only a small percentage of that group would likely default on their payments, lenders typically set a cut-off rate that excludes many potentially good borrowers.
In total then, that’s 8.3 million people who suffer from financial exclusion. For many, it’s a catch-22 situation. If you’ve never had credit, it’s harder to get credit. With limited options, many turn to sub-prime lending – at the cost of unfavourable interest rates. And the vicious circle is complete.
The FCA estimates that high-cost credit – with an APR equal to or above 100% – was used by 6% of all adults in 2017
As an industry, we can – and should – be doing more. Firstly, because improving financial inclusion will benefit society, secondly because this group represents a huge untapped market.
To offer better credit options, you need to reduce risk. To do that, you need to know enough about everyone to feel confident they can make payments without getting into difficulty. In short, perform more detailed affordability assessments.
New and alternative data
One way or another, traditional models have been failing this group and businesses. To change, there is a need to innovate. Looking at new data sources available, as well as the insight from trended data, is a great place to start.
Take the bank account data now available through open banking APIs. Bank account transactional data provides a detailed look at a person’s income and expenditure – the day-to-day reality of their financial situation. Pair this new insight with the tools needed to process and interpret it, and you get a much more robust view of a person’s affordability. Providing the consent is there, – which requires transparency, and work to ensure customers are aware of the potential benefits – we believe open banking, alongside other non-traditional data sources, could be transformative for thin-file and marginally declined applicants.
Such new types of data may even allow us to move these customers back into mainstream lending. For example, self-employed people will typically have small, frequent deposits into their bank account. Being able to identify and categorise these deposits as income will make lending much more accessible to this group and therefore much more inclusive. Other examples could include young people who are only just eligible for credit, or those who are new to the UK, who typically have a thin credit file.
Obviously, the use of this type of personal data is only possible with consent. Our research, which explores the attitudes of people to share their data, has clear evidence that people will share their data if there is evident value to them. And value in this sense is depicted in the role of easier access to finance.
Beneficial, new data
As well as bank account (open banking) data, potential new data sources include government data and rental data.
Rent is a serious financial obligation – according to the Office of National Statistics it accounts for around 27% of an average salary. So, it stands to reason that factoring rental payment history data into affordability assessments can help give you a more accurate picture. It will be instrumental in helping them to secure a mortgage.
New forms of analysis, ranging from psychometric questioning to advanced machine learning and artificial intelligence, can also improve automated assessments by uncovering person-specific insights that are traditionally overlooked. Of course, this potential is only possible with the best quality of data. In short, the better the data (including breadth of data incorporated into the decisioning process), the better the customer outcomes.
Frequent data updates
Updating bureau information more frequently could also lead to more accurate affordability assessments. Credit bureau receive refreshed consumer data in periodic batches. So, while lenders make credit decisions in real time, the information used to make that decision can be a snapshot of how each active credit account looked several weeks ago.
A lot can change in that time, and we believe more frequent updates could be especially beneficial in high-cost, short-term lending, when loans can be applied for, received and paid back all within the monthly-statement cycle. Real time CAIS is in our development plan, with the next steps to work with the industry to see how this can best work and the opportunities this can enable.
As data increases in volume and complexity, you need to scale up technology too. Data and the scale we’re seeing today requires advanced technology, such as machine-learning and categorisation engines. Work carried out by our DataLabs means we can now analyse up to 12 months of bank statement data in under a second, categorising income and sub-categories of income, as well as committed and discretionary expenditure. It is also proven to uncover previously unseen income and enhance scorecard performance. More specifically it can advance underwriter productivity (by 66%), performance of affordability (by 98%) acceptance rates, and bring more efficiency through the reduction of time taken to source credit scores, income and expenditure on consumers – plus much more.
We’ve talked about machine learning at length (see our guide: advanced analytics). But one pertinent advantage machines have over humans is that they’re incapable of discrimination of which is a particularly pertinent part of this blog. Like all our processes, our decision engine is built to be fair, accurate and transparent, with the customer’s best interests at the heart. We remove any bias from the data and the engine continually assesses it to determine whether any new bias enters the data over time. This process allows you to be certain of the customer receiving the best possible outcomes.
To ensure our processes and technology are used responsibly we use the FACT approach, meaning any output must be fair, accurate, customer centric and transparent.
As a side note, we believe machine-learning systems should be used with discretion. Transparency around how data is used is important, and as advanced analytics systems don’t lend themselves to easy explanation we would advise the use of more traditional models where these suffice. An example of this would be where a person’s credit history is solid enough to formulate a decision.
To conclude, new data alongside the requisite new analytical techniques and advanced technological power, offers significant benefit to all, especially consumers.
To make the most of this opportunity you need to ensure you:
1. Have the best data – including data assets (perhaps requiring new data), but also data quality
2. Partner with the right provider who aligns to your brand promise, but equally advocates the best and most effective methodology
3. The right decisioning capability to automate all of this, therefore enabling real-time, fair, accurate and efficient decisions that can be made, and acted on, in the moments that matter