Long reads

A vision for data driven lending

Tom Renwick

Tom Renwick

Propositions Manager, Atom Bank

“The past is a foreign country - they do things differently there" - so runs the infamous first line of L.P. Hartley's novel, The Go-Between. And so the unprecedented economic circumstances associated with Covid-19 are set to alter the lending landscape: lenders are dealing with material impairments, degradation of loan portfolios and spikes in forbearance requests. With the potential for cyclical lockdowns and business restarts on the horizon, lenders are having to monitor credit risk with limited visibility. Historic data sets previously used to drive a credit decision are less predictive of future circumstances.

Advanced data and analytics capabilities – the access to and mining of transactional data; real-time monitoring and advanced decisioning are integral to the solution. For traditional, incumbent institutions, the crisis represents perhaps the single most compelling opportunity to digitalise their systems. And for challenger banks and alternative lenders, an opportunity to steal a march: placing the plethora of open data sources available at the heart of their propositions.

Open banking – challenging information asymmetries

Bank ownership of customer data has long given incumbents a competitive advantage in terms of pricing and risk scoring. Yet as economic activity has progressively moved online, an increasingly long trail of data has been produced and advances in technology and financial regulation have made data portability economically viable. Open banking is challenging the existing competitive advantages associated with lending, curtailing incumbent banks' privileged access to customers and addressing the information asymmetry between borrowers and lenders.

Historically, existing bank account providers could better observe income, cashflows and existing loan performance than other potential lenders. Open banking is now enabling faster and better credit decisions: real-time granular transactional data can be mined to verify income, derive affordability, identify risk indications such as gambling and high cost lending, mitigating the risks associated with outdated information and lagging data. A broader picture of the risk profiles of applicants enables lenders to:

  1. Make faster, near real-time decisions by facilitating automated affordability assessments and replacing manual processes;
  2. Increase acceptance rates by reducing false negatives and allowing an assessment of thin-file applicants;
  3. Reduce defaults due to predictive forecasting.

The opening up of transactional data will lead to increased competition in UK retail banking and facilitate new and innovative customer propositions. It is likely that broad access to transactional data may ultimately lead to the unbundling of products that are typically sold together by banks at present, such as overdrafts and current accounts – upending traditional current account economics.

Innovative fintech lenders, overlaying their services over existing current accounts, will be able to anticipate when a customer is likely to go into overdraft and offer an alternative, better priced line of credit to replace long-term overdraft use. This competitive dynamic is likely to impact both the quantum and price of incumbent banks' overdraft products, which could lead to a material reduction in their profitability.

Financial performance data – unlocking the supply of credit for SMEs

Across the UK, small businesses often struggle to access the finance they need. The Competition and Markets Authority report that 50% of small businesses only consider one lender when seeking a loan, with 25% of them put off from ‘shopping around’ by arduous and frustrating application processes. The average time to decision for a small business is between three and five weeks, with the time to disburse cash significantly longer. This is in part because data asymmetries between lenders and would-be borrowers are particularly acute in small business lending.

Evaluating creditworthiness in the current crisis is challenging. Certain industries, such as wholesale food distributors, have not only fared well, but struggled to meet exponential demand. Others - hospitality and tourism – even the most casual observer cannot fail to have noticed a struggle. But within even the same sub-sector, some businesses will be more or less suited to a faster recovery. Individual business models are heterogeneous, those with a strong online presence may be able to ride out an exogenous shock. Lenders therefore must go beyond sector or even sub-sector analysis to assess the creditworthiness of individual borrowers.

But traditional sources of data often used in assessment are antiquated in evaluating business resilience. Making a credit decision based on last year’s annual accounts in today’s changing tempestuous environment is relying upon reprehensibly out of date data. Real-time, data-driven decision making is required.

To that end, a number of fintechs have developed integration-as-a-service offerings to access and process transactions direct from accounting software packages. For small businesses, the submission of digital, accurate data allows for a streamlined onboarding experience. For lenders, management accounts data offers a substantially richer, accurate and insightful view of company performance than out-of-date annual accounts. This data can be used to generate financial statements, affordability ratios, debtor and creditor concentration analyses – in real-time.

This data being made readily available to challenger banks and other lenders will allow for better access to more diverse and competitive sources of finance. In the future, owing to reductions in cost to serve, customers who are willing to share a combination of management accounts, open banking and ecommerce data are likely to benefit from more favourable pricing.

Towards personalised and dynamic lending

It has long been recognised that the application of machine learning methods has the potential to improve outcomes for both lenders and customers. Although risk based pricing – offering different prices for customers with different profiles is becoming the norm, traditional banks and lenders typically employ pricing models that pivot on product rather than the on customer: seeking to optimise the net interest margin associated with a particular product line or variant.

Marginally more sophisticated institutions may devise pricing strategies that maximise the lifetime value of a customer across all product lines.  But both approaches are blinded by a limited recognition of the customer needs.

Whilst exceptional digital experiences and service grab the headlines, incontestably, the less sexy issue of price remains one of the principle motivations for customers to switch banks. The rising threat of credit losses, impact of impairments and the anticipation of negative interest rate are compounding uncertainty in product-pricing performance for both the short and long terms.

Lenders need to assess the trade-offs between multidimensional business objectives: margin, bad debt and volume, whilst developing more sophisticated strategies that recognise that pricing models should distinguish between different customer profiles and needs. Access to richer data sets will allow for welcome opportunities both to optimise and enhance pricing performance and price more precisely and empathetically.

Covid-19 has provided an opportune moment for lenders to deepen their relationships with their customers. Innovative fintechs are now using increasingly diverse types of unstructured data to better understand customers and deliver more personalised, dynamically priced products and services.

Placing open data integration, categorisation and enrichment at the heart of propositions will provide lenders the ability to continually monitor customer needs, pre-empt borrowing and react in real-time to changes in behaviour – offering actionable nudges and personalised solutions delivered at the right moment. At Atom we are developing smart lending products which are tailored to small business customers by cashflow profile, with the repayment schedule reconciled to the predicted income profile of the business.

Looking ahead

Perhaps the last great bastion of valuable information is data on transactions and financial behaviour. Historically, unparalleled access to this data was the sole preserve of the big banks, owning the transactional layer provided them exclusive access to inbound and outbound activity. But we are approaching an inflection point – and Covid-19 has only accelerated the need for lenders to move from how they have historically managed credit risk, to a future state which is reviewing real-time financials.

The implications for banks of developing a detailed, timely understanding of the financial behaviour of customers are far-reaching: the ability to rebuild longstanding, and occasionally antiquated - pricing strategies; make more informed, speedier credit decisions and to design better and faster product interventions to support customers.

Whilst there has been an inexorable rise in its usage within financial services, there remains a long way to go to realise the vision of data driven lending.

Comments: (0)