Consumer lending: understanding risk models, regulations and business opportunities

What is the state of play on consumer lending in Europe, and why is this product so important for banks?

Consumer lending is extremely popular within Europe, with more than EUR 1,000 billion being lent to people at any time. UK and Germany, for example, lead this market with a total of over EUR 500 billion in circulation. And banks love this market, as they underwrite most of it: the total average interest rate is above 5%, with refinancing costs at a nearly zero level. Furthermore, the regulators ask for only 8% as equity capital to secure the overall lending sum against risks. Some financial institutions generate a return-on-equity ratio of above 70% in this business, which is a great number for a time when a lot of the other business models do not work for banks anymore.

How does this business work today? What are the risks involved?

Money lending is a highly regulated business. Lenders have to adjust their models to adhere to regulatory authorities, such as BaFIN in Germany, or the Financial Conduct Authority and the Prudential Regulation Authority in the UK. These financial supervisors provide specific requirements for the risk models, regarding consistency and explainability, therefore, most banks prefer to use simple models based on a maximum of only 20 data points. Often, these risk models include credit agency scores as a somehow neutral proxy for the creditworthiness of a consumer.

In any event, what the banks really want to know is the future degree of repayability of a consumer loan: is the borrower able and willing to pay us back? Ability is about creditworthiness, Willingness is about identifying fraud. For this reason, as a bank, you need to address both issues to make sure you cover default risks appropriately.

How do the PSD2 and Open Banking affect risk management within consumer lending?

PSD2 and Open banking create a wealth of data at the fingertips of the consumer, and the latter is enabled to decide with whom to share their account information. Credit Agency scores will be less critical since banks can tap into recent customer data and evaluate first-hand information. A consumer bank account data from the past six months provides a good understanding of the financial behaviour of that person. Transaction data can be classified, and machine learning models will extrapolate spending habits and compare them to the borrowers income situation.

On the contrary, consumer credit agency data is often outdated - it will provide historical data, but no recent information or an actual financial snapshot. Here, at Risk42, we evaluated more than 500.000 consumer loans, and we found out that a risk model fed with a combination of a consumers bank account data and the corresponding online application data will by far outperform old legacy models based on credit agency score data.

Even better, these new scoring models create new market segments for banks, by making it possible to serve customers that do not have a credit agency history, or people with a bad credit history but in recent good financial standing. A vast majority of consumers are willing to provide credit lenders with their bank account data, if, in return, these lenders make a decision on the customer loan in real-time and transfer the credit sum instantly. There is a significant new customer segment with attractive margins for the bank, and a great opportunity for customers that have been rejected by the legacy models used today.

Are there any challenges in creating these new risk models?

There is an added risk, since bank account data reaching back only for a few months can be falsified, and there is also a risk of bank account takeovers not being recognized. To deal with these cases, you have to add separate fraud detection models evaluating specifically the likeliness of a fraud attempt. Within an online credit application, there is extra data you can add to these fraud detection models, for example which device the applicant is using and how it is being used. Moreover, the patterns of how accounts are being used after a takeover.

In conclusion, it will be a clever combination of both new credit scoring models based on user-provided data as well as separate fraud detection algorithms that will make credit agency data obsolete within consumer lending.

About Roberto Valerio

Roberto is Founder and CEO of Risk42 Software, his second successful venture within the Risk & Anti-Fraud Software industry. For the last years, he has played an active role in online payments and risk management. Furthermore, Roberto provided numerous publications regarding new tools and methods used within the fraud prevention space. He was part of the European Advisory Board at the Merchant Risk Council. His background is within business administration and programming.

About Risk42 Software

Risk42 offers a sophisticated credit & risk scoring platform to enterprise customers within e-commerce, financial services, and retail banking. Clients are Payment and Financial Service Providers, Retail Banks, and Price Comparison Sites for Financial Products. The company was founded in September 2018 and is headquartered in Hamburg, Germany. It serves large markets within Europe.