You are here

Consumer Lenders Embrace Alternative Data, Machine Learning to Be Competitive

Unsecured personal loans are hot. Consumers like them because they have a simpler repayment structure than a credit card and are quicker to obtain than most other loan products. Noticing this demand, lenders (both banks and non-banks alike) are increasingly offering these loan products. Even lenders who have been focused traditionally on credit cards are now offering these closed-end installment products.

A recent white paper I wrote for Equifax looks at the challenges that lenders of unsecured personal loans are facing, whether they use alternative data or alternative credit scores for marketing and decisioning, and the extent to which they have incorporated the use of advanced analytics or artificial intelligence (AI) techniques into their business.

Here’s a quick look at what I found, based on my survey of 18 of the largest U.S. unsecured personal lenders. You can download a free copy of the report here for more details (courtesy of Equifax).

Challenges: Lenders making unsecured personal loans expect external fraud risk to be their top challenge in the coming year, as tactics used by fraudsters applying for loans online continue to evolve. Other key challenges involve identifying and attracting new qualified borrowers and retaining existing customers, reflecting the competitive nature of this part of the consumer loan market. The biggest wildcard, however, is whether these lenders will have to deal with a long-expected, but not yet realized, economic downturn. Many of these lenders are already making plans for this eventuality, since they expect that financially stressed borrowers will prioritize other payments (such as those for cell phones and auto loans) before paying back unsecured personal loans.

Use of alternative data and alternative credit scores: The use of multiple types of alternative data for marketing and loan decisioning is increasingly table stakes for unsecured personal lenders, particularly those that are non-banks and originate substantially all of their loans online. The 72% of consumer lenders in Aite Group’s survey that use alternative data report that the key factors leading to their adoption include the need to better assess the risk of a given applicant and the ability to identify more qualified applicants and offer better terms than they could using traditional data alone. When evaluating a certain kind of alternative data, the accuracy and reliability of the information, as well as the lender’s confidence that the data is compliant with applicable regulations, are top of mind.

Advanced analytics and AI: Lenders of unsecured personal loans are increasingly incorporating advanced analytical approaches and AI techniques such as machine learning to better harness the large amount of in-house and third-party data that is available for marketing and loan decisioning. The majority of consumer lenders surveyed are using these techniques, and additional lenders are planning to incorporate such use in the near future. A major caveat to this, however, is how lenders are using machine learning for loan decisioning. Most consumer lenders are using machine learning to identify which attributes should be incorporated into underwriting models rather than using it directly in the models themselves. The key roadblock for these lenders is the inability to explain adverse actions.