The auto-lending industry stands to benefit from artificial intelligence in much the same way as insurance companies, particularly when it comes to underwriting and risk management. According to Deloitte, nearly $500 billion in new loans and leases are originated annually, and 86% of new car purchases rely on borrowed money. In this article, we discuss how AI startups aim to facilitate different processes within the auto-lending industry, looking to two well-funded startups as examples of what’s possible in the space:
- Zest AI: AI-enabled credit underwriting software for lenders. Its ZAML product could allow auto lenders to approve more car loan applicants without added risk for the lender.
- LendBuzz: A fintech company primarily serving immigrants and international students living in the US, providing them with auto loans despite their lack of credit histories within the US.
Zest AI – AI-Enabled Credit Underwriting
Zest AI was founded in 2009 and has raised $217 million in funding. In 2017, they announced their ZAML platform, which combines predictive analytics and natural language processing capabilities. The company claims the ZAML platform can help lenders process loan applications and ascertain applicant risk.
Using the ZAML platform, lenders could:
- Input customer data, including financial transactions, location data, or data from credit bureaus and assess creditworthiness and level of risk.
- Recognize financial and behavioral patterns within that customer data. Users can then analyze these patterns to determine how much risk a customer might pose to their company.
- Review automatically generated credit scores for customers and use this information to personalize pricing for each driver based on their perceived level of risk of defaulting on their loan.
Zest AI lists a case study on their website in which they claim they helped Prestige Financial Services reduce losses and loan defaults while maintaining the same rate of approval ratings. Prestige used Zest’s solution in addition to their existing underwriting methods. The case study states that Prestige saw a 33% decline in credit losses and a 14% increase in approval ratings for borrowers.
The 2-minute video below from Zest AI discusses how ZAML works. It emphasizes how the platform can help lenders maintain regulatory compliance while facilitating the underwriting process:
Lendbuzz – Auto Loans for International Students and Ex-Patriots
Lendbuzz is a Boston-based fintech company founded in 2015. Unlike Zest AI, which sells its software to auto lenders, Lendbuzz offers loans directly to consumers. The company has raised over $150 million mostly in debt-refinancing from several banks and insurance companies; although, they have raised $30 million in venture funding.
The company claims they can approve applicants with little to no credit history, in contrast to banks and more traditional financial firms that rely on FICO and other credit scores to approve or deny applicants. We’ve written extensively about the potential benefit that AI could bring to credit underwriting in past articles, and the ability to onboard applicants with thin credit profiles is certainly one of its greatest draws in lending.
That said, Lendbuzz takes a unique approach to marketing its software, focusing on ex-patriots and international students living in the US. These potential customers don’t have credit from US-based credit bureaus, but they need to purchase cars to get to work or school.
Lendbuzz claims to use an applicant’s background information, including their employment history and educational background, to determine the creditworthiness of an individual. They also state to use “personal background;” some AI-enabled underwriting companies will use an applicant’s public social media data to help determine creditworthiness, while others (Zest AI included) will use public record of court appearances and arrests.
Special Considerations for AI in Auto Lending
Although predictive analytics seems to be a fit in the lending industry when it comes to credit underwriting, the technology will likely face legal challenges before widespread adoption across the financial sector. There are many ways in which a machine learning-based credit model could inadvertently screen loan applicants by factors such as race and gender, which is illegal in the US.
The problem is that it can be difficult to determine how a machine learning model comes to the decision it does about whether to approve an applicant or not—something that is easy to figure out with more traditional credit models. Governments may look to regulate machine learning in lending to ensure that algorithms aren’t inadvertently “biased” toward certain groups of people and so that lenders remain in compliance with national regulations.
For more on the challenges of adopting AI for credit underwriting, including in the auto lending industry, read our article on AI Transparency in Finance.
Header Image Credit: PYMNTS