Last year, Upstart Holdings (UPST 2.76%) faced challenges in finding buyers for the loans its AI-powered platform approved. The fintech company also saw a drastic slowdown in consumer lending, with muted demand for its offerings from banking partners, which added to its struggles.
Some of these problems were alleviated in May when Castlelake, a global alternative investment manager, agreed to buy up to $4 billion in loans from Upstart. The news triggered a massive rally in the stock, which gained more than 400% in a few short months.
The stock gave back some of that gain following the release of its second-quarter report in August. The lender saw a continuing loan decline through its platform during Q2 amid a challenging economic backdrop. Although loan volume fell, one number from its recent report stood out to me: 87%. That was the percentage of the loans Upstart originated for which the process was automated end-to-end, requiring no human involvement at all. Here's why this metric is so critical to the company's long-term success.
Upstart's automated process and AI lending models could change consumer lending as we know it
Upstart uses its home-grown artificial intelligence (AI) models to price credit risk and help more consumers access loans than ever before, and often at better rates. Its goal is to improve on Fair Isaac's traditional FICO scoring model, which it believes shuts some worthy borrowers out of the financial system, or rates them as riskier than they deserve.
One crucial component of Upstart's business is the highly automated nature of its lending activity. By automating the processing of most of its loan requests, Upstart can operate more efficiently and increase its contribution margin, which is its net revenue from fees minus costs for borrower acquisition, verification, and servicing.
Upstart management keeps track of fully automated loans as a percentage of total loans originated. This number represents the number of loans originated where at no point an employee has to get involved -- from the initial rate request to the signing of the loan agreement. Loan automation is one reason Upstart could quickly scale up from 300,000 total loans in 2020 to more than 1.3 million in 2021.
Loan growth in recent years has slowed down due to other factors, but management has continued to automate much of the lending process. From the beginning of 2021 through its most recent quarter, the percentage of fully automated loans at Upstart has risen from 71% to 87%.
Data source: Upstart Holdings. Chart by author.
Management has achieved this while maintaining low fraud rates by increasing the accuracy of its AI models, eliminating previously manual processes, and growing the number of repeat borrowers on its platform. While there is a limit to how many loans can be automated, this high rate should help Upstart maintain healthy margins. Its contribution margin grew from 47% in Q2 2022 to 67% in Q2 2023.
Upstart faces some challenges in the near future
Upstart's models continue to perform well compared to the traditional FICO scoring system. Thus far, the fintech has done a solid job of identifying borrowers with lower default rates across the credit scoring spectrum, showing that it is doing an excellent job of identifying credit-worthy borrowers that FICO passes over.
However, it faces challenges in the near term due to the economic conditions that have significantly slowed down demand for consumer loans. According to Federal Reserve data, U.S. banks have reported tighter lending standards and weaker loan demand from businesses and consumers in the second quarter. This slowdown comes as banks cite a more uncertain economic outlook and expected credit quality deterioration.
This slowdown in demand has significantly impacted Upstart's business, which has seen the transaction volume of loans (in dollars) in the second quarter fall 72% from last year. Fellow consumer lender LendingClub experienced a similar hardship as originations dropped 48% in the second quarter.
Upstart is in a highly cyclical business. When interest rates are low and loan demand is robust, it can make more loans, earn more fees, and deliver stellar earnings as it did in 2021 and 2022. However, when interest rates rise, and demand for consumer loans is tepid, it puts a lot of pressure on its top and bottom lines.
The company has worked to diversify away from personal loans and into larger automotive and home lending markets. It has forged partnerships with 61 auto dealerships to offer its loans and recently began dipping its toe into the home loan market by offering home equity lines of credit. These markets, plus personal lending, give the company a $4 trillion market opportunity.
Image source: Getty Images.
A stock with excellent long-term prospects if it can navigate near-term challenges
Upstart's automated loan process continues to improve and could be a key driver of its return to profitability and long-term growth in auto and home lending if it can navigate headwinds from a slow lending market and tepid demand for its loans.
However, if the U.S. enters a recession, we could see a further slowing in consumer lending and an uptick in delinquencies. How Upstart's models perform in that particular environment is a big question mark, and its models will face their biggest test ever if we get a deeper recession than what experts predict. Upstart stock price reflects investor concerns and trades at around 5 times sales, which is on the low end of Upstart's valuation since going public.
Investors should take a patient approach to the stock right now, keeping an eye on its model performance over the next few quarters in the context of the broader economy. If we get a recession and Upstart's models prove better than traditional credit scoring, it could gain favor with more lending partners who want in on its success -- which could be a good time to build a position in the fintech.
