I’d like to know what everyone’s thoughts are on Openlending vs Upstart’s ongoing expansion into the autolending market. I see them as competitors.
Openlending (LPRO) was created in 2000. The company was formed to provide automated risk based loan pricing and underwriting for credit unions in near-prime auto lending and refinance. They collect fees/profit share for certifying loans. They also partner with insurance companies to provide auto loan default insurance for the credit union. They are business to business only.
They recently began expanding out of credit unions, to now include captive OEM lenders (two so far).
In Q1 2021, they certified 33318 loans versus 28024 loans in Q1 2020.
“More than 275 lenders used the Open Lending platform last year to originate more than $1.7 billion in auto loans.”
They are also a NAFCU silver partner for autolending decision platforms since 2019.
They state the return on assets (ROA) for the credit unions is about 2.25% (last slide:
https://www.nafcu.org/sites/default/files/education-conferen…
In one highlighted example, “The return on loans covered by Lenders Protection has been 2.82%, 54 basis points higher than expectation”.
https://www.cumanagement.com/articles/2019/12/loan-zone-auto…
They claim to have 20 years of data analysis and underwriting experience across 380000 loans ($8 billion), including through the Great Recession.
They say they have over 2 million risk profiles established from their data (slide 9). https://investors.openlending.com/static-files/345336b7-6abb…
Openlending creates a Lenders Protection Score. They “released the LP Score in November of 2018 allowing them to predict non-prime default rates with 99.2% accuracy.”
https://www.openlending.com/post/nafcu-services-names-open-l…
The score is from using Auto FICO and the following alternative data points:
LexisNexis Riskview score (Not familiar with this. ?anyone know much about Lexisnexis?)
Loan to value
Geographic location
Vehicle depreciation
Payment to income
As an example, “Harnessing the power of LexisNexis data, we might consider a 580 FICO score with a positive LexisNexis score more favorably than a 620 FICO borrower with a lower LexisNexis score and price accordingly. Even a 0 FICO score may be viewed positively with a strong LexisNexis score.”
https://www.openlending.com/post/the-power-of-data-in-near-p…
A key difference setting them apart from Upstart: they do NOT appear to use any AI/machine learning. Indeed, they are NOT actively hiring for any ML engineers. These are their only open positions currently:
Software Developer - Full Stack
Austin, TX • United States
Senior Accounting Manager
Austin, TX • United States
Database Developer Intern
Austin, TX • United States
Azure Data Engineer
Austin, TX • United States
Software Developer in Test (SDET)
Austin, TX • United States
DevOps Engineer
Remote • United States
Office Assistant- Part-Time
Austin, TX • United States
Business Intelligence Developer / Database Administrator
Austin, TX • United States
Their CEO has also specifically said “Artificial intelligence is another buzzword flying around the credit union blogosphere. Many think it will be leveraged to make decisions that are currently made by humans. Maybe someday, but today’s applications help your loan officers make better and faster decisions.”
https://www.openlending.com/post/data-technology-help-make-m…
Key differences with the alternative variables that (presumably) Upstart will be using with their machine learning algorithms:
Education history, job/employment type, cost of living, bank account transactions, repayment data flow, macroeconomic changes (like unemployment rate), consumer’s interaction with the loan application.
I speculate that Upstart will also use multiple models/algorithms in auto, modified from their unsecured loan underwriting for:
Fee optimization— optimizes assignment of origination fees
Income fraud— quantifies potential misrepresentation of borrower income
Acquisition targeting— identifies consumers likely to qualify for and have need for a loan
Identity fraud— quantifies the risk that an applicant is misrepresenting their identity
Time-delimited default prediction— quantifies the likelihood of default for each period of the loan term
Other commentary:
-Openlending was founded in 2000…and it took 20 years (!) for them to certify a cumulative 380000 loans?
If Upstart successfully replicates its flywheel model from personal loans into auto lending and exceeds 380000 loans underwritten within a couple years rather than 20 years, that will be a VERY powerful validation of the use of machine learning/AI.
-Openlending is focused on traditional credit union branches as a place of loan origination and does not seek to improve the consumer experience at all. Upstart’s mission however, includes ‘effortless credit’; they acquired Prodigy to streamline the consumer loan experience/net promoter scores with a seamless digital loan application either online or at the dealership, and they also will be performing “title management (placing and releasing liens, and transferring titles), disbursement of funds, and collections activity, including asset repossession.” I believe Upstart’s strategy for effortless credit and true risk will drive a better overall value proposition as it caters to both consumers and lenders, rather than Openlending’s sole focus on the lender’s side.
-It is interesting to see from LPRO’s latest quarterly report that one single insurance partner provides Openlending 65% of their total profit share revenue! Talk about customer concentration! (analogous to Upstart’s current concentration risks)
-The addressable market optionality for Openlending is very likely limited to Auto. They can’t expand out to personal loans, mortgages, etc without another 20 years of data collection in each category to build their algorithms, at least without the adoption of ML/AI.
-The 250 billion near prime market is likely big enough for both Upstart and Openlending to grow, despite being direct competitors.
I look forward to what your opinions are on Openlending and Upstart.