UPST: auto refi

So, I admit I was initially puzzled as to why Upstart was focusing so heavily on auto refinance rather than auto purchase lending, especially as they acquired Prodigy earlier this year, which is really about car purchase loans.

Well, I watched an auto lending webinar given by Upstart to prospective bank partners plus reviewed a couple brochures – and it makes sense. Auto refi is the lower hanging fruit for UPST, possibly with better unit economics than auto purchase.

I’d like to share my notes as follows:

Val Gui gives the webinar talk. He is the VP of automotive lending at UPST. He has 10 years of automotive experience, including managing 12 car dealerships that had combined revenue 450 million. Has past hands on experience in a startup, was cofounder/COO of instamotor (which focused on auto financing).

Banks struggle to build a large profitable portfolio of auto loans, although the market is massive.

Building a robust auto loan segment helps with adapting to the current environment of:
-Record cash deposit levels
-Demand for loans overall was falling due to stimulus
-Branch traffic is down, exacerbated by COVID
-Low interest/low yield environment
-Accelerated consumer demand and acceptance of digital interaction

There are two auto lending segments of opportunity with current trends/macro environment. Auto refinance, and auto-loan origination

There are over 1.37T outstanding auto loans. (It’s actually 1.42T as of the latest Fed report)
Auto refi comprised 4.5% of auto loans in 2019 (1.2 million transactions) vs 95.5% new auto loan accounts (25.1 million transactions).

Auto refinancing has not been well explored as a growth channel, despite the following benefits:

  1. easier to find/market to consumers who have existing auto loans, as it’s much more difficult to find the consumer who is seriously in the market to buy a vehicle,
  2. less risk to lender as consumers have a track record of making payments on existing loan,
  3. cross-sell opportunities available when you save the consumer money with refi and establish trust.

Val Gui gave a webinar poll to the banking audience:
47% say they offer autorefi but it’s adhoc/not significant (just a product of convenience).
29.4% don’t offer but would like to.
23.5% market it and it’s a meaningful part of portfolio.

Why is auto refi difficult? Barriers to entry:
lien perfection increases transaction costs,
smaller loan sizes make unit economics unprofitable,
and the entire process can be complicated for consumers who might need to make trips to DMV, notary, etc.

Well, the angle of attack to this is just like with personal loans:

  1. lower the rates to reflect true risk as many loans are overpriced, and 2) make it a frictionless customer experience.

Digital innovation drives greater efficiencies (Savings on DMV registration, titling costs etc that typically eat into margin)
Upstart’s AI/ML models reduce marketing costs by knowing which borrower to target, lowers friction with instant verification, and lowering loan rates.
Upstart can also service the loans by identifying at risk borrowers and proactively outreach to those at risk/manage potential losses.

The crux of this, as usual, is the ability to reduce rates without increasing defaults, as consumers want to reduce monthly payment expenses. That is their primary focus.

TransUnion reported that of consumers who refinanced, 86% saved over $10.
This indicates a willingness for consumers to lessen monthly expenses, even if the amount is as low as $10.
The report added that many refinanced for less than $30, underscoring these were not edge cases.

This was confirmed with Upstart’s studies which found that:
27.5% of auto loan borrowers can save over $20/month.
12.2% can save 0-$19/month.
60.3% likely have no auto-refi opportunity given their existing loan APR is not mispriced enough.

(Keep in mind, these refi offers are with keeping the SAME loan term and not extending loan duration, so it actually saves the borrower money overall - unlike other predatory refi companies.)

Upstart has found that $20/month savings is the threshold where borrowers generally find it meaningful enough to accept auto-refi.
Some do accept auto refi savings of less than $20/month, but not as many.
However, marketing dollars should be targeted to those who are most likely to accept (the $20+/month savings group).

This means there is a massive opportunity to grow from today’s 4.5% refi share of auto loans into a share of 27.5% (expand the market by a multiple of 6).

27.5% of 1.4 trillion outstanding auto loans = 385 billion auto refi opportunity.
Assume $20,000 average loan size, and that’s 19.25 million consumers to target.

Moving onto auto loan origination (purchase auto lending):
There are over $600B in auto loan originations annually.

35% of auto loans are direct purchase, which are profitable but hard to grow.
This is where customers go directly to the bank/branch to get a loan, and then go to the dealership with that financing. But dealerships can flip that loan away from the consumer who physically walks through the door, by offering a different deal.

65% of auto loans are indirect purchase. These are so heavily competed, there are only razor thin margins.
This is where the consumer goes to the dealership without financing arranged. Then, the dealers are acting as an intermediary by offering loan options from its banking partners. The dealer ultimately decides if that consumer is provided your loan for financing or some other bank.

What you need to win in purchase auto lending:
-instant pricing for all applicants
-Pricing/approval via soft credit pull
-Instant offer re-evaluation for multiple vehicles/multiple structures
-True per offer, risk based approvals and pricing

Like with auto-refi, Upstart’s platform offers all of the above.

Upstart’s Prodigy can also help benefit the dealer, bank, and consumer:
allows dealer to interact with consumer on their own terms,
allows consumer to shop online/offline,
allows the bank to compete directly via loans offered on platform

Webinar and brochures:……


Another terrific Upstart analysis. However, the one thing I didn’t see is just what is the potential revenue from this opportunity? What fee would Upstart collect on a $20k loan? Of the 19.25M existing loans what percentage would be a reasonable estimate for Upstart to place with lenders? Once the bow wave of existing loans that will potentially be refinanced are placed, what is the anticipated annual run-rate?


Revenues per auto loan will appear larger due to auto loan sizes being larger (on average probably ~20K, which is double that of personal loan size at ~9K).
But the margins for auto is likely smaller than personal loans, because after paying borrower acquisition costs they still have to account for title/lien/dmv etc fees.

Currently, management is guiding for similar dollar contribution profit per auto loan, compared to personal loans.

Q2 call:
"But with respect to auto…It’s still very early days…with respect to margins, I think it’s still very early. We’re still in the early stages of engaging with banks and engaging with investors and understanding the effects of fees on the conversion funnel and things like that.

But I think our expectation remains that – the loans are clearly bigger and so there will be a higher dollar revenue per loan.

And we’re expecting probably a similar dollar sort of contribution profit per loan as we see in personal loans.
So the percentages will be a little bit different because the loan size is bigger, but that’s our current going assumption."

There’s 19.25 million auto refi customers to target. On a dollar contribution profit basis, autorefi market alone is going to be roughly 5 times the size of the nonprime personal loan market (if we assume they get the same profit per autorefi as auto purchase loans)

I think the biggest risk to UPST is the speed of auto roll out.

If 10-15% sequential quarter to quarter rate of growth in personal loans is sustained into 2022, by the end of next year they might have 2 million total personal loans transacted for all of 2022.
At 10K size per loan, that’s $20 billion.
The personal loan TAM is about $90 billion in originations annually.
About half of that is the non-prime segment that UPST can realistically capture. THat means UPST would have a near 50% market share of that segment if they achieve 2 million loans in a year. I dont’ think the rapid personal loan growth can keep up past a certain threshold market share.

Q2 Call:
“We started in January, offering our auto refinance product in a single state, then expanded to 14 states by the end of Q1 and have now expanded to 47 states…Upstart-powered banks have now originated more than 2,000 auto refinance loans in 40 different states.”

It is critical they move faster into auto lending. However, they’ve only completed 2000 auto refi loans since January 2021. They currently have 5 bank partners in auto lending.
And although their prodigy platform has transacted 1 billion in vehicles in Q2, we don’t know when they’ll bring auto purchase loans onto that platform.

Auto guidance is what I’m paying close attention to over the next few quarters. We can hope the stellar management team conitnues to execute, but never count on hope. We should rely on the numbers they provide.


Very insightful post jonwayne235, thank you. I have a question with regard to your comments about the non-prime segment of the personal loan market:

“The personal loan TAM is about $90 billion in originations annually.
About half of that is the non-prime segment that UPST can realistically capture. THat means UPST would have a near 50% market share of that segment if they achieve 2 million loans in a year. I dont’ think the rapid personal loan growth can keep up past a certain threshold market share.”

That would translate to $45 billion for the non-prime segment. Do you think it’s possible that Upstart’s AI platform could expand that TAM due to it’s superior ability to recognize borrowers in the non-prime market who are acceptable loan recipients?

That is to say, is it possible that the reason the non-prime TAM isn’t larger is because traditional underwriting parameters are hampered by the limitations of using just the FICO score? Resulting in otherwise qualified non-prime borrowers being denied loans? And that Upstart’s ability to approve more of these borrowers could expand the TAM?

Although even if it did I agree that saturation in the personal loan market is definitely a situation to keep an eye on.



Do you think it’s possible that Upstart’s AI platform could expand that TAM due to it’s superior ability to recognize borrowers in the non-prime market who are acceptable loan recipients?

Mike, it’s a good point to think about and very possible that previously rejected (but creditworthy) nonprime borrowers are now given loans via UPST that they wouldn’t have received from any other lender.

We also know repeat borrowers are becoming an increasing portion of their revenue each quarter, which decreases borrower acquisition cost and itself might even be expanding their market segment.

However, we do know there remains a macro headwind against the personal loan market ever since COVID. Perhaps this can reverse going into 2022, but macro is generally difficult to predict.

Best case scenario for our company would be TAM increases due to Upstart literally creating its own markets with nonprime personal loans expansion as you described, and “making the market” in auto refi/auto loans.
As I had mentioned previously, auto refi is <5% of auto financing transactions but 27% of all outstanding auto loans are potential targets for UPST.

I also watched a presentation by Experian on the ‘state of auto finance market in Q1 2021’ (you can watch here…) recently and they had data showing subprime borrowers in the auto market have been squeezed out due to high prices since COVID.

For example, deep subprime & subprime used loans remain near record lows with prime/superprime comprising nearly 54% of used loans. In 2019, prime/superprime were 48% of the used loan market.

Meanwhile for new auto loan financed car purchases, average credit scores of approved borrowers have surged from 718 in 2019 to 726 in 2021!

We can hope Upstart will be able to draw the creditworthy portion of these nonprime consumers back into the auto lending market with its AI models giving them their rightly deserved lower APRs.

I like to stay conservative though, and so my investing actions in UPST is going to be rooted in the numbers/guidance provided each quarter.


I have come to believe, by reading up on stories you can glean in some of the trustpilot commentary, that upstart is actually making a new and underserved market.

I have a buddy who was starting his own business… for the first three years, when he wanted loans, no bank would give them. Now that he is successful they come to him and offer him loans, lines of credit, and at almost 0 interest rate – and he has no need.

The point being that there is a whole segment of the market that needs access to credit, are worthy recipients and they cannot get it because the risk systems for banks are broadswords, not scalpels. It is just not worth it for legacy systems to identify and provide those loans, because it would be people intensive, and they lack the data, or lack the data in an organized, effective system.

To me, it’s a beautiful thing that Upstart is doing - they are taking a massive efficiency gap and growing it into a new market and likely improving the lives and economics of whole populations. It’s something that should likely show up in CFPB and other government agency statistics, eventually.