How UPST differs from fintech competitors

I recommend viewing Senior VP of business Dev at Upstart, Jeff Keltner’s latest video:

In the beginning parts of the video, Keltner explains clearly that fintechs began competing in consumer lending via the goal of lowering costs to achieve profitability and beat out the banks. They planned to operate online/digitally, to forgo the costs of physical bank branches (a bank branch typically needs $25 million in deposits just to stay breakeven). But, eventually this failed - it wasn’t enough.

Towards the end of the video, timestamp 25:37:
"what’s the best piece of advice I’ve gotten in consumer lending? A mental model that our confounder Paul G put out.

consumer loans are a simple equation:
default loss + operational cost + required return = cost towards consumer.

And lower cost to the consumer, wins.

It’s a simple framework for a complex set of actions but helps me think about the truth in lending all the time, and keep in my mind that I think of as the true north which is the consumer…we’re always thinking of better product offerings for the consumer"

If you watched the video from start to end, this last statement by Jeff Keltner really clicks in place. It goes to show why all other consumer lending fintechs hit a growth wall and failed (LendingClub, OnDeck, GreenSky etc). And those that are still growing, are doing so by taking excessive risks (like Upgrade, Avant). These fintechs were too short sighted; focused on short term numbers/growth, but not the long term on what really helps the consumer. Remember, companies that tend to focus on “consumer, consumer, consumer” typically win out in the end. It’s that customer obsession that Jeff Bezos outlined in his 1997 shareholder letter that brought Amazon its success (…).

Let’s go through examples:

OnDeck: “OnDeck, the once highflying online lender that fell on tough times made worse by the COVID-19 pandemic, is being sold to Enova International for about $90 million. After enjoying a huge surge following on its IPO in late 2014, OnDeck fell on tough times as competition heated up, the cost to acquire customers surged, and the money it made on loans dwindled. As of late 2019, OnDeck was in turnaround mode, trying to save a business that had a market value of $1.9 billion at its height. Then COVID-19 happened and its efforts failed to produce meaningful improvement. Cost-cutting eventually brought modest profitability, but investors were seeking margins more in line with those generated in the tech sector. At the end of June 2020, 39.5% of OnDeck’s loans were at least 15 days past due, up from just 10.3% three months earlier. In May, the company temporarily stopped originating new loans."…

OnDeck stock fell 95% from its IPO before its fire sale acquisition.

“GSKY is a financial-tech company that arranges loans for home-improvement projects and medical procedures. However…Rising loan defaults and concerns over future growth opportunities have prompted investors to rethink valuations for a class of companies that uses better technology to determine creditworthiness and offer loans over the internet and mobile phones.”…

And published yesterday:
“A federal watchdog agency has ordered an Atlanta-based financial technology company to pay a $2.5 million civil penalty and refund money to consumers after enabling contractors and other merchants to take out loans on behalf of thousands of consumers who didn’t authorize them. The CFPB issued a consent order requiring the company, GreenSky, to refund or cancel up to $9 million in loans for customers “harmed by its illegal conduct.” It also ordered the company to implement measures to prevent fraudulent loans.”…

GSKY stock is down 77% from IPO.

“When Lending Club went public in 2014, it was worth more than $9 billion…in 2016 when the founder and chief executive of Lending Club, Renaud Laplanche, was pushed out over loan sales that violated company policies. He and a subsidiary of Lending Club were ultimately charged with fraud by the SEC and fined, and he was banned from the securities industry. (He did not admit or deny the charges and has since founded another marketplace lender, Upgrade)
“When [the industry] started out they all said they were better [underwriters] than traditional lenders,” says Jennifer Thomas of the asset manager Loomis Sayles. The platforms thought that using technology to analyse new data, they could beat the banks at their own game. “Then the loans had tremendous losses.”…

“…While companies like LendingClub, Avant and GreenSky used their novel technologies to drive impressive user growth, they have ultimately failed to live up to their promises of disruption. All have struggled with funding costs. Without the benefit of their own deposits, like traditional banks, the fintechs have had to rely on strategies like securitization, which is not only costly, but in times of severe market uncertainty, tends to evaporate altogether.”…

LendingClub stock is down 86% from IPO.

As for the private company Upgrade, although it is growing rapidly it has taken on excessive risks.… Looking at the chart, UPGR was already seeing about 25% increased cumulative net loss rates than anticipated as of October 2020 for its 2018 and 2019 securitized trusts.

So what is Upstart doing differently?

Refer back to the simple mental model of consumer lending:

default loss + operational cost + required return = cost towards consumer

Reduce default loss:
No other lender uses true AI/ML underwriting models to find risk. For pools of securitized loans, Upstart’s realized loss rates were only ~half of those predicted by KBRA, and over the same time period the realized losses for the same pool of loans were only on average 5% different than UPST’s internal forecasts. (Goldman Sachs analyst report and see here:…)

Even COVID19 couldn’t remove Upstart’s lead in outperformance. A number of Upstart’s bank partners, many of which work with numerous platforms in the unsecured personal lending space, noted that Upstart’s loans were significantly outperforming loan loss targets and were the most stable in terms of performance from February into late 2020. (Goldman Sachs analyst report)

Required return:
As you know, fintech lenders bundle their loans and sell to institutional investors. But "Relying on asset managers seeking high yields has a downside…The cost of capital of the hedge funds is high and that pushes you into risky lending…In the short term, that can work…but in the long term, you have to figure out figure out how to compete with the banks.”
However, Upstart manages to reduce that cost of capital via confidence in its AI/ML underwriting. Upstart went from working with just six institutional investors in 2015 to over 100 today. Strong institutional investor demand for its securitized loans, thanks to data showing far lower loss rates vs other lenders, helps drive their desired return rates down, as they don’t need to compensate/anticipate larger than expected losses.

We also know Upstart is doing something unique versus all the other fintech lenders: it’s seeking to partner, NOT compete with all the 5000 banks in the country willing to use their underwriting model. Over the long term, this will be EXTREMELY powerful in lowering cost of capital. Bank retained loans are generally funded with depository capital, which has lower cost of funds (typically around 1%) than funds from institutional investors/hedge funds.

Reduce operational cost:
Upstart isn’t just focused on cost cutting by forgoing the bank branch; or simply using technology by online lending. With its various AI/ML models, Upstart has been able to reduce costs by driving increased conversion rates via frictionless/fast/easy verification and fraud models.
70% of loans are completely automated.
Its fraud rate has been a mere 0.30%.
Upstart also has AI/ML models to figure out which methods of advertising work best for each consumer, which reduces wasteful ad spend.
And, it has positioned itself to expand into auto via recent Prodigy acquisition which will reach customers at the dealership to reduce further marketing cost.

Its successful underwriting models in lower rates, increase conversion rates as well: Internal Upstart data suggests that each 100bp reduction in interest rate increases conversion by 15%.

“Everyone is chasing the same borrowers with 700-plus [prime] credit score, that usually come from the same marketing channels…So the companies running those marketing channels — the aggregators — take a lot of the profits." But Upstart doesn’t worry about that. Being able to accurately assess risk in the near/subprime consumers means they don’t need to always compete against the Prosper/Marlette/LendingClub/Sofi’s of the world (for these lenders, their borrowers average >700 FICO.
So, Upstart gets lower operational costs to pay to aggregators (at end of 2019, only 5% of Upstart’s borrowers had above 700 FICO score and as of March 2021 its borrowers averaged 674 FICO).

And finally: this goes without saying, but doing things illegally to juice growth/profits will increase operational costs in the long run with fines/getting shut down…but as we see from the above examples, many unscrupulous fintech lenders will do exactly that and commit fraud when they hit their growth walls (LendingClub, Greensky…and many other examples not mentioned above such as MoneyLion and OppFi).
Upstart has made it clear with its core company values “Do the right thing even when it’s hard” that I sincerely doubt they would ever go that route.…)

To end, I’d like to highlight the immense respect I have for cofounder Paul Gu…He was like 20 years old when he dropped out of Yale to create Upstart, and its his founding principles in lending that is driving the ‘north star’ of this company…and middle age 40-50 year old distinguished and experienced executives like Jeff Keltner and Dave Girouard take Paul Gu’s advice to heart!