UPST delinquency rates?

Has anyone been keeping up with Upstart’s current securitized loan delinquency rates?
I just looked on KBRA to check on them and this is looking like an enormous red flag to me…

On January 18, 2022 a new pre-sale report came out from UPST onto KBRA’s site (https://www.kbra.com/documents/report/61852/upstart-pass-thr…).

I put a screenshot from the report here (https://ibb.co/RNR8vh1) to help visualize this but it is a very disturbing trend. Delinquency rates on loans packaged before 8 months ago are exhibiting nearly twice the rate of delinquencies seen on loans packaged in 2020, 2019, 2018 etc that had experienced by the same months of seasoning.

For example, see below. I compared 2019 loans to 2021 loans at roughly same number of months seasoned. We see at 5 months, the 2021 loans (4.29%, bolded in table) are 1.84 times more delinquent than the 2019 loans (2.33%, bolded in table).

Deal      Months  Delinquency%
2021-ST3  8       3.85
2021-ST4  6       3.71
2021-ST5  6       4.03
2021-ST6  5       **4.29**
2021-ST7  4       3.38
2021-ST8  3       2.16
2021-ST9  2       1.04

2019-2    5       **2.33**
2019-1    4       1.90
2019-3    1       0.49

I specifically chose the 2019 set of loans to compare as that would strip out COVID/stimulus; which in theory would benefit the 2020 loans. Although I think the 2021 loan delinquency rates are just as worse when compared to the 2020 loans!

And, when I look at competitor LendingPoint (https://www.kbra.com/documents/report/61196/lendingpoint-202…), which issued a new securitization report on Jan 14, 2022 for close comparison, I do NOT see a huge uptick in cumulative net loss rates like what I am seeing for UPST’s 2021 pass-through trust loans. So I don’t think I can reason this as “normal return” to usual macroeconomic environment with COVID stimulus fading away, if UPST’s peers are not experiencing the same.

What’s going on here? Is it a problem with UPST’s AI underwriting model? Was the model tweaked or learning in a way that is causing them to lend too loosely (but allowing them to transact more loans in the short term)?
In the longer run, if delinquency rates are spiking too high for UPST and eroding its underwriting competitive advantage versus peers, how can they maintain their bank partnerships in the future?
I didn’t look too closely at all the UPST individual reports but if someone can dig around the details and come up with good rationale for this finding (that makes this concern go away) I would very much like to hear it.

For the time being, what I am seeing is just another red flag to pile on to my list of reasons why I will not hold UPST through Q4 report and rather wait for management’s numbers/guidance/explanations (hopefully analysts will ask about delinquency rates on the conference call).

58 Likes

Hi Jon,

This was raised by Wedbush analyst on December 22, 2021, and I have heard some on the buyside have inquired about it. I am sure it will be something that is discussed on the conference call:

Wedbush analyst David Chiaverini initiated coverage of Upstart with a Neutral rating and $160 price target. Though he is encouraged by the company’s high growth with strong potential, and notes that the stock has pulled back over the last few months, he notes an uptick in delinquencies on recent 2021 vintage securitizations that seem to be deteriorating at a faster pace than its 2018, 2019, and 2020 vintages. Though he argues that a premium valuation is warranted given the company’s high growth potential, a widening premium to peers is difficult to justify until it is seen if the recent spike in delinquencies is temporary, Chiaverini tells investors.

22 Likes

Thanks, akhenatong.
I stopped following UPST closely after I sold the position a second time in early December so I definitely missed that.
It does look like going from December’s report to January’s KBRA report, it hasn’t gotten better. So this new January 18 report provides new information since what the Wedbush analyst noted in December, that the trend is worse.

If we compare Dec 3’s new issue report (https://www.kbra.com/documents/report/60014/upstart-pass-thr…) to January 18’s report (https://www.kbra.com/documents/report/61852/upstart-pass-thr…:slight_smile:

December  Months  Delinquency%
2021-ST3  6       2.88
2021-ST4  4       2.93
2021-ST5  4       2.70
2021-ST6  3       2.08
2021-ST7  2       0.96
2021-ST8  1       0.00

January   Months  Delinquency%
2021-ST3  8       3.85
2021-ST4  6       3.71
2021-ST5  6       4.03
2021-ST6  5       4.29
2021-ST7  4       3.38
2021-ST8  3       2.16
2021-ST9  2       1.04

The two additional months of seasoning for each 2021 UPST pass through trust is quite dramatic in spiking the delinquency rates across the board.

I believe their loan transaction numbers have continued to grow massively (they might do 491K for Q4, and I am now projecting 578K loans for Q1 at the current January run rate of trustpilot reviews), but if all this hypergrowth is at the cost of higher default rates in the long run…then it’s hard to distinguish them as any different than what LendingClub was doing in its earlier years (fast growth, loose lending criteria, then defaults crushed their business).
If I think of it another way, had UPST in 2021 managed to keep delinquency rates steady like in 2020 and earlier years, then they might not have grown so fast in the last few quarters?
Could it be possible that recent and ongoing hypergrowth is ‘artificial’ in this sense?

It is unfortunate that delinquency rates are a lagging indicator - only time can tell how their loans are actually performing.
Hopefully near-future modifications to their AI underwriting models can correct this course and return them to delivering outperformance in defaults versus traditional lending peers.

6 Likes

It’s true that the uptick in delinquencies is worrying.

But wouldn’t some of that happen naturally by going further downstream on the FICO score?

I believe Upstart goes into the high 500s now - 580 was the last figure I saw.

I imagine at that point, the delinquencies would go up.

The real question is:

  1. Is their AI learning from that? I still have faith they are, and they will adjust. But of course that’s just faith.

  2. Are the higher rates compensating for that? Obviously if I’m holding a package of loans with a higher delinquency rate, I would be just fine about it if their rates are higher, right? What matters is the overall yield and money I make.

Upstart did discuss expected loan delinquency upticks due to macro conditions changing. Some of these other lenders such as SOFi play only in higher FICO score and others like LendingClub are lending primarily to past borrowers because they know they are “safer.”

John, what would be helpful is to understand the delinquency rate of loan packages with similar FICO scores and similar rates. But especially similar FICO scores.

Otherwise IMHO you are just comparing Apples and Oranges and scaring shareholders.

62 Likes

John, what would be helpful is to understand the delinquency rate of loan packages with similar FICO scores and similar rates. But especially similar FICO scores.

Otherwise IMHO you are just comparing Apples and Oranges and scaring shareholders.

Hi Mizzmonika,

Ok, I don’t want to come across as rude but I am a little irked by your statement.

Your expectation is for me to go another extra mile to dig into each individual KBRA report and analyze the FICO characteristics and rates and initial loss expectations vs current loss across multiple vintages?

For a company whose stock I chose not to hold a long term position in for several weeks already? The only reason I looked into delinquency this morning is because I was holding a short term trading position in UPST (which I got rid of after these findings).

And then say I am just trying to scare shareholders? Would you prefer I not share any of my thoughts in the first place?

Maybe the better question you could ask instead is why didn’t anybody else, who actually owns UPST, bring up the delinquency rates back in December. If you hold shares in a company isn’t it your own responsibility to understand and keep track of how the business is performing? It is much preferred to do your own due diligence instead of relying on others…what I really want is someone else (multiple people) to dig into the numbers and either confirm or refute my concerns. I asked for this in my initial post within this thread.

This board is supposed to be a crowdsourced discussion!

2) Are the higher rates compensating for that? Obviously if I’m holding a package of loans with a higher delinquency rate, I would be just fine about it if their rates are higher, right? What matters is the overall yield and money I make.

In regards to your statement here, if you looked at the KBRA report itself, just glanced at the graph even, you can see the cumulative net loss rate is also spiking for 2021 trusts…so I don’t think rates are compensating for defaults…did you read it before asking this. Thanks.

60 Likes

Appreciate your posts Jon Wayne, but you shouldn’t be irked by Mizzmonika’s comments. You came in with a pretty alarming post on a day when Upstart is getting destroyed while literally just admitting you didn’t do the extra work to make sure your comparison was valid. I don’t think this is helpful for people that are deep underwater on their position and prone to reacting emotionally right now. We should absolutely discuss concerns about the company, but you should really make sure your research is buttoned up and not get upset if someone pushes back. You seem to have a psychological bias against Upstart that stems from the previously enormous position you had based on other research you did going into Q3 earnings. Again, let’s keep discussing the company, but I don’t think half researched posts that create panic during a huge sell off are helping anyone.

65 Likes

I think this thread should take a time out.

I personally see both sides and wish all well but right now emotions are running extremely high and camaraderie and sticking together in a tough time is important.

76 Likes

I’m responding because I started a small position in Upstart (UPST) a few days ago and it’s a long call option. I may buy a few shares shortly as well but it is still going to be my smallest position until I see how their earnings look like.

With that said, I really like what they are trying to achieve or disrupt the lending space for sub-prime borrowers. Couple of things I didn’t like about their AI hype and had also kept me away from investing earlier are.

#1: I had seen a lot of folks with high FICO scores and good net worth receive exorbitant rates when applying for personal loans.

#2: In the last quarter Upstart mentioned fraud as a detriment.

For #1, there was some messaging that since they target borrowers with low FICO scores, their AI doesn’t care about borrowers who are more credit worthy since their margins would be lower.

However, I do have some experience developing ML models( being used by services that cater to millions of users) and think that ML excuse for #2 is not a very good one as fraud and anomaly detection are pretty much integral to those models. I have friends working on the edge areas of ML and AI and I have discussed some of those aspects with them too. So, I feel Upstart’s ML and AI will mature over time and be better in addressing both the above issues.

If that happens, it will raise the level of confidence in their platform and service. And hopefully someday they may sell that service as SaaS.

Just my 2 cents!

ronjonb

@ronjonbSaaS on twitter

15 Likes

Yes with a triple somersault to Big Broadway Dan in his post above.

Jon: your research and effort and altruism in sharing your work and process is beyond gracious and generous and valuable. Please, for the sake of all things Holy, do not withhold from us your analytical gifts and offerings in the future. They are extraordinarily helpful.

And yes, everyone’s nerves right now are completely fried; and anything perceived, rightly or wrongly, as adding unnecessarily to the surging flood of doubt and confusion and uncertainty and fear is like a nasty bulldozer of suffering mocking us for our human frailties, susceptible as we are to being scared and worried.

So yes, this is a “both sides are correct” issue in that we must continue to do our best offering sound analysis and reasoning, and we should all consider the context of severe pain and ask “will this help people make a useful decision” going forward, investing-wise.

Jon’s post about Upstart’s delinquency rates made Monkey’s poophole pucker with additional fear, but given the 73% drop he’s already suffered, it didn’t change the long-term outlook on his investment, so it was a useful post, even if it did appear to be somewhat “underresearched” relative to the usual high standard Jon has spoiled us with. But Monkey knows all that and ain’t about to toss rotten bananas in anyone’s direction because the intention for all the posts was inherently positive. Hence Dan’s plea for peace and collaboration is supremely wise given his limitations as a bald ape.

Our businesses are superior. Their numbers are superior. Their future opportunities are superior. Stay focused on just those facts and grab a hammock. Time will do the rest.

Hugs,

Monkey (long UPST)
@cxddesign

109 Likes

Thanks for the heads up on delinquencies, Jonwayne-- I know you were heavily invested in UPST and kept the board well informed of any developments for a long time-- While you are in no way obligated to, I appreciate that you have kept us so informed about any position changes you have made-- You’ve been very transparent in that regard.

It sounds like you held a small momentum position or “trading block” of UPST and I’m thankful for that as your due diligence led to discovery of an analytic worth observing. Before last earnings there was a thread created where someone essentially said-- “Okay, we all love UPST, but let’s exercise a thought experiment and brainstorm potential risks.”

One of those risks brought up was if UPST suddenly began showing increased default / delinquency rates relative to peers. Obviously that could indicate a fundamental flaw to the technology and thesis.

Personally I think it’s the largest risk with UPST-- especially as it pursues automotive and mortgage lending in the future. Being that I consider this the greatest list, I’m embarrassed to say I have made ZERO effort to track it-- I’m glad someone did. The other risks related to macroeconomics and lack of ARR are much less concerning to me.

What I essentially saw in your post about UPST was “Hey I just discovered a potential red flag, this isn’t a primary holding of mine, but I feel it is actionable and wanted to share. Also, I recognize this may be explainable somehow, but that will take more digging which I don’t have the time or incentive to perform (as you no longer hold the stock), but for the good of the order, I wanted to share.”

We should all do our own research and understand the risks for our holdings-- a major breach in a cybersecurity client, supply chains woes for manufacturers, increased delinquencies for a lending software provider.

Ultimately, there is a lot to benefit from in this thread-- we all benefit from information sharing, and it’s important to have our ideas and sources challenged. I’m sure jonwayne, or Saul, or anyone else doesn’t want everyone to blindly invest how they say-- we have to challenge each other and come to our own convictions. For this reason-- IMO-- it’s important for Saul, or anyone else, to share, and to be challenged. We will all benefit in the long run from a well vetted crowdsourced conclusion.

There isn’t much to grin about in the portfolio these days, but once during a medical mission trip to Central America I discovered there’s a lot to be thankful for anyway-- (and not just the banans CMFMonkey :wink:

I transitioned into hypergrowth from REITS and Energy at exactly the perfect time to wipe out my annual gains and finish last year down 20%-- I’m now down 29% YTD for 2022. Yikes. It’s ugly, a little scary, and… all part of the ride. I’m still invested in companies I believe in, with high growth, strong financials, and long runways ahead-- invested in the largest and most powerful economy in the world. What should I change from that equation?

MillennialFalcon

PATH
PLTR
S
BILL
UPST (12.8)
VERI
DOCN
FUBO

43 Likes

Hi Jon,

When I read your post I immediately thought, as other people did too apparently, maybe default rates are up because they are moving into lower socioeconomic groups, and the interest rates are higher and will compensate for it. However, I want to thank you very much for the post nevertheless because it made me think “I don’t know whether this is really bad news or is meaningless, but what it does do is point out to me how really hard it is to guess the future for Upstart, for guess is what we have to do, compared to our other companies like Zscaler, Cloudflare, Datadog, Monday, ZoomInfo, etc, where we have much better ‘knowledge’ about what the future holds”.

I decided to greatly further reduce my already smallest position in Upstart. It may do great, it may do poorly, but it’s just a guess, not because of anything wrong with the company, but because it’s not a SaaS company like the ones I mentioned above.

Thanks again,

Saul

Links to the Knowledgebase for this board is in the Announcements panel that is on the right side of every page on this board. (It’s in three parts)

50 Likes

Hi Jon,

Each trench of securitized loan consists of different composition of grades of loans.

With reference to page 20 and page 21 of the new issue report, there are charts showing the composition of loans (prime Vs sub-prime) for 36 months, 60 months and 84 months.

For the chart referring to loan with tenure of 36 months , the chart shows that sub-prime portion increase from 50% in Q3 2017 to 85.55% in Q4 2022. In other words, the prime portion reduce from 50% to 14.45%.

There is similar trend for loans with 60 month tenure.

Each trench of loan is consisted of different composition of Prime and sub-prime loans. The higher the portion of sub-prime loan, the higher the expectation of delinquency rate.

Each trench of loan is consisted of different combination of 36 months, 60 months and 84 months of loans. The more the portion of loans with longer tenure, the higher the expectation of delinquency rate.

41 Likes

Upstart’s delinquencies spike, assuming that there isn’t a good explanation for it, is very disturbing. The competitive advantage of better risk assessment and therefore better underwriting and therefore lower rates for borrowers AND lower delinquencies for lenders is Upstart’s reason for being. These are the incredibly strong value propositions that will lead to the disruption of lending markets and financial riches for Upstart and its investors. This is true because borrowers and lenders will do what’s in their best financial interest.

Three months ago and after their Q3 earnings results, I wrote a post about why I was sticking with my UPST shares:

https://discussion.fool.com/ok-bear-i39ll-take-the-bait-now-firs…

If the delinquencies continue to spike relative to other risk models and if Upstart’s AI risk model is worse or no better (it needs to be not just as good but better) than other risk models then we have lost the value proposition for the lenders and we have a broken investing thesis. Obviously, offering lower rates to borrowers without lowering delinquencies will lead to a destruction of Upstart’s business because lenders will do what’s in their best financial interest and not do what’s not in their best financial interest.

I can accept that UPST earnings will be lumpy, and I can accept that the stock will be volatile. As long as the value propositions for borrowers and lenders remain intact, I can be patient. Upstart has been able to claim that its AI model performs better than traditional underwriting methods. This has been the case for several years including during the stress of the pandemic. Now that claim is being called into question which may foretell problems in Upstart’s ability to get new banks. Even the perception by the lenders that the AI is no better will hurt adoption. I’ve decided to sell most of my shares with uncertainty about Upstart’s AI model effectiveness now in question.

GauchoRico

39 Likes

This is a fascinating report and I appreciate you sharing it.

Delinquency rates are definitely on the rise and it is of concern. After reviewing the information here’s what I pulled out as a counter argument and would love to hear your thoughts.

On page 20, it shows that the have been underwriting increasingly “risky” loans, defined as C, D, E, and F loans with an inflection point starting in Q2 2020 and steadily rising. In fact, from 70% Q2 ‘20 to 86% Q3 ‘21 of 36 mo loans and from approx 55% to 76% of 60 mo. loans.

Increasing your risk factor, you would assume a higher delinquency follows. The other reading that I find interesting is comparing the KRBA base case projection vs current (pg 12). All 2020 loans appear to be performing 4-8% better than the projected base case as of August ‘21). None of the 2021 loans have been reviewed.

Also, if you compare projected net loss (pg. 31) vs. current net loss for each, they expect 5% around month 13, although trending higher (pg. 13), it would appear they are still on track to meet or exceed that threshold.

Final observation, on pg. 11 it shows 2022 ST1 vs 2021 ST10. Of note, it shows less D and E grade loans, in turn showing a lower interest rate and a lower expected loss. This may be nothing, but part of my thesis is that they are able to reach more prime credit borrowers at a good rate, expanding their target market. Alternatively, it could be the opposite and the AI tightening up on loans to the D and E borrowers due to the higher delinquency rate. Hopefully, we’ll get more color from the company.

Side note: Man has this been a difficult two months.

31 Likes

I decided to dig into the numbers

The good 2020-ST1

From page 7 of 2020 ST-1 report

https://documents.kbra.com/report/28335/abs-upstart-pass-thr…

Three largest categories out of a total of $64.0M

29% Grade B
25% Grade E
18% Grade C
18% Grade D
8% Grade A, AA
0.21% Grade F

The W.A. Interest rate for 30 months is 15.70%, and for 60 months, 16.28%

It looks like KBRA call Grade A and B prime and C, D, E subprime.

the bad 2021-ST6?

For 2021-ST6, where the spike in delinquency rate (4.2%) concern came from

https://documents.kbra.com/report/51088/upstart-pass-through…

Three largest category

42% Grade E
20% Grade B
15% Grade D
14% Grade C
7% Grade A, AA
3% Grade F (vs 0.21% Grade F in 2020-ST1)

Collateral interest rate 19.63%. About 3% higher than 2020-ST1

It also says in the report

“UPSPT 2021-ST6 has a greater percentage of E Grade loans than UPSPT 2021-ST5, which results in a higher weighted average APR for this transaction.”

What I saw

It seems that Upstart packaged this pass through security differently. There is a larger composition towards the E and F category. As an investor I certainly would hope that the delinquency rate stay stable, but this may be just how they put this package together.

As an investor in the market

Having to dig through numbers like this make Upstart a definitely more complex story than SaaS companies. I don’t know how one should handle this but I am ok with holding my (still) 20% position in UPST despite the 75% drop from it’s high. I think the thesis for me hasn’t changed like GauchoRico said as I was digging through the numbers.

But I understand everyone’s decision and conviction is different. I am very glad that jonwayne brought this to the board. It tests the thesis which as an investor one shouldn’t shy away from despite all the emotions involved in the market.

88 Likes

Does Wedbush read this board? Or is Jonwayne in fact Wedbush?

“Wedbush Lowers Price Target for Upstart Holdings to $110 From $160 on Elevated Delinquency Rates, Maintains Neutral Rating”

6 Likes

I am a newbie, and still learning the ropes of stock investing. I dont think I should even post here, given my lack of investing credentials. So I am afraid I really can not add any substantive information to the discussion in question. However, Just wanted to add one piece of information. Apologies if this is not of much help and please delete if this is deemed OT.

I checked the analyst ratings on Upstart and this is what I see. Please note each of these ratings are AFTER the earnings report ( Upstart reported on Nov 8th…and these ratings are all within the last 2 months, with 4 rating in the last month)

These were in the last 1 month

  1. Piper Sandler: Arvind Ramnani - BUY - $ 223
  2. Wedbush: David Chiaverini - Hold - $ 165
  3. Jeffries: John Hecht - Hold - $ 175
  4. Morgan Stanley James Faucette: - Hold - $ 200

These were in the 2 months ago

  1. Citigroup Peter Christiansen buy $350
  2. JMP Securities Andrew Boone Buy $315
  3. Barclays Ramsey el assal buy $285
  4. Goldman Sachs Mike Ng buy $290
  5. And then of course, Nat schindler from Bank of America who I believe was one of the first to double downgrade upst when it went to moon, and first gave it target price of $300…and then revised his sell rating again, but even his revised target price was…$250

At this point, being a novice, I am not really sure how useful this information is, or even whether such analyst ratings play a role in stock price movement. However, I am sure the downtrend started the moment the analyst downgraded it. I may be way wrong here…but if the macro environment was lets say bullish (which it clearly isnt), I can not but help think that UPST price would be much higher if even the “sell” side analyst rating has a price target of $255.

I just wanted to share this and seek the input of the thought leaders in our board, as I am still very much in the learning phase. If this is off-topic, I can certainly understand if this gets deleted.

Thanks a lot Saul and everyone,
John

8 Likes

Analysts ratings and price targets are completely and totally meaningless. They upgrade after big advances and downgrade after big declines. The best thing you can do is ignore them entirely and learn to read a price chart.

4 Likes

Having to dig through numbers like this make Upstart a definitely more complex story than SaaS companies. I don’t know how one should handle this but I am ok with holding my (still) 20% position in UPST despite the 75% drop from it’s high. I think the thesis for me hasn’t changed like GauchoRico said as I was digging through the numbers.

Thanks chang88 for looking into the composition of the latest securitization. From my perspective as an investor, the thesis may have changed. I just don’t know if it has and can’t be sure. I decided that I didn’t want to wait for things to become more clear and decided to take my ~15% position down to 3.4% with 3/4 of my remaining position subject to getting called away at $100 next week.

Yes, as Saul wrote, there are other high quality, hyper growth companies with better visibility into the future. These other companies are down as well and I have a lot more confidence that they will continue to grow revenue for a long time (and future stock prices should eventually reflect that growth). Yes, UPST is down a lot more but I think it’s fair to say that UPST never should have reached $400/share. So investors might want to be careful not to anchor on that all-time high for UPST. If I miss on future UPST possibilities then so be it.

GauchoRico

28 Likes

Can’t help but feel like this skepticism is heavily biased by the horrendous price action. The company put up strong enough numbers in Q3 for many people to hold their position all the way down from $300+ to now 95 bucks. I think a lot of people, myself included to a degree, have been looking for any possible reason to bail b/c it’s been such a miserable experience these last few weeks. I could easily be wrong, but this thread feels like an easy excuse to sell and I think there’s more psychological / emotional aspects at play here than necessarily investment logic. But, then again, that can just as easily apply to someone like myself whose stubbornly holding on to his shares…

22 Likes