Upstart Moratorium Over - My Estimates

I would like to go through a brief exercise in numbers to derive the growth expectations embedded in the Wedbush price target ($75) published this morning.

Analyst David Chiaverini is pricing UPST shares at 20x next year’s (2023) EPS (implying $3.75 EPS for 2023) discounted 10% per annum back from 2026, thereby implying $5 EPS in 2026 ($5 / 1.1^3 = $3.75).

DISSECTION OF WEDBUSH ESTIMATE

I’ll be discussing my own assumptions and basis in a moment, but first let’s go through the Wedbush model to answer the question, Is a $75 price target reasonable, or just playing into momentum?

  1. Let’s convert net income (EPS) to EBITDA. $5.00 EPS will equate to about $5.10 in EBITDA/share, this is proportionally consistent with 2021, and given no long term debt, fixed asset amortization, or significant tax burden given the stated goals for reinvestment in the business, this would seem reasonable to extrapolate forward.

  2. We will estimate diluted shares outstanding in 2026. Current (EOY 2021) diluted shares outstanding are 94.7M. Let’s back out the current authorized $400M buyback at the current share price–presuming they exercise the buyback given share price weakness. Diluted shares fall to 91.3M (EOY 2021). To project this forward to EOY 2026, let’s presume two things:

…A) Quarterly dilution of 0.8% (80 basis points), which would track their three full quarters they’ve reported as a public company

…B) The buyback is sustained for 2022, 2023, 2024, 2025, and 2026, in similar amounts—meaning a total additional buyback of $2B. In my experience, share buybacks become part of corporate culture. And given UPST’s profitability, I would expect that a sustained buyback is totally reasonable, if not conservative. I say “conservative” because in the face of depressed share prices, they may reasonably accelerate the buyback, curtail SBC, or both.

These inputs result in a 2026 ending diluted share count of 90.0M. (1.008^20 quarters * 91.3M – ($2B / $117)).

  1. Now we can deduce that Wedbush is modeling a 2026 EBITDA of $459M (90M * $5.1 EBITDA/share).

  2. I will also make one more deduction for their model; EBITDA margin remain constant at 2021 levels (27.3% 2021 EBITDA Margin).

So what does that mean for top line growth expectations for 2022, 2023, 2024, 2025, and 2026?

Calendar year 2021 adjusted EBITDA, reported, was $231.9M. This means that Wedbush is anticipating 98% EBITDA growth from 2021 to 2026, which is 14.6% top line CAGR for 2022-2026. ($231.9M * 1.146^5 = $459M).

That’s too simplistic, though. Given that (materially lower) EBITDA margins and (materially higher) revenue growth have already been guided for 2022.

Incorporating guidance, let’s make these two changes to more accurately predict their model:

  1. 2022 Top line growth will be 65% and straight line CAGR will apply from 2023-2026 only

  2. EBITDA margin will contract in 2022 to 17% (per guidance) given their investment in their auto roll-out, and not recover given the continued roll-out to new markets (SMBs, Mortgages, etc.).

This now leads to the following growth assumptions, still constrained to arrive at an EBITDA of $459M in 2026


		2022		2023-6 (4-year CAGR)
Rev. Gr %	65%		17.8%
Revenue	$       $1.4B		$2.7B
EBITDA %	17%		17%
EBITDA $	$238M 		$459M

2026 EBITDA = $459M
2026 Earnings = $450M
2026 Market Cap = $9B
2026 EPS = $5
2025 Share Price = $100/sh ( based on “next year’s” (2026) earnings)
2023 EPS = $3.75 ($5 discount 10% per annum)
2022 Share Price = ~$75/sh
Implied PEG Ratio = 1.12 (20 / 17.8)

Okay, so that’s roughly the model that Wedbush is using. They’re taking 2022 Revenue and EBITDA guidance as given, with no beats. Then, presuming stable margins given the new market penetration for the outyears, and applying a 17.8% top line CAGR for 2023 through 2026, leaving $459M of EBITDA in 2026.

IS THE WEDBUSH MODELING REASONABLE, AND WHY?

No, it is not. For two reasons, which happen to be the two primary inputs: revenue and margins

Revenue:
The 2022 revenue forecast is, of course, the absolute bear case for Upstart. Upstart has beaten revenue guidance by 4.6%, 22.9%, 6.3%, and 16.0% (12.5% average) in their four quarters as a public company (oh the joys of non-SaaS). Girouard and Datta were utterly ecstatic on the call, but what can we point to for supporting a beat on revenues? Auto, for one. Datta stipulated that the $1.5B in auto originations was due to (paraphrasing), an early first-half surge that gives them confidence to telegraph the $1.5B. I read this as, “we are seeing acceleration in the business which will be material in the early part of the year, allowing us to be very confident in our guidance.” So, again, auto is almost certain to overshoot.

Upstart began last year with 15 bank/credit union partners, and began this year with 42. They also trebled their dealership footprint YoY in Q4 and 41% sequentially. This is massive acceleration. Unfortunately we don’t have personal/auto breakouts, so I’m assuming continued growth in these fronts is a big piece of the 65% forecasted growth.

At any rate, history indicates that 65% will be low end of 2022 growth. If I apply a 10% beat (below average), that gives us just over 80% YoY growth in 2022. I’ll forecast 80%, then. Regarding 2023-2026, it’s hard to be bearish given the commentary around auto and the references to scaling “Our auto loan funnel looks a lot like the personal loan funnel in 2019” (mentioned several times on the call). Throw your dart where you will, but I see a business still in its infancy, bringing in bank and dealer partners (capturing TAM) at a rapid clip that is outpacing reported revenue, and partnering with entire auto companies–Subaru and Volkswagan both announced Upstart as their preferred digital retail partner, both in just the last 12 days! Was that in guidance?

I feel like an ~80% growth durability for 2023-2026 is completely reasonable. No, this isn’t SaaS, but this is a fee-based business tethered to a cyclical lending market. Yes, loan origination is hurt in a recession, but the trend for loan origination for all of history is up, up, up. The pie is enormous. Upstart is anticipating 0.2% market penetration in 2022 (one-fifth of one percent of the auto loan market, $1.5B out of $727B), and is at 4.3% penetration in personal loans. How exciting is that? And they are actively moving into new, broader markets for SMBs and Mortgages. And there’s currently no discussion around moving internationally—but the opportunity is there. Their AI aims to do one thing—accurately price risk. Are there other industries outside of direct lending that can benefit from a better risk pricing mousetrap? How about insurance?

Margins:
The 10% contraction from 27% EBITDA margin in 2021 to 17% margin to 2022 is driven half by the auto-rollout, which is the result of a lower margin profile for funding loans from the balance sheet, auto model maturation, customer acquisition costs, and other sub-scaled unit costs. These auto-related elements are expected to hit contribution margins to the tune of 5%. And the remaining 5% margin contraction is due to workforce investments, to which they underspent in 2021 and want to increase by 150% this year. When asked about these items, Datta said: “there’s no fundamental reason why our business over time won’t return to our existing profile and, in fact, beat it.” And also, “I suspect by 2023, [auto] will be accreting to the bottom line, not reducing our margins.”

I can garner that 17% is bear case for 2022, so let’s go with 18%. Then returning to 27% through to 2026.

UPSTART PRICING MODEL MORE REASONABLE THAN WEDBUSH


	        2022	2023	2024	2025	2026
Rev. Gr %	80%	64%	51%	41%	33%
Revenue ($B)	$1.5 	$2.5 	$3.8 	$5.3 	$7.1 
EBITDA %	18%	20%	22%	24%	27%
EBITDA ($M)	$275 	$501 	$833 	$1,281 	$1,914 

2026 EBITDA = $1,914M
2026 Earnings = $1,876M
2026 Market Cap = $69,337M (PEG of 1.12)
2025 Share Price = $770/sh
2022 Share Price = ~$578/sh

WHAT YOU’RE SEEING WITH UPSTART IS A BINOMIAL OUTCOME

On one hand, you have a cyclical company with a high supplier concentration for personal loans, venturing into unproven markets, facing the possibility that the “big boys” will just in-house a solution, demonstrating that they are vulnerable to global pandemics that shut down national and global economies, and have an underlying technology that is still somewhat viewed as fringe. This company has a value of $807 Million (book value).

On the other hand, you have a company whose fee-based business model is largely insulated from interest rates, continues to surmount economic headwinds surrounding fiscal stimulus, auto prices, student loan forbearance, etc., has a mousetrap that is relatively superior in all macro conditions, is venturing into vast, inefficient, and ripe-for-disruption loan markets, has a massive advantage in its data-tested learning models, trades at a reasonable multiple to its earnings given its growth, is massively profitable (just reported 27.3% EBITDA margins), has unimpeachable, forward-thinking leadership, is growing at head-spinning rates, and faces mouth-watering optionality for new geographic and lateral markets. This company has a value of $1 Trillion (because, why not?).

Sentiment, and little else, is what assigns probabilities to the binary outcomes above. Right now, sentiment is heavily weighted toward the former. Depending on your own personal sentiment, Upstart may fall somewhere far different on your own spectrum. Allocate accordingly.

Eric Przybylski, CPA

167 Likes

Hi Eric,

Thank you for posting your detailed estimates of what might go into the Wedbrush pricing model for UPST.

I’ll point out one flaw I noticed in your analysis: your estimate of diluted share count is substantially off. You wrote: We will estimate diluted shares outstanding in 2026. Current (EOY 2021) diluted shares outstanding are 94.7M.

Going to the recent PR for 4th quarter earnings however, I see that number is actually the weighted average of diluted shares outstanding for fiscal year 2021, and for a more number more relevant to end of year we should use the figure they give for weighted average of diluted shares outstanding for the fourth quarter 2021, which is 98.8 million, a significant increase over the number you use.

Also, as opposed to using a calculated forward estimate for next quarter, we can use their forward guidance given for first quarter 2022: Diluted Weighted-Average Share Count of approximately 95.9 million shares, which appears to reflect the reduction in share count one would expect from them spending the $400 million allocated for share buybacks.

I also question your assumption that share buybacks will continue apace for the next 5 years. In the earnings call they talked about the buyback, how it was unusual for a company this early in their growth cycle, and they basically said they just saw it as such an investment opportunity given the low prices, they thought it a great way to invest the money. They didn’t indicate any plan to make that a recurring event, though if the share price continues to languish, I’d expect them to allocate more money for it.

I also saw somewhere in the earnings release or Q&A that they do expect to continue to rely heavily on stock based compensation for the foreseeable future. Given that they took on quite a bit of debt last year (increase from $63 million to $695 million year over year), it would seem to me that adopting a policy of regular buybacks at this stage of the game is premature and in conflict with their stated policy of continuing stock based compensation.

Long story short, we should be starting from a diluted share count of ~96 million in first quarter 2022 (post authorized buybacks) and, IMO, assume increasing share count from there, as opposed to decreasing, though if the stock price stays low, perhaps it will be roughly stable.

25 Likes

Thanks for the correction, Ben. Certainly the year-end diluted share count or, better yet, the forward guided count–as you point out–would have been preferable to my use of the full year weighted-average fully diluted count.

Although, I would hesitate to call this ~4% variance “substantially off” and a “significant” difference as you have done–especially when you consider the takeaway from my analysis is that UPST is potentially underpriced by orders of magnitude. I certainly hope the insinuation does not dissuade anyone from considering my analysis.

But, regardless, I was wrong and I appreciate your scrupulous review of the essay portion of my Level 3 CFA exam.

Eric

41 Likes

Although, I would hesitate to call this ~4% variance “substantially off” and a “significant” difference as you have done

Yeah, okay, 5% off on 1st quarter 2026 diluted share count isn’t that big. However that would grow if I’m right about them continuing to issue more shares than they buy back.

Your reported historical dilution of .8% per quarter also seems wrong:

First quarter 2021: 91.4 million diluted shares average (m.d.s.).
Second quarter '21: 94.8 m.d.s., growth from prior quarter: 3.7%
Third quarter '21: 96.1 m.d.s., growth from prior quarter: 1.4%
Fourth quarter '21: 98.8 m.d.s., growth from prior quarter: 2.8%

Simple average of the growth in diluted shares last year is 2.3%.

First quarter 2022: estimated 95.9 m.d.s. as the company buys back $400 million worth of shares, a 2.9% drop count from prior quarter, which is nice, but will only offset a little more than one more quarter’s typical growth and if diluted share count continues to grow like last year we could be looking at a net increase of diluted shares of about 4% over the course of 2022 despite the $400 million buyback.

If this were to play out like that through 2026, we’d see diluted share count grow from ~99 m.d.s. Q42021 to ~120 m.d.s end of 2026, 33% more than your estimate of 90 m.d.s.

I’m just using rough numbers here as opposed to going for a false precision.

7 Likes

Hi Ben,

I appreciate you providing some additional color on share counts. While I don’t agree that UPST management will continue to authorize (and execute) share repurchases, and I do agree that share dilution will continue, these changes will not dramatically impact the analysis.

I’m not sure where you were going with the false precision statement, after all, these are all models based on information that is extraordinarily hard to pin down.

I appreciate the effort here by Eric to redirect our thoughts on UPST financial numbers based on the company performance, opportunity and TAM.

Perhaps we can call it a model with a +/-34% error bar?

That would put the central expectation for 2026 UPST shares at $50x.00 <2026 Share price<$103x.00. (2026 modeled price target at $770.*

Wiggle it up or down a few ticks based on whether you think they will continue to suffer dilution or allocate to share repurchases.

That’s an acceptable range of outcomes for this shareholder. Eric, thank you for the analysis.

Long UPST.

*Let’s also assume inflation continues it’s runaway path of 8% per year between now and then, too. $770 in 2026 currency at 8% discount rate is equal to $524 in today’s dollars. Eric, what was your modeled discount rate for inflation? (I may have missed it in the model assumptions).

25 Likes

Hi Eric, Thank you for the detailed analysis. It is important for us perform these analyses, as you have done, so we may try to look at these investments with an unbiased view.

And also thanks to Ben for pointing out 2 items which I was pondering as I read the analysis:

  • Stock based compensation on a quarterly basis is larger than the share buybacks and will continue to be for the near (and maybe long term) future
  • The buybacks appear to be somewhat opportunistic - based on management’s commentary – and counting on them being consistent through 2026 is not what I read from the comments

This leads to different share count for the analysis of Wedbush data and notably different results on which the end analysis is based. And yes, in most investment analysis a 5% initial variance IS notable, especially one which compounds from there (as Ben pointed out).

Back to my first paragraph stating “It is important for us perform these analyses …”. It is equally important to refine the analysis when data changes or errors are found in our underlying foundation. So when Ben (or another Saul reader) identifies a flaw in the basis of the analysis, it would behoove us to listen and revise and not to dismiss and argue.

I too am long UPST and am looking for significant growth in the company and the stock price. But I don’t agree that it is “potentially underpriced by orders of magnitude”. Underpriced? Yes. By orders of magnitude? Maybe a multiple, but “orders of magnitude” would imply that UPST should be priced at over $1000 today.

So again I thank Eric for his good work in on this and putting it out there for us to review and discuss. It has given us some datapoints to work with, think about, and build upon. And thanks to Ben for providing public feedback to stir the discussion.

Now, if we can only take feedback and work with it, and not reject it as dissenting criticism, then maybe our models will continue to be refined and we won’t get slapped with another 2 week moratorium on all things UPST.

Best,
Jff

11 Likes

Eric,

I largely agree that Upstart faces a binary outcome. You gave quite a range: On one hand, you have a cyclical company… This company has a value of $807 Million (book value). On the other hand, you have a company which…has a massive advantage…is massively profitable…is growing at head-spinning rates…This company has a value of $1 Trillion (because, why not?). I’ll assume the ~1B to 1T range is hyperbole, but the point holds that the true value of Upstart (like most companies) is really difficult to determine – and I grant that as you said, sentiment…assigns probabilities to the binary outcomes. You concluded: Right now, sentiment is heavily weighted toward the former. I would disagree to some extent. Upstart is still up something like 5x+ from its $20 (or was it $26) price at IPO, less than 1.5 years ago. And with currently a ~12b market cap, I don’t think Upstart’s valuation ignores the upside.

You gave a model where Upstart reaches almost 2b of EBITDA by 2026, and estimated a market cap of ~69b. I have no problem with that and would agree it is potentially conservative. Maybe they grow even faster and churn out 3b of EBITDA by then and are valued at a 100b market cap in 2026. Let’s call that roughly best reasonable case. I grant, it’s not impossible that Upstart could be even higher, but worst case would be bankruptcy and I’m not going to look at that possibility either.

So what’s the middle-ish reasonable case? What if they have another year like 2020 in there where some macro event causes their revenue to take a big hit – maybe for even longer than one quarter? Maybe revenue growth even goes negative that year. Maybe instead of:


	        2022	2023	2024	2025	2026
Revenue ($B)	$1.5 	$2.5 	$3.8 	$5.3 	$7.1

we see something like:

Revenue ($B)	$1.5 	$2.5 	$2.3 	$3.3 	$4.5

This isn’t a horrible scenario! Getting to 4.5b revenue is still a tall ask, but it gets them to perhaps closer to 1B in EBITDA instead of 2B. If in this case perhaps the 2026 market cap is only 30b or 40b, we’re no longer talking about a super impressive CAGR from 12b now, but we still probably beat the market and aren’t too disappointed.

The worst reasonable case is that despite Upstart being dominant in personal loans, the personal loan market doesn’t really grow that much and they’re already close to saturating what they can (or will be by the end of 2022), and also that auto just doesn’t work out as well as they are hoping. In this reasonable, but bad, scenario, maybe the trajectory looks like this:


Revenue ($B)	$1.5 	$2.0 	$2.4 	$2.2 	$2.6

Even this isn’t total disaster – they still end up with 3x as much annual revenue as they have today – pretty good when revenue isn’t even recurring! They just kind of hit a wall after 2b. Maybe there’s a macro event in there too. And if EBITDA gets a little strained, maybe they don’t end up with more than 500m in 2026. Because of the lack of growth, the multiple gets a little compressed, and Upstart’s market cap ends up not much higher than today’s ~12b. Maybe it’s even lower as the market gives up and moves on for growth elsewhere.

Conclusion: I believe the current ~12b market cap is a reasonable weighing of these possibilities. The $35b+ market cap they reached last year was, in my opinion, not reasonably weighing the downside. All the market could foresee in 2021 was “trees growing to the sky.” Upstart’s tree has the potential to grow really tall. But it also has the potential to run into struggles. I decided Upstart completely disrupting the entire credit market was not something I want to bet on – not so much because I believe they can’t do it, but simply because I have no way to evaluate whether or not they can. It’s just not something I know much about.

Summary: I think we agree the error bars for UPST are wide. We somewhat disagree about whether the current valuation is pessimistic or reasonable. You’re willing to bet on one of the better cases, and I don’t think this is a horrible bet. I simply like other bets (companies) better because I find their trajectories easier to predict. I’m pretty convinced (cloud) software “is eating the world” – I am not convinced at all about how credit markets will act/change/evolve.

Bear

PS - I’m sensitive to the fact that Saul will probably want us to wrap this up and not obsess over valuation. We can tweak models all day, and there’s no use arguing minutiae, which is why I conclude that like with every company, this is a judgment call, and we all have to decide what we expect and how much we’re willing to bet on it.

PPS - I think this exercise is good to do from time to time, but mostly it reminds me that the error bars are almost always so wide that if we find the best companies, the numbers are a matter of time.

88 Likes

PS - I’m sensitive to the fact that Saul will probably want us to wrap this up and not obsess over valuation. We can tweak models all day, and there’s no use arguing minutiae, which is why I conclude that like with every company, this is a judgment call, and we all have to decide what we expect and how much we’re willing to bet on it.

I agree with Bear! This is a long thread obsessing over valuation and arguing minutiae, rather than dealing with how the company’s business is actually doing.

I’m not sure why Upstart seems to bring out arguments over minutiae more than any other company, but it does. Perhaps because it’s so complicated and opaque.

BUT LET’S BRING THIS DISCUSSION TO A CLOSE UNLESS YOU FEEL YOU HAVE AN ESSENTIAL LAST COMMENT.

Thanks for your cooperation.

Saul

14 Likes

I would like to correct the 4.3% figure for personal loan market penetration from the first post in the thread. Upstart’s latest slides have personal loans as a $96b market (originations). They originated $4.1b in personal loans in q4. That would annualize to $16.4b, equivalent to 17% market penetration.

This market is expanding, and I believe that Upstart enables that expansion. However, I think it is important for people to understand that Upstart defines their markets in terms of originations, and it is important that people use apples to apples and don’t compare Upstart revenue to the size of the originations market in order to get an idea of market penetration. This was a mistake I made when first looking at the company, and seeing this mistake again, I think it is important to point it out.

Sorry for the additional post after Saul asked for the thread to be closed. I think this point is worthy of Saul’s invitation to add something else, if important.

20 Likes