2023-08-28 10:22:00 ET
Summary
- Upstart Holdings suffered a steep decline after their Q2 '23 earnings, which has only gained momentum since.
- The main reason: the Q3 outlook was marginally shy of consensus, and given the vertical ascent of the share this year that implicitly assumed raised guidance, profit-taking was rife.
- The article explains why the Q3 outlook is perfectly creditable. In fact, the company's near-term conservatism is rational in the face of increasing consumer stress.
- As for the medium term, the article details why Upstart's competitive strengths have only got stronger.
- The perfect macro-economic storm the company recently faced has subsided. Upstart stock is a "Strong Buy" before the sails catch wind. Target $84 by end 2024.
Upstart Holdings (UPST) can be described as nothing if not volatile. Whenever a stock finds itself in a violent tug of war between the bulls and bears, it's because there's a fundamental misunderstanding of the company by the two camps. Often, it's because the devil's in the details.
Upstart share price (seekingalpha.com)
This article delves deep into the weeds of the company’s latest results, that led to a savage correction in share price after a vertical ascent this year. If the reader endures the tortuous detail, he/she will emerge with a far better understanding of why Upstart is a “Strong Buy.” It explains how management’s decisions are both rational and desirable from a shareholder’s perspective: in addressing the scars from the perfect macro-economic storm from which Upstart has emerged; setting the stage for an ascent in future revenue and earnings growth, by solidifying the company’s competitive strengths.
Upstart's Q2 2023 Report
An analysis of the results will segue nicely into the company's operational strengths. For Q2 2023 , Upstart delivered a beat on both top and bottom-line estimates, with net revenue and adjusted EBITDA coming in at $136.8M (beat by +0.4%) and +$11M (beat by +$14M), respectively. While Upstart's revenues declined by 40% YOY in Q2 '23, the sequential growth was +35% from Q1, probably the company's low point in the final phase of a perfect macro-economic storm.
The surprise in EBITDA sprang from a jump in operational efficiency, particularly in marketing costs that fell a significant 76% YOY (from $105m to $24m). Upstart stated this remarkable improvement was achieved because a whole 88% of the loan grants in the quarter were fully automated and required no human intervention. This speaks volumes for the operational leverage now embedded in the company (this ratio was 64% in Q2 2019). Related to this point, 38% of loan originations were derived from repeat borrowers, indicating the hassle-free instantaneous decision provided by Upstart is highly appealing to borrowers.
The result: the contribution margin was a record 67%, up from 47% in the same quarter of the prior year. (Contribution margin is a non-GAAP measure defined as revenue less borrower acquisition & verification costs, akin to a 'gross profit margin' in a manufacturing concern, thus a great measure of operational efficiency.) Doubtless this ratio will fall as originations grow strongly and marketing dollars target less obvious borrowers, but the improvement is testament to Upstart's cost-cutting success during the drought of 2022.
Delving deeper into the revenue line, it's important to be aware of one accounting convention employed, as it sheds light on both the market’s misunderstanding of next quarter’s forecast and improvements in the accuracy of the AI model.
Q2 2022 | Q1 2023 | Q2 2023 | |||||
Revenue from fees, net | 258 345 | 117,141 | 143,689 | ||||
Interest income | 28 974 | 45,315 | 33 916 | ||||
Interest expense | (2,313) | (7,132) | (4,282) | ||||
Net Interest income | 26,661 | 38,183 | 29,634 | ||||
Fair value and other adjustments | (56,844) | (52,397) | (37,557) | ||||
|
|
|
| ||||
Total revenue | 228,162 | 102,927 | 135,766 |
Each quarter, revenue begins with fees, or commissions generated from the run-off of previous originations (the duration of a typical personal loan is two years) and new loans. This is described as ‘Revenue from fees, net’. Following is the net interest income generated from loans Upstart retained on its balance sheet , for R&D (for testing and to develop new AI models such as auto-loans) and loans retained when illiquid credit markets precluded passing them on.
Then comes an item termed "Fair value and other adjustments" which reflects impairments to fair value of their loans held on balance sheet. As seen from the three quarters above, this charge-off has risen - predictably - as rates rose, but it also contains impairments to R&D loans where actual loss rates were higher-than-expected. (More on this weakness, now perfectly resolved in the UMI section below.)
The above was necessary to explain the market's disappointment in the Q3 outlook. Prior to the last earnings release, the consensus forecast for Q3 revenue was $155m, which included a positive net fair value adjustment of $10m; analysts assumed loans held would yield a net interest income of that quantum, net of some minor impairment.
However, Upstart guided for 3Q revenue of a net $140m after a negative $10m fair value adjustment. Hence in actuality, Upstart's revenue guide (before the fair value adjustment) was $150m versus $145m consensus.
The explanation above was necessary to rebut article headlines such as this , following earnings: "Upstart Holdings' stock plunged 34% after it offered a disappointing forecast." Nonetheless, a bear could claim Upstart's steep slide was warranted, as embedded in the 401% rise this year-to-date (up to 2Q earnings, see chart above), was an implicit expectation the company would raise guidance substantially. And yes, there is some merit in the argument.
However, there's a very valid reason why that wasn't the case. In fact, once the reader understands the rationale for Upstart's near-term caution, she/he will agree that this very conservatism that disappointed the market sets the foundation for a far more secure and promising future.
This conservatism will only be fully understood with reference to Upstart’s UMI Index , an addition to Upstart’s methodology in October 2022.
The UMI Index
As any investor is aware, COVID created havoc with numerous economic variables. For Upstart, the impact was felt in the largesse of government assistance to consumers during the pandemic (resulting in default rates markedly lower than that predicted by their AI-model), followed by higher-than-expected default rates when government assistance was withdrawn.
In other words, 2022 was marked by an underperformance of Upstart's AI model, where on a given vintage the expected yields (as time progressed and cashflows received) were lower than targets set at the outset. NB. Both phases are visible from the graph legends below.
Upstart Loan Performance to Aug 2023 (ir.upstart.com)
Doubts in the model’s accuracy were rife, where critics claimed Upstart's model only worked in a declining rate environment, one that had prevailed over the last decade and from which their AI model was trained; the model was hence useless in the tightening cycle recently invoked by the Federal Reserve.
Naturally this gave pause to credit funds, the purchasers of Upstart's loan portfolios, especially since the Fed was hiking aggressively resulting in the sharpest rate-hike cycle in almost two decades (graph below). Hence it should be no surprise that the resultant drought in credit markets in 2022 became Upstart's annus horribilis .
But management proved fit for the existential challenge posed. Upstart accommodated this macro-economic uncertainty by the addition of the UMI (Upstart Macro Index) , a real-time variable that monitored the macroeconomic impact on the financial health of American consumers.
With this new dial that better reflects the macro-economy, Upstart incorporated a macro-overlay (includes numerous variables such as unemployment, inflation, fiscal largesse etc.) to its AI model that is distinct from the consumer’s individual financial stress (reflected in the model’s data).
The addition of the UMI has been an unequivocal success, resulting in a marked improvement in the model’s accuracy. The vintages (Q4 '22 onwards) since the adoption of UMI have led to outcomes almost precisely on target. In other words, the cashflows net of loss rates derived from the loan portfolios originated since Q4 '22 have delivered exactly what was expected of them (as predicted by Upstart's AI model) up to August '23. This explains why the recent quarter - despite disappointing market expectations - bears testament to the company's mounting competitive strengths. Credit providers such as Castlelake (Reaches Purchase Agreement for up to $4 Billion) began to embrace Upstart loans in earnest, now that the model's integrity had been vindicated, following a period of grave doubt.
Smaller "fair value adjustments" should impact future revenue
Another positive is investors can look forward to far smaller "fair value adjustments" to future revenue. Note that 2Q '23 charge-off was a material $37m, offsetting the $29.7m net interest earned. Now that the model's predictive power has vastly improved, there will be less need for unexpected charge-offs on the R&D loans Upstart used for AI-training. This was explicitly stated by management during the last earnings call , when quizzed about the surprise -$10m charge forecast for 3Q '23:
CFO Answer (my italics)
Yeah. Hi, James. Yeah, I would say a major factor in that number is the fact that by and large, our balance sheet now particularly after the ABS deal we completed, which -- where we cleared out most of our recent sort of core personal loans, the remaining balances are R&D and a lot of it is very seasoned. And some of those older vintages of, say, auto loans, auto loans from year-and-a-half years ago, their charge-offs are starting to grow, both because they were originated at a time where our models were still sort of in calibration and as well, they were originated in an environment that was -- and has rapidly deteriorated.
As a result, after this final charge-off of an estimated $40m in Q3 '23, an investor can remain reasonably confident Upstart has purged its balance sheet of the scars inflicted from the unprecedented rate-tightening cycle of recent. Granted, there could still be mark-to-market adjustments of loans held if rates continue to rise, but those adjustments will be relatively minor, thus permitting a far cleaner revenue trajectory.
Near-term caution explained
Upstart could be criticized for a tepid outlook for the current September quarter where the sequential growth of 1% was found wanting by the market. However, there’s a perfectly valid reason for this caution. The UMI index rose from 1.49 in March to 1.68 today.
UMI Index Aug 2023 (upstart.com)
As stated on the conference call
Our best measure of borrower delinquency trends, the Upstart Macro Index has tread water over the past few months, and in fact, even seen a more recent seasonal increase versus earlier '22 and '23 levels, as we came off the favourable tax refund seasonality that ran through April.
This conservatism is entirely warranted. It is far more important for the AI-model to deliver accurate results with expected loss rates rather than to chase marginal business when consumer strains are evident. The credibility gained with credit funders, thus securing a healthy and vibrant credit flow will reap enormous rewards when the consumer’s financial health renders him eligible to shoulder additional loans.
Valuation
Latest remarks from the Fed at Jackson Hole strongly suggest that inflation has peaked but a vigilant Fed will remain on watch. The CME FedWatch , which reflects market expectations about future changes to U.S. interest rates indicate the rate-tightening cycle is nearing its end . But even if there is one, or even two more 25 bps hikes ahead, the acute disruption in credit markets caused by the recent rate-hike cycle (see Fred graph above) is behind us. By all estimates, 2024 will be more benign to Upstart than this year.
The environment will permit Upstart to capitalize on the remarkable improvements made in its business model by attracting borrowers at lower APR rates, on a hassle-free and instantaneous loan approval process; offering precisely calibrated portfolios to credit funds searching for a risk-adjusted return. Remember high rates do not extinguish a $162bn personal loan market, they just make it more discerning in terms of affordability for the borrower and attractiveness of yield for the lender.
Earnings model for 2024
The updated earnings model below assumes revenue of $1,000m for 2024. In simple terms, it assumes revenue moves up from $200m in Q1 by $25m per successive quarter to achieve $300m for Q4. It’s worth noting these numbers are far from outlandish, as this September quarter’s guide is $150m and Upstart recorded revenue of $287m in Q4 ‘21 and $314m in Q1 ’22.
2021 | 2022 | 2024E | |
Revenue: | |||
Revenue from fees, net | 801,275 | 907,272 | 801,275 |
Interest income and fair value adjustments, net: | |||
Interest income | 20,634 | 105,580 | 60,000 |
Interest expense | (3,274) | (10,843) | (10,000) |
Fair value and other adjustments | 29,954 | (59,565) | 15,000 |
Total interest income and fair value adjustments, net | 47,314 | (64,828) | 65,000 |
Total revenue | 848,589 | 842,444 | 1,000,000 |
Total operating expenses: | |||
Sales and marketing | 333,453 | 345,776 | 300,000 |
Customer operations | 117,579 | 187,994 | 110,000 |
Engineering and product development | 133,999 | 237,247 | 90,000 |
General, administrative, and other | 122,677 | 185,290 | 110,000 |
Total operating expenses | 707,708 | 956,307 | 610,000 |
Income (loss) from operations | 140,881 | (113,863) | 390,000 |
Other income (expense), net | (5,174) | 9,473 | (7,000) |
Expense on warrants and convertible notes, net | (1,976) | (4,684) | (5,000) |
Net income (loss) before income taxes | 133,731 | (109,074) | 378,000 |
(Benefit) provision for income taxes | (1,712) | (409) | 18,900 |
Net income (loss) | $135,443 | ($108,665) | $359,100 |
Net income (loss) per share, diluted | $1.43 | ($1.31) | $4.22 |
Weighted-average number of shares o/s | 94,772,641 | 82,771,268 | 85,000,000 |
The cost metrics in the model are derived from the contribution margin, currently at a record high of 67%. There is some slippage here (47% used as an average over 2024) as marketing dollars chase more elusive borrowers, but overall costs reflect the much higher degree of automation prevalent today. A tax rate of 5% is used, given the tax shield of cumulative losses. Shares outstanding rises only modestly from 83.1m today to 85m in 2024. (Note that this fell from 95m in 2021 as 20% of the staff have been laid off since, reflecting the cost-cutting measures implemented) The result, earnings per share of $4.22 for 2024.
From 2024, Upstart should be able to grow revenue at 15% and earnings at least 20% per annum, as the operating leverage inherent in the company acquires scale and the superiority of its AI-model continues to gain acceptance in a vast market estimated to be $162Bn for consumer loans and $780bn for auto-loan s. This growth rate merits a PEG ratio of at least 1x; using a TTM P/E of 20 provides a share price of $84 by the end of 2024, after which the share price will move in tandem with earnings growth of 20% pa.
Summary
A critic may claim that the $250m/quarter revenue was achieved when the UMI index was far lower at around 1 in 1Q '22, whereas the current ratio is closer to 1.5, so the revenue forecast seems highly optimistic and so is the earnings projection. The rebuttal would be:
Yes granted, a higher UMI reflects greater consumer stress but note Upstart grew revenue by 35% in Q1 ’23, whilst the UMI rose from 1.4 to 1.5. Upstart has increased the breadth and scale of its operations significantly since December ’21: the number of partners in their lending network has grown from 57 to 100; auto-loans is a now an established vertical, with 61 dealerships offering Upstart-powered loans.
In summary, Upstart has survived a perfect macro-economic storm through which the company built on its core strengths both in attracting and securing borrowers, as well as forging links with credit sources. Now it’s time to demonstrate those strengths via earnings.
The recommendation is to Buy Upstart now, before the sails catch wind.
For further details see:
Upstart: A Perfectly Creditable Outlook Perfectly Misunderstood By Market