- Note: This was published for the Beating The Market community previously, hence the conversational language.
The Inexplicable 40% Decline Explained
For those that have been investors in Affirm (NASDAQ:AFRM) since March of 2021 with us, I would like to say, “I commend you, and I have to imagine the investing gods are looking upon us favorably for our tenacity and grit in holding this incredible company through the ups and downs.”
And there have certainly been many ups and downs.
In 2026, when Affirm is $300+/share (nothing is guaranteed), I will spend more time writing about our harrowing journey with Affirm, full of ups and downs. There will be time for that down the road.
Today, we will remain in the trenches. Today, we will consider the nature of Affirm deeply. Indeed, on this Saturday (when this was written), I’ve spent my entire day working through this thesis for the 1,000th time, ensuring that this minor hiccup has not derailed the company or our investment thesis.
And, as of typing this, I feel entirely confident that nothing has changed for our investment in Affirm.
The decline on Friday was precipitated by one investor backing out of a deal to buy Affirm’s loans. On a normal day, this might have created a couple percentage points worth of declines; however, during a time of heightened panic and fear in the market, this minor event, in the grand scheme of things, created what superficially looked like complete panic.
In short, Friday’s decline was not so much a reflection of the quality of Affirm’s loans as it was a reflection of the state of asset markets, and we will explore this in just a moment.
Delving In Further
I would like to very briefly explain the inexplicable 40% decline from yesterday:
- In short, Affirm was selling its BNPL (Point of Sale) loans to Wall Street on Friday. During this sale, one investor backed out of their agreed-upon purchase of a tranche of Affirm’s loans. This caused Affirm, and in sympathy Upstart (because why the hell not), to trade down violently.
- Affirm’s most recent securitization [from February] had a weighted average FICO score of 686, which denotes a tranche of loans composed of prime borrowers. Moreover, FICO does not accurately assess the risk attached to tranches of loans from Affirm, nor Upstart, as an aside. We will explore this further in just a moment, as this is arguably one of the most important components of the Affirm thesis.
- Couple this event with a general market malaise (the QQQ was down 2% and is presently in a bear market; 20%+ decline from highs) and short sellers capitalizing on fear, and we’ve seen a ~40% decline in the span of 72 hours.
Now, the natural tendency for investors might be to think, “Oh, no! Affirm’s loans are being rejected by Wall Street. This is bad! The smart money knows something we don’t.”
To start, the smart money also created the greatest financial crisis since the Great Depression just 13 years ago. These “smart individuals on Wall Street” brought the global financial system to its knees. What makes us believe these “smart individuals” have any ability to truly assess the quality of loans underwritten by the technologists at the helm of Upstart and Affirm? I’d rather bet on Paul Gu and Max Levchin than bankers on Wall Street any day!
The “smart money” has been in the Fed’s ear, urging them to make this or that decision. (Ackman was vocal at the bottom in March of 2020, and he’s been vocal at the top in late 2021, placing bets in both directions while having a direct influence on the Fed). $4T printed, rampant, middle class destroying inflation, and impending economic crisis later, and I’m supposed to care “what the smart money is doing?!?”
The smart money, with all due respect to those working on Wall Street, the vast majority of whom are simply trying to provide the best lives possible for their families, is anything but.
The smart money is just like you and me.
We have the same data in almost all cases.
The smart money has access to the MorningStar loan data, the same KBRA data, that we have access to.
They have access to management’s comments.
They have access to investor presentations.
They have access to 10-Qs and 10-Ks.
In the same way that we have access to these resources.
And, at the end of the day? Most of the “smart money” never bought Lowe’s in the ’80s. Most stayed away from Amazon in the early ’00s.
Most didn’t buy Intuit (INTU) when Microsoft released its “Money” product in the early ’90s.
Most didn’t buy Netflix on the dip in 2012, from which point $10k would have become $510k by 2018.
Wall Street understands that Affirm, as well as Upstart, as an aside, are leveraging unique, proprietary, and cloud-deployed AI-based models to assess the risk of each tranche of loan.
And they are investing on, more or less, the same grounds as us, using virtually identical data.
In short, just because Wall Street is in a sweaty panic does not mean we should be as well.
Speaking of data, let’s examine recent Affirm securitizations to determine whether the sweaty panic has been warranted.
Examining The Quality Of Affirm’s Loans
Affirm’s recently publicly published tranche of loans provides for us a roadmap as to the quality of the company’s lending heading into this challenging economic period. I will share the data before I comment on it so as to not influence your perspective when reviewing it:
As can be seen above, Affirm’s latest tranche of loans was prime on a weighted average score basis! Its “weighted average” FICO score was 686!
In fact, I once read that, “Affirm only accepts very high credit scores,” as one of the critiques of the company, and the data above substantiates this.
And it was much higher for non-interest bearing loans (723) vs interest bearing loans (676), indicating that there was higher yield on the loans that may be riskier.
Additionally, the investor who was purchasing Affirm’s loans was buying the AAA-rated (very prime) portion of the loans.
“A major investor in the top-rated portion of the deal, also its largest tranche at more than $400 million, was said to have backed out at the last minute due to general market volatility that may have led to wider risk premiums than the company wanted, three people with knowledge of the matter said. That AAA slice was almost entirely sold when the transaction was halted, two of the investors said.”
Their reason for no longer participating in the deal is impossible to entirely discern, as only they know the reason for their actions, but I suspect it has something to do with the following ideas.
Troubles In High Yield Credit
The central factor for this investor parting ways with Affirm’s tranche of loans was likely the rather violent sell off in the High Yield credit markets.
This spike in volatility and, correspondingly, yields either 1) spooked the investor or 2) created opportunity for the investor elsewhere that may have been more attractive; hence, they decided to go with more proven high yield offerings.
Moreover, as yield on higher risk credit increases, Affirm’s loans, which are probably usually less proven than alternatives out there, may need to be offered at a lower price (and thereby a higher yield) to attract investors, or, at the very least, this one such investor who decided to allocate capital elsewhere (not to Affirm’s loans).
And this idea brings us to the most important part of the Affirm thesis.
The Most Important Part Of The Thesis
As mentioned above, due to the relatively unproven nature of Affirm’s loans, to attract investors, Affirm, in some cases, would have to reduce the cost of their loans, and thereby drive up the yields.
For instance, let’s say Affirm offered a tranche of loans whose composite price was $100 with a 9% annualized yield.
Due to 1) the unproven nature of Affirm’s loans and 2) the increasing yield being offered out there (as risk elevates due to macro factors; remember risk = return), Affirm may have to sell that same tranche at $95, and thereby make the yield more competitive with other higher-yielding products. In this scenario, Affirm would lose money on the loans, creating worse economics for the company (and thus, a rationale could be made for a declining share price).
But here’s the most important part of the thesis:
Affirm will likely not do this. They understand the risk of their loans. They understand what the return should be. If the market rejects their loans, then they will simply hold them and generate revenue and cash flows on the yield of the loans. As Michael Linford, CFO of Affirm recently remarked,
Credit outcomes are a choice for us, and the question that we ask ourselves every day and are monitoring when we look at credit outcomes, “Are the credit outcomes we’re seeing consistent with the model when we made an underwriting decision?” We get that signal 34 days after we originate a loan, so every day we look back 34 days, and say, “Ok, we expected this much loss. Are we in line?” If we are, we pass go. We collect $200. We continue to operate the business. If we ever see this deviating, we have to react.
That’s the nature of our product. It’s so valuable. Every transaction is underwritten, and the duration is so short. We sit here with our hand on the spigot. The good news is the credit outcomes today are great, and that can change very fast, which is where we have world class monitoring.
For now, what we see are outcomes very consistent with what we want, and when you look at the rise in delinquencies, it’s consistent with what we wanted to have happen. [Upstart has said the same thing, as an aside]. We did this on purpose. The credit capital markets can withstand that level of loss. We’ll take it, and we’ll drive really great results for the merchants, and that’s what we’ve been doing.
You’ll see us behave very differently if the outcomes start to deviate from the models, and we’re in complete control of that.
Source: Affirm Investor Presentation
Affirm will doubly not start taking huge losses on their loans in a desperate attempt to get them off their balance sheet because Affirm’s loans will prove to outperform similar tranches of loans in terms of risk/return (on a losses/approvals basis), and this is the thesis underpinning our ownership of Upstart as well.
As we’ve seen in the past from Upstart, Upstart’s loans outperform other similarly rated loans.
How could this be?
Again, this is the most important part of the thesis.
In assessing a tranche of loans from Affirm or Upstart based purely on FICO scores, one cannot accurately capture the true risk/return of that tranche of loans.
This is borne out in Upstart’s relative outperformance of similarly FICO-rated loans out there.
It’s borne out in Upstart’s absolute domination of the Credit Karma marketplace.
When Affirm and Upstart take their tranches of loans to the credit markets/Wall Street, they fit these loans into the prescribed FICO rating scheme, and it generates a number, on which investors base their decisions.
But we believe, so our thesis is structured, that this representation of risk does not capture the entire picture.
We believe that investors are fitting a round peg in a square hole.
And this idea is illustrated very clearly in the Upstart model. Upstart’s model targets the swath of consumers that do not have access to Prime Credit, but who are worthy of Prime Credit. According to FICO, these individuals often fit within the 600-700 credit score range, which strongly correlates with credit scores attached to some portion of Affirm borrowers.
These are often individuals with a short credit history or who are immigrants or who are individuals who are re-entering the financial system after experiencing life obstacles. It’s truly a noble business model, as an aside, that heretofore could not be created because the technology did not exist. In developing the very strong ability to successfully target these individuals, Upstart is able to offer even cheaper credit to the Prime-est borrowers.
Upstart is then further able to offer cheaper credit to the least prime borrowers.
The same ideas apply to Affirm’s use of data/AI/ML to underwrite credit!
The entire industry’s economics are restructured. Indeed, as we’ve said often, the reorientation of the economics of the consumer credit ecosystem is only possible through “Google-like data operations”!
True differentiation can occur in this realm for companies that can very powerfully leverage data science/AI/ML. Affirm can offer a credit product mankind has never seen (up to $17.5k loan with no late fees, no activation fees, no deferred interest, no accruing interest on revolving lines of credit) because of its technological prowess, which very, very few other companies in this realm have! (most are just “the smart money” on Wall Street).
But this is not reflected in FICO!
And as such, two companies can both present a tranche of loans with a ~700 FICO score, and those two tranches of loans could perform entirely differently!
As we’ve seen in the past from Upstart.
Upstart and Affirm must jam their loans into exclusively FICO-based rating systems so as to interact with the broader financial system, but this does not convey the true risk/return profile of the tranches of loans.
Only the companies, i.e., Affirm and Upstart, truly know the nature of risk of their loan books.
Is Management As Panicked As Investors?
Well, let’s turn to the evidence.
For one, 24 hours after Affirm and the prospective investor in its loans parted ways, Max is biking:
Now, of course, that is not exceptionally material. John Foley (PTON) bikes every day!
As for Affirm’s perspective on the tranches of loans it’s recently sold, let’s turn to the recent investor day presentation with Affirm, in which Max Levchin specifically commented on how Affirm would fair in a recession:
I’m moderately excited about the likelihood of an oncoming recession because if I hear one more, “What will it look like when you get tested?”
We’ll get tested and be okay.
I cannot say the same thing about our competitors, but we’re very comfortable with our ability to manage risk, and that’s fundamentally what credits and payments is about.
In a separate note I’m writing, I recently illustrated:
Affirm, as an example, is prolifically featured on the websites of:
Product market fit is probably an understatement.
Affirm is closer to “World Domination” than “just product market fit”.
Affirm simply does not need to lower its lending standards to drive growth.
Its distribution is staggeringly big, and its product appeals to a wide swath of the population.
I don’t care if you make $125k/year.
Paying $25/month for a $1,500 grill in a no hassle manner is appealing, and this is just scratching the surface of Affirm’s long term value proposition, which we’ve illustrated ad infinitum, and into which I won’t start delving in this forum.