$CVNA Kerrisdale Rebuttal - Missing the Forest and the Trees
Context: https://www.kerrisdalecap.com/wp-content/uploads/2023/06/Carvana-Co-CVNA.pdf
To be clear, I have no idea who Kerrisdale is and what they do. My main focus is on figuring out what companies do and what matters. Generally I find short reports rather lacking in both dimensions and this case is no exception. Sometimes I wonder if it is intellectual dishonesty given the $ on the line, as for whatever reason “Short Report” tends to hold more sway than a random VIC long post, but I digress.
Let’s do micro points first then high level shall we.
Trees:
We’re going to just rapid fire go down the list of claims and evaluate truth/relevance. CVNA for whatever reason seems to have a lot of misunderstanding of data even from those who should be in the know and have an incentive to be in the know. I frankly have no clue why, but let’s dig in.
“Over its history of burning billions of dollars of investor capital to manufacture topline growth, Carvana has never generated sustainable profits or free cash flow”
Why yes of course this is true and it has been cleanly shown on historical financial statements for some time now. Shift merged with CarLotz. VRM is struggling and had to entirely shift their business model. KMX is facing earnings pressure from just an initial foray into omnichannel. CVNA is clearly struggling with their cap structure. One could look at these outcomes and assess that the used car industry was simply caught up in Covid mania, but such simplistic painting of an ecosystem is rarely accurate. One could alternatively look at LAD calling out their plans to move towards omni and e-comm over the next 5 years, look at the fact the largest used auto player in KMX felt the need to depress earnings as they shifted to omni and invested in a better digital platform, and the fact CVNA was growing >100% per year prior to Covid. These people are not morons, commerce is an increasingly online activity as clearly shown in numerous other verticals. Cars are no exception and I would bet any day of the week >10% of used car sales are purely digital within 5-10 years. The Garcia family and others would match that bet and have.
So why not make money doing so? Fundamentally selling cars is a rather commoditized experience. Having a slick website is not enough to drive someone to a dealership to make a transaction. Physically building out a network is extremely cash intensive and we’ve yet to see KMX even really try, instead opting for their own store network and 3P logistics with limited delivery radius. CVNA is currently doing more delivered vehicles than all other players combined and multiplied by 2 in a world where total market share is measured by 1% improvements. Even assuming massive share loss, if the world is doing >4m used car sales digitally and CVNA can grab 1 million at $2000 EBIT/car, that’s a pretty good opportunity. Ploughing money into such an opportunity to guarantee that when people think of online car sales they think of CVNA, doesn’t strike me as a particularly bad idea. See the below from Bespoke Intel.
Fundamentally if CVNA instead took a decade long approach to scaling the economics simply don’t work. Everything rests upon the distributed network blitz which is why they did this via CVNA and not Drivetime. Companies often live within a dichotomy of growth/cut mindset and doing both is exceptionally difficult or impossible. Hard to know CVNA for being irresponsible with costs when capital is free and the opportunity is tens of billions.
“As the prospect of bankruptcy loomed, last year management began slashing costs, shrinking its operations and finessing working capital to try to generate positive free cash flow, and still failed.”
Really I’d say Q4 was the opposite if anything. The quarter was kitchen sinked. Q3 was the first time they actually started cutting anything and they were quite clear the goal wasn’t immediate FCF, that was never the target nor the expected target which is rather relevant when shorting a company valued like an option.
“Any company can grow quickly and take share if run irresponsibly on costs, especially if capital markets are willing to foot the bill.”
Please do try and make sure to let the executives at VRM/Shift/etc know they are morons.
“Rather than representing true disruptive change, Carvana is a flawed player, armed with tools no better than the competition it seeks to disrupt and led by a management team which lacks seasoned automotive, operational experience.”
I’d wager whoever wrote this at Kerrisdale has never had the pleasure of interacting a physical auto dealership? A rather simple exercise, compare the time it takes to buy a vehicle on CVNA versus just about anywhere else. I’m unaware of any “Joe’s Auto” where you can have financing and transaction complete on any car you can dream of within 15 minutes, but please do tell.
Additionally I would encourage Kerrisdale to look up a company called “Drivetime” if there are concerns about multi-cycle used vehicle operational experience.
“After repeated attempts to improve liquidity through a bond exchange failed, last week the company conveniently pre-announced “better than expected” 2Q EBITDA, not because of sustainable, fundamental improvement in its core business, but primarily due to large one-time loan sales – a move which reeks of pumping shares ahead of a potential equity offering.”
Two parter. First off, the bond exchange would be purely positive for creditors if they believe the company is fundamentally involvement. The only reason they would not exchange is if they believe the company is fundamentally solvent, or believe they will have better negotiating leverage in the future. Said leverage will certainly be lower if ADESA is used as collateral for financing via other parties, as the implied recoveries in such a scenario are far below par. APO/Pimco/DK are not morons as well, so I’m sure they have a good view on solvency given continued open market purchases of bonds.
On AEBITDA, if we call 2Q AEBITDA $4,000 per car from Other GPU (being quite aggressive and assuming very high GoS), that implies >$2,000 of GPU from retail/wholesale. Per Alternative Alpha only ~5% of current available inventory was also available at the start of Q2, so I don’t expect significant write downs/write ups of inventory, implying much of that >$2,000 of retail/wholesale GPU is in fact real.
Also per Alternative Alpha, the spread between CVNA ASP’s and KBB values has increased somewhat dramatically over the course of the quarter and is currently $200-400 above Q1, indicating improvements in retail GPU. ASP has also clearly climbed as they finished selling through old inventory.
The reasons for this are quite simple, CVNA’s inventory has rapidly de-aged. Again per Alternative Alpha.
Others looking at Q4/Q1 retail GPU as any semblance of normal is part of why I had >40% of my personal account in Carvana. Inventory was simply far too large which caused vehicles to be much older and depressed GPU’s by >$300. The difference between >90 day sales and <90 day sales was ~$800 per unit at some points in Q1. The one off loan sales do muddy the waters, but retail GPU is absolutely spiking and will continue to spike as the company improves days to sale, rolls out fees, and optimizes their cost structure. Q3 will likely be even better as total reconditioning’s are up massively QoQ which takes time to flow through and leads to lower fixed costs per unit.
“The reality is that the industry is overrun not only by scrappy local players constantly sacrificing profit to guard market share, but also sophisticated multibillion dollar behemoths actively building out their own data analytics and software capabilities to adopt the most cutting-edge tech-enabled industry standards”
Again I would encourage the folks at Kerrisdale to try and buy a car online via different providers. I would also heavily encourage Kerrisdale to put themselves in the shoes of a subprime buyer and evaluate options. The industry is anything but “cutting-edge”
“There is relatively little cost associated with making and selling loans versus the advertising, customer service, nationwide logistics network, and corporate overhead associated with selling used cars. In terms of profitability (or lack thereof), Carvana has historically made money selling used car loans at a
premium while losing a tremendous amount selling used cars.”
This is quite silly logic similar to “Costco only makes money selling membership cards”. I’m curious how many membership cards Costco could sell without Costco stores and how many loans CVNA could sell without selling cars. The only time selling loans is an issue if they literally cannot sell the loans, and even then at steady state you have competitors such as KMX simply holding the loans and riding the positive spreads. A company will make money how they make money, and if anything the most differentiated part of CVNA is the extreme ease of in-house high margin financing which KMX has rapidly tried to copy and LAD has in their roadmap. Both with no ETA on full credit spectrum support and often even losing money on subprime.
Knocking a company for the drastic advantage they have built is questionable to me.
“The ADESA acquisition was consummated at the peak of the market in early 2022.”
This is strictly false and the stock was down >70% at that time. If CVNA had truly exploited market mania and raised equity at $400 we wouldn’t even be having this conversation. Their issue was being late on ADESA and underestimating macro pain. Hardly a + in their favor, but let’s keep the narrative consistent.
“Carvana management likes to say that it currently has access to $1.5bn in total committed liquidity (excluding unpledged real estate assets). In reality, of this amount, only $488m as of 1Q23 is cash on hand with the balance primarily short-term revolving capacity available only for financing vehicle inventory and funding customer loans.”
How this is still a debate it beyond me.
Yes the floor plan is for inventory financing. CVNA intentionally lowers cash on hand by funding inventory via the balance sheet to lower interest expense. They are not taking out loans for fun, they will do it when it is needed. You would not go get a payday loan right now in case you need one in 6 months, that would be very very stupid of you. The fact finance professionals call this out blows my mind. The floor plan will be used as necessary.
“1H23 results (both reported 1Q23 and recently pre-announced 2Q23) should not be treated by investors as indicative of normalized operations because they are heavily distorted by abnormal, one-time events.”
Yes, but it turns out you can normalize below $6,000 of GPU to say $4,500-$5,000 of GPU with $1,500-$2,000 incremental EBITDA/Vehicle and get a company that is very very valuable with volume.
I would also note that 2020 to current is anything but normal, so trying to extrapolate a cost and margin opinion out based on 2020-2022 results instead of a fundamental build up of the economics is suspect. A stock does not go $400 to $4 to $24 on the back of easy analysis, so pretending it’s easy is quite indicative of an inherent ego. I have 50% of my PA currently in CVNA and will readily admit there is a possibly it can $0. Opening a short position with certainty the company is worth $0 and nobody else gets it is ignoring the market entirely.
“An S&P upgrade of securitizations sponsored by Carvana on June 1st drove a +22% one-day gain – a particularly absurd move as the ratings action was based on the performance of customer auto loans which Carvana does not hold on its balance sheet. S&P did not upgrade Carvana’s debt, it upgraded Carvana’s customers’ debt.”
My dear friends at Kerrisdale, you yourself call out YTD auto loan performance and Carvana’s economics being tied to auto loan performance as a bearish indicator. Make up your mind here.
“On conference calls, management spins these efforts as an opportunity to “improve efficiency” but we believe in actuality, the company has been guided since the spring of last year by the first order principle of preserving liquidity to stave off insolvency.”
First order principle of preserving liquidity is antithetical to what CVNA has been doing. Alternative Alpha has consistently called out and increasing regionalization of CVNA’s operations as they work towards localizing vehicle production around their demand centers. Since announcing cost cuts they have launched markets in OR/WA/IL. They have drastically increased coastal reconditioning in MA/OR/CA. They have shifted their logistics network to better match these regional developments.
Alternatively they could have simply exited these markets instead of investing in growing them. If you would like a look at what staving off insolvency looks like, take a peak at VRM. CVNA is aiming quite ambitiously at profitable growth. Calling it deceptive liquidity preservation is simply incorrect.
“Case in point: while 1Q23 headline results appear to be driven by Carvana’s pivot toward better unit profitability, sequential improvement in EBITDA and key metrics such as non-GAAP GPU were heavily distorted by transient, abnormal pricing dynamics and the delayed timing of loan sales”
Thanks I can read, as can everyone else hopefully. 1Q also didn’t really see a large benefit from loan sales. Instead of opining about 1Q perhaps we should focus on the data showing improvements in 2Q?
“At an investor conference last week, Carvana provided an update to its 2Q23 guidance highlighted by raising total GPU to $6,000+ (up 1,000 from $5,000+ prior) and adjusted EBITDA of at least $50m from simply “positive” before. Financial articles celebrated the seemingly impressive “record profits” and shares closed +56% on the day. But these metrics are heavily, if not entirely, driven by transient, one-time factors and not a function of improving industry or Carvana fundamentals.”
This is just simply false. Yes loan sales play a role, but try to provide some evidence of the claims fundamentals aren’t improving. I have provided data that ASP’s and spreads versus KBB are improving in Q2, Kerrisdale has provided speculation and narratives.
“The facility also includes certain criteria for what vehicles are eligible to be borrowed against which may complicate Carvana’s stated inventory management strategy. Per the inventory agreement, the criteria includes cars that must be within the 11 previous model years and have less than 150,000 miles. Carvana has described wanting to increase its vehicle sourcing from customers (versus auctions) as these cars are generally more profitable. The problem is, according to S&P, the average used car on the road is now 12.5 years old, up more than three months from last year’s average and the sixth year in a row the average age of cars on US roads has risen. This means there are incrementally fewer cars that are floor plan eligible within Carvana’s preferred sourcing channel.”
This is pure data illiteracy. The average car on the road is not the average car being purchased. The average CVNA inventory vehicle is ~6 years old, a far cry from 12.5. Most people simply won’t buy a 12.5 year old vehicle and the next step in that chain of ownership is being sold to a 13 year old nephew or the scrap yard. CVNA has no issue financing their inventory within this criteria and ~3% of vehicles would not fit these criterion. Bringing this up is purely narrative spinning and not substantive.
“Carvana has also drastically reduced the size of its inventory to reduce sales cycles and avoid the risk of declining car prices. As its 4Q shareholder letter explains, however, “purchasing fewer retail vehicles means fewer low age units are added to the website which other things being equal, increases the average age of our inventory.” [emphasis added]. Carvana does not disclose the average age of its vehicles (itself a red flag given the metric’s importance) but using third party data providers, we have found a material change in the age of Carvana’s cars on its website. In 1Q20 nearly 90% of the vehicles listed on Carvana’s website were between 2 and 7 years old. That figure has dropped to 63% as of 1Q23. In sum, floor plan financing, which again should not be misconstrued as a source of funds for debt service, is being drained by the reduction and aging mix of Carvana’s vehicle inventory.”
Long blurb, apologies.
This is a rather amusing total misunderstanding by Kerrisdale. When Garcia says age here he means days spent in their inventory, not model year. Age as defined by time spent in inventory is now a third of what it was in 1Q, at 44 days versus >120.
Additionally 2-7 years is entirely arbitrary and not relevant to the point they’re making about the floor plan which they themsleves called out as 11 years.
And further, those older model years actually have better vehicle margins (Per Alternative Alpha once again)
“In short, $2bn in unpledged real estate most assuredly is not the same thing as $2bn in actual liquidity.”
Yes and nobody is or has been modeling it as such.
“Furthermore, even if macro conditions improve, any resumption of growth would almost certainly require Carvana to once again increase inventory, reincur logistics costs, run more shifts at reconditioning centers, and out-shout competitors by advertising. Positive operational cash flow isn’t happening anytime soon if the company decides to try growing again”
Yes which is why we measure things on a per unit basis. If Kerrisdale would like to bet CVNA has negative unit economics I am entirely open to a 1:1 50/50 odds.
“Based on multiple interviews with former Carvana employees, a concerning lack of fundamental automotive and logistics know-how at the senior management level contributed to poor execution as market dynamics shifted in late 2021.”
I would ask Kerrisdale to point to the national auto logistics experts in the room and explain how they aren’t at CVNA, the only national auto logistics dealer. I’m sure they can do better at pointing out issues than Tegus calls.
Some random talk about CVNA using a pricing model to price their inventory that is not sensitive to sudden 500bps hikes in interest rates.
I would ask Kerrisdale where the automotive pricing experts are that can accurately predict interest rates and vehicle demand and why they aren’t a billionaire yet.
I also find this excerpt amusing “Consumers who sell to Carvana have more information about their driving behavior and the true quality of a vehicle than any statistical mode”. According to Kerrisdale the average consumer is a better pricing expert than CVNA. It is also amusing that direct purchasing from consumers is called out as a large boon by other players such as LAD and KMX doing very similar things as opposed to the traditional auction route. Apparently the entire auto industry is too incompetent to price in a margin of safety on vehicle acquisitions?
“What were our differentiators early on, are no longer differentiators at all. Like, everyone will deliver a car to you now, everyone has a return policy...Everything that made Carvana what it was…CarMax? They’ve already caught up.”
Kerrisdale please name me 10 auto dealers that will deliver a car to my house, if I’m a subprime buyer name 5.
“Vroom, Shift, Driveway, AutoNation, CarBravo, and CarMax all have easy to navigate websites that provide a similar buying and selling experience to Carvana (literally, all of them have some version of “Shop, Sell/Trade, Finance” dropdown menus across the top of their sites and a search bar for make, model or keyword).”
As we know automotive delivery is determined by website design, which CVNA of course is still far better at if you actually go through the effort of trying to buy a car from these players. The gap has absolutely closed, but CVNA does not need to trounce KMX/LAD to have vastly more scale, and the gap still does exist.
“Over the past year, Carvana has dramatically narrowed the size and breadth of inventory (reducing customer selection) and added return fees, incentivized pick-ups, and begun charging a higher delivery fee when a customer wishes to purchase a vehicle that is further away (reducing customer convenience).”
CVNA has >40k vehicles available for purchase and offers more vehicles with free shipping than are physically able to be put in any lot. While the relative convenience has certainly decreased, pretending there is any parity between selection of >40k vehicles and thousands of free shipping vehicles with a traditional auto lot is silly.
“Carvana’s version of this model [centralized logistics] does not scale effectively or seamlessly. The acquisition of ADESA was an acknowledgement of these shortcomings. Far more brick-and-mortar infrastructure (over $2.2bn worth) was needed to properly scale, physically reduce the distance to its customers, and improve profitability by lowering logistics costs. In effect, Carvana needed to look and act more like a local dealership group.”
ADESA is primarily being used to augment the CVNA network in locales where it makes sense. IRC’s were heavily concentrated in the eastern and southern US with most of reconditioning’s happening in Texas/North Carolina. ADESA is being used for reconditioning in MA/OR/Los Angeles. It is hardly becoming a local dealership group and is more so fast tracking CapEx and building network density.
“There’s a reason why the used car market exists as a highly local, fractured industry. To succeed in a cutthroat, low margin retail business, particularly through downturns, it behooves a company to have a granular understanding of the nuances that drive purchase and sale decisions, i.e., a decentralized model of local dealers making quick, locally driven decisions.”
I wonder how Kerrisdale would explain the consolidating market trends in the auto industry and also why they for some reason think human judgment is better at vehicle pricing than algorithms despite the constant market shift to algorithms.
“Given little separates Carvana from industry peers in terms of future growth rate, consumer experience, and increasingly hybrid business models, we believe Carvana should be valued on the same basis as other auto retailers”
Kerrisdale I implore you to simply try the product. It was not growing 100% per year just because.
Using peer multiples for EV / Sales of ~.45x and EV / Gross Profit of 3.0x returns an enterprise value for Carvana of less than $5bn, insufficient to cover the debt, implying limited to no value with the balance sheet as presently constituted. Carvana equity is worth $0.
Please for the love of god stop doing blind EV/Sales multiples with no consideration of other factors and the fact sales are uniquely depressed versus market expectations, for your sake.
Enough of the trees, apologies for the length, let’s talk the forest.
Forest:
Fundamentally the Kerrisdale argument is that CVNA is an irrational business model with too much debt and no competitive differentiation. There is a lot of narrative spinning and some amusing clear misunderstandings, but that’s the basic gist.
I think on the current core economics they simply don’t know what they’re talking about. The focus on finance GPU is a bit of an eyeroll as that isn’t what any rational buyer is playing for. The expectations for CVNA to work are that the core retail/wholesale economics will improve and continue to improve with ADESA and better focus. Alternative data is showing clear improvements here, as does basic logic of the synergies with ADESA. Launching reconditioning in Los Angeles for example is just a pure $ save for selling vehicles in and around LA. The fact Kerrisdale does not at all acknowledge these developments with ADESA leads to a rather surface level thesis on unit economics with no numbers. In this case getting numbers wrong can be disastrous, as if the business has fundamentally viable unit economics or not will determine drastic swings in equity value.
I clearly believe the numbers work and have data and bottoms up analysis to support that position. I can always be wrong, but at least I’m not going off narratives.
Their points on competitive differentiation also reek of poor logic and being out of touch. Conceptually I would implore readers to imagine what it’s like to buy a vehicle as a subprime buyer. Shopping numerous dealerships and getting declined financing. Spending hours of your time trying to find a vehicle that works and then getting denied by the back office is a total chore. CVNA solves this issue elegantly and charges a premium to do so, hence the rapid growth. Even for prime buyers the convenience is unmatched still. Quoting VRM or Shift as comparable is a joke. Additionally, CVNA and KMX could both gain plenty of share and still sell a large portion of cars. The market is extremely fragmented as Kerrisdale points out, compare CVNA to a mom and pop dealership and the differences are immense.
The too much debt issue is fair. I do not disagree that the capital structure is crushing, but if the core economics work (which I believe they do), then more scale fixes all ails. Whether CVNA can obtain said scale is an open question of course, but the odds of doing so increase drastically if they can prove EBITDA and give themselves a path to being able to obtain said volume while maintaining liquidity. This is why the stock has rebounded drastically as buyers get more comfortable with the core economics. The bond holders which Kerrisdale quotes as an issue are also very interested in the core economics being viable and invested believing they are. Said bondholders also have minimal interest in owning a used car logistics company versus collecting on their >20% IRR debt investments. Lots of flexibility can be afforded if the economics fundamentally work.
I personally will continue tracking unit volumes via Alternative Alpha as that is the key open question. Kerrisdale’s concerns are about 6-18 months late and more of a narrative exercise than analysis. I wish them the best but a tough first day.
As always if you’d like to subscribe you can do so below, up to you. Hope it was helpful.
Now that's a sound reasoning, based on facts and fundamental data analysis.
🙏