Summary:
-Low value-to-weight of used cars makes it infeasible to profitably offer pooled inventory and ship nationwide
-Used car market comprised of many idiosyncratic purchases, impossible to monopolize all inventory
-High price of cars means consumers will spend more time looking for the best deal and there will always be another seller nearby offering similar inventory at a similar price
-Comparison sites provide sellers with minimal inventory same attention as those with more
-CVNA isn’t a low cost operator, they add steps in the value chain and provide a different service
-Comparisons to AMZN are misguided, quickly compare it to Gamestop and Copart as well
———————————————————————————
Carvana, the self-titled Amazon of used cars, has become a magnet for growth investors and short-sellers alike while growing rapidly. The company has been running national ads at a high frequency recently (I am not in a Carvana “market” and see them often), proposing to consumers a better car buying and selling experience, along with its vending machines. With its recent rise in the minds of consumers nationwide, good reviews, rapid revenue growth, and progress towards GAAP profitability, it would seem that it is making headway on consolidating the used car market. While I would like the company to succeed in its goal (I am happy to be wrong because I expect to remain a used car buyer most of my life and I won’t place any bet on the stock), for if it really can provide all used car inventory to every consumer nationwide at the lowest price it would save us all some time and effort, I believe there are slight but significant differences between this market and that of other ‘consolidators’ which will thwart Carvana and prevent it from capturing value for investors.
The ‘bull’ thesis for Carvana, as I understand it, is as follows:
/The used car market is fragmented, with many buyers and sellers. Dealers purchase inventory at local auctions or from trade-ins and sell locally, often with high-commission salespeople and in brick and mortar stores. Their competition and high cost structure allow for sleazy tactics, inflated prices, and an experience that consumers despise. Further, each used seller has a limited selection of inventory, which means consumers have to shop around and spend a significant amount of time comparing options. Enter Carvana, which forgoes physical stores for a website and its hub-and-spoke inventory/distribution system, allowing consumers access to pooled inventory nationwide, lower costs, and higher convenience. Carvana will pick up your car when you sell to them and will deliver to your house when you buy from them, also allowing for test drives and a 7-day return window after any purchase made. Given its lower cost structure, higher inventory turns than a local dealer, and the economies of scale inherent in its system, it can purchase inventory from consumers at a higher price if it chooses to, and can sell to them at a lower price than competitors. Combined with its higher selection of inventory, it will make for a marketplace platform similar to Amazon, and will eventually consolidate and monopolize used car purchases and sales nationwide./
It is an alluring story and the company’s expansion has bolstered the market’s belief. Disproving a narrative which is expected to play out in the future is nearly impossible and I don’t expect to convert any believers, but would like to quickly compare Carvana’s business to that of other successful marketplace-type companies (Amazon, Gamestop and Copart). Amazon is the most-cited comparison, and while I haven’t seen any mention of Gamestop or Copart, they are better but flawed comparisons. While superficially similar, Carvana is inferior in some respect to each and I believe these differences will meaningfully detract from its results.
While many pessimists may mention the current stock valuation, if the optimist view plays out, investors could pay up to $100bn for the company and make out fine, while if the bear narrative plays out, almost no price will be worth paying. I don’t think there is much space for the firm to settle in between these outcomes in maturity (the bear case often concludes that the firm is structurally unprofitable and I tend to agree). Though I usually dislike the terms bear/bull as they ignore all nuance and lump analysts into opposing groups, they probably work fine here given the two most likely binary outcomes.
For your reference, you can see examples of ‘bull’ narratives from professional investors here: 1, 2, 3, 4, 5. The ‘bear’ cases are here: 1, 2, 3, 4, 5, 6.
On to the Amazon comparison. In book retailing, Amazon bypassed brick and mortar book stores with lower costs and better selection by centrally warehousing books and allowing consumers to purchase online. Consumers could thereby save time and browse more books from home with more convenience. Barnes & Noble has comfortable stores, but obviously can’t provide complete book selection at each of its locations. Books are a frequent and recurring purchase for readers, so having the best selection is likely the most important factor for every book consumer (each person requires a broad selection as they will purchase many different books over their life). If you want a book that isn’t in your nearest Barnes & Noble location, they either ship it to the store from their warehouse for pick-up or directly to your home. The brick and mortar locations became a detriment and unnecessary additional cost as online buying took off, so Barnes & Noble had to choose between catering to those who wanted an in-person experience or immediately eliminate its retail locations to match its costs with Amazon and move towards e-sales. It got caught in-between and has been facing same-store sales declines ever since.
The crucial point here is that Amazon could eliminate costs and certain activities which were deemed unnecessary, primarily offering books at local stores nationwide. Shipping costs for a single book from Amazon to a consumer can be a decent chunk of the total order price, but adding more books to an order doesn’t affect it much, reducing the shipping cost per book. Another important point is that production of books (and most everything else) is centralized, reducing inbound shipping costs per unit. Lastly, Amazon’s most profitable revenue comes from third-party sales, whereby other sellers retain inventory and it merely charges a fee for platform usage. This is high ROIC business and allows total product selection on the site to grow much faster than it would if Amazon had to purchase all inventory on its own behalf.
Carvana doesn’t experience these efficiencies and has to add steps to its value chain to provide its nationwide selection, primarily inbound and outbound shipping. It takes inventory of the car from the consumer’s house and delivers it on a single-space truck to the local hub where it is refurbished, sorted, then sent to another hub via 9-car trailers at which point it is delivered from that hub on a single-space truck to the buyer’s house. Often local markets are far enough from hubs, requiring the additional step of delivery on 9-car haulers from each market to the closest hub. Compare this to the typical used-car dealer which buys at auction or via trade-in locally then sells it locally- if the consumer wants delivery to their house the dealer can hire or buy a single-space truck to do so.
Unlike Amazon, which eliminates certain activities, Carvana has to include many more to offer its selection as it attempts to procure its inventory from millions of unique customers. How much extra will the typical used-car buyer pay for a specific unit up to 3,000 miles away? Would they be willing to accept a similar age/make/price nearby, even if it isn’t exactly what they want? A potential issue with any cost-efficiencies achieved by higher volumes through Carvana’s fixed-costs is that it is questionable whether the facilities are even needed or what value they provide to consumers.
Unlike package sortation facilities, Carvana’s refurbishment centers (IRCs) likely have very limited scale economies. Each car needs to be diagnosed and prepared by mechanics, and there are only so many cars a mechanic can fix per day. In package sortation, higher volumes directly reduce costs per labor hour and per building as more pieces run through the facility and pieces moved per labor hour can become quite high. There are also facilities at a few different transportation companies which can sort packages without much human labor. Whether the car refurbishment is done locally as it is with most dealers, or at a Carvana IRC, the number of cars refurbished can’t be increased without directly increasing labor hours. For example, if we say that a mechanic can only check and fix 8 cars a day, every ninth car coming through the building will need another full-time employee and will require additional space in the center. To my knowledge, there is no machine or function which can drive refurbishment costs per unit to levels much lower than a local operator can achieve with a single mechanic working out of a garage.
We can quickly dismiss any scale economies associated with higher car volumes per single-space truck or 9-car hauler. This equipment costs below $100k and $300k, respectively and these capital costs won’t dissuade any potential entrants. The supposed barrier to entry would come from the required purchase and operation of a large network of these haulers to compete in any local market with Carvana, which I don’t believe is the case. This contrasts with the package delivery business in which a nationwide network is necessary to compete in any market, as consumers in each city will require delivery to many other cities around the country and globe.
We are therefore left with a reduction in hub costs per unit sold as the primary source of potential economies of scale, which I believe will be offset by long-distance transportation costs. Ignoring the labor costs associated with the sortation of increasing car volumes onto 9-car haulers in each hub, along with the question of how many times a typical hauler load is taken apart and rebuilt as it moves cross-country (or whether each hub just waits until it can fill a hauler with cars destined for a certain hub), there are likely cheaper methods of transporting heavy loads, namely by rail. If we are attempting to minimize transport costs per car, and we need to send cars across the country, why not rent space on a rail which can move approx. 800 of them at once? This is how new cars are sent long-distance as well as most low-value to weight commodities. If it is most efficient for new cars, which have higher values per weight, it likely is for used cars as well unless the shipping distance is low or if the cars need to be re-sorted along their trip.
Further, last mile delivery is usually the most expensive portion of any trip, particularly for bulky or heavy items. It seems implausible that providing inbound pickup and last-mile delivery in the total cost of a used car could provide for a lower total cost than a local dealer with inventory held on a lot (they have no need for delivery trucks, and again, if a consumer is willing to pay for delivery it can be provided).
Below is one of the Carvana flywheels they present in their presentations:
It seems to me that this flywheel undermines the bull case. It proposes that more refurbishment centers around the country leads to having more inventory closer to customers, lowering delivery times and increasing selection. If they can lower delivery times with local inventory, then why would a nationwide pool be of value to consumers? If they are shipping nationwide inventory through their hubs, then how would delivery times be cut? This graphic seems to admit that it is possible to provide most of what a used-car buyer will need with local sourcing. If higher inventory levels locally could push out competition or provide cost efficiencies, why has the used-car market always remained fragmented despite large dealerships existing?
Though the Amazon comparison has obvious flaws, Gamestop’s used game business is a better one. In its prime, Gamestop held a good portion of the used-game and device business. There were few other options for people to turn to when they wanted to sell an old game, and few stores had the same selection and reputation for reasonable quality. This comparison comes closer but I believe it misses the mark for a few reasons.
Used games and even consoles are relatively small-value items and most people keep old consoles indefinitely. Individuals cannot easily sell their games to anyone but friends (on eBay for example), because you cannot know the quality of the game prior to purchase- there is little recourse if you buy a dud. With its retail outlet, Gamestop and a few others offer dirt cheap trade-in prices which consumers are willing to accept (why put in additional hours/effort to get a few bucks out of an old game?) and generate high margins reselling these old games slightly below new prices. It can easily and cost-effectively ship old games to its central refurbishment centers and back- shipping costs per unit are reasonable and don’t affect margins much. Consumers also buy and sell games in large quantity- you can sell fifty old games to Gamestop at once or try to sell each on your own, one by one. This gives consumers another reason to dump all their old games in a price-insensitive manner to Gamestop as they have little to no bargaining power.
On the other hand, there are many services to verify the quality of a used car- for example you can get a Carfax or a mechanic’s report. While such services would be price-prohibitive to use for old game sales, they are worthwhile when purchasing a $15,000 used car. Many of these third-party verification services allow and in the future will make it easier for one-to-one consumer transactions to occur. While I can’t easily sell an old game on eBay for a decent price, I could be very competitive with a used-car dealer posting my listing on a site like CarGurus if I have a Carfax. Many people are willing to put in the effort to maximize the selling price of their old cars given the difference it can make, and I don’t think there is much reason for loyalty to any particular buyer. In selling a used car, consumers are primarily concerned with getting the best price. The bull argument here would be that Carvana will have more selection, so you will have to trade in with them to get the car you want and with their cost structure they could offer the best price- I don’t believe either is the case.
If we envision a future in which Carvana monopolizes used-car inventory, how would it prevent local transactions between reliable and verified consumers? If I have a working used car, it’s verified to be functional, and I can post/sell it to anyone within 200 miles (and I’ll drive it to them to test-ride), I am almost certain to sell it in a reasonable period of time. I can offer a better price than Carvana because I can sell above trade-in but below retail value (as I don’t have many costs associated with selling my car). There is little value to nationwide inventory pooling when websites can already show consumers all local inventory from every possible seller. Such websites offer a platform for people selling a single car to reach every potential buyer in a wide radius, on par with every other dealer. I believe it is highly unlikely that the used-car buyer will not find a car they want (or are willing to accept) within their local market. Most used-car buyers are price sensitive and willing to compromise- I don’t think many of them require Atlanta’s pool of inventory when living in Sacramento, for example.
As the last example, we can quickly examine Copart. This is a popular GARP-compounder stock, but for those who aren’t familiar, Copart operates an online salvage auction and provides local lots for sellers to hold inventory and for buyers to inspect it prior to auction. There are only a handful of companies with similar services nationwide, and as a result, it’s an oligopolist. It may seem to be another decent comparison to Carvana.
Copart generates the vast majority of its revenue from service fees, meaning it charges others to sell their inventory on its website and store cars. They also provide verification, title, and other services. Only around 15% of their revenues come from purchases and sales of salvaged cars on their own behalf. They don’t have an oligopolist position in trading salvaged cars- they have their dominant position by offering a marketplace for many others to sell cars. Their sellers are mostly insurers trying to get rid of damaged cars, and looking for the outlet with the most potential buyers, they often sell through Copart. Most of its salvage sales are local, meaning an insurer with a damaged car in Louisiana will probably sell the car to a buyer in Louisiana. Copart benefits sellers nationwide in that it can provide the largest insurers with a lot and auction anywhere they operate- it is not often selling cars from one state to another, or shipping them cross-country (some rare buyers do this at their own expense). We can think of Copart as offering many local-auctions rather than one giant nationwide marketplace. Local operators have a difficult time competing because large insurers would prefer to deal with one auctioneer nationwide than a different one in each state, offering Copart more inventory locally for most prospective buyers nationwide.
Unlike insurers dealing with Copart, there is no particular impetus for a used-car seller to sell to Carvana. If they are offered a higher trade-in value, the typical used-car seller may go with Carvana for a purchase, but monopolizing the industry would require a stranglehold on selection in all local markets. I believe the Carvana bull case falls short when considering the need to pull all potential sellers onto its website. I imagine Copart understands the difficulty in taking inventory of all its seller’s salvaged vehicles- it would likely be impossible due to the capital requirements as well as the fact that competitive inventory can pop up anywhere at any time. As long as new cars are sold, there will be consumers with myriad selling options and Carvana would have to somehow funnel all these future used-cars into its inventory.
If we attempt to consider a used-car market with the lowest possible costs and prices for consumers, it would probably comprise many local-sellers with limited costs and a willingness to accept low profits (which is exactly what it looks like today). Similar to a Copart auction, which is remarkably efficient, used-car buyers currently have access to hundreds of thousands of potential units within their market and little need to pay for cross-country shipping. I believe in the future there will be better verification services, less friction/sleaze, and tools which help consumers find the best possible deals, but I can’t fathom how any firm would profitably dominate this industry nationwide.
I didn’t think it would be worthwhile to bring up the financials, as it often leads to quibbling about how they will change in the future (not to mention the fact that Carvana rents most of its space from related-party Drivetime or to question how the Up-C structure could affect owners or the ongoing insider stock sales). I would only say that most of Carvana’s progress towards profitability has come from its receivable securitizations and the value of the associated residual interests held. Despite management’s recent publications (1, 2, 3), there is little information to be gleaned from its securitizations, as there are no associated public filings. I don’t know whether it is reasonable for a related party to pay above-par for its receivables, and these transactions generate non-cash earnings subject to the whims of Carvana’s estimates. Some info on its recent securitization here- 1, 2. On another note, there is also risk associated with its financing of inventory and questions as to how its growth could be affected if credit ever dries up. From my calculations, they are not yet close to generating a profit from selling cars and it is extremely difficult to estimate what value to place on the firm’s potential finance/insurance revenues if they will never generate superior returns on capital from selling cars.
There are many intelligent professionals with a different opinion than I on this company and stock; don’t take anything written above as a recommendation or suggestion. I haven’t and won’t take a position in any of this company’s securities.