Why are you investing in Carvana stock? – Clifford Sosin
- Clifford Sosin had nearly 40 percent of his portfolio in Carvana after the stock rose (2019).
- His buy limit: no more buying once a position exceeds 25 percent of the portfolio (2019).
- Shorts made up less than 2 percent of his portfolio (2019).
- Carvana had almost 20,000 cars on its site, versus about 150 on a local dealer lot (2019).
- In his scenario, Carvana sells 15 million cars a year in 15 years (2019).
You are currently viewing a placeholder content from YouTube. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More InformationCarvana (CVNA)
CarGurus (CARG) · local car dealerships
Alphabet (Google, GOOGL) · Amazon (AMZN) · Uber (UBER) · Toyota (TM) · BMW (BMW)
Why is Carvana Clifford Sosin’s biggest holding?
TL;DR: Carvana grew to nearly 40 percent of his portfolio, above his 25 percent buy limit. He sees a better offer, better unit economics and high barriers (2019).
How big is the position?
[00:00:06] 
[00:00:09] 
[00:00:14] 
[00:00:16] 
So, as to why I like it. We talked earlier about this idea that it’s the big things. Over long periods of time, it’s the big things that matter. And so, in the case of Carvana, is this a better value proposition for customers? And the answer is definitely, by a wide margin. And then the other question is this a business that has really good unit economics in delivering that? Is it a more efficient way to operate a car dealership? And the answer is yes, absolutely, by a wide margin. And then the last question would be, are there barriers to competition here that will allow you to preserve those economics over time? And the answer is, it seems to be absolutely there, it’s just that this is an immensely difficult business to duplicate and it gets much, much harder in the presence of Carvana. It was really hard enough to build without Carvana, now that Carvana is there, it’s almost impossible.
And so, if I’m right about those three things, Carvana is going to be a very, very, very big company and a very, very profitable company. And it’s got a great set of people who run it, really among the best. And so, it’s a long journey from where they are to where I think they’re going to eventually be. Frankly, at this point where they need to be to justify what is a lofty price, relative to the current performance of the business. But I’m reasonably optimistic that they can do quite well.
What sets Carvana’s managers apart?
[00:02:42] 
[00:02:45] 
So I find that often enough I’ll meet with people who run a business and I’ve had a chance to think and have a perspective on their business that might be a little different. And maybe I’ve thought about things in a slightly more detailed way about some aspects of the business because I’m just not distracted by actually running the business. And at Carvana, it’s never the case. I think I’ve thought about this stuff, I think I have some interesting insight, and they explained to me, “Oh, yes, we thought about that, but did you think about this? You have to factor that in you know, Cliff. Don’t worry, we’ve got this.”
And so, they don’t actually say that last bit. And then they say, thank you for the insights, that was really clever, but here’s the other thought. So, they’ve thought this through very well. They’re very, very bright there. When you meet people from Carvana, who are in a role, and you start talking to them, you realize this is definitely the most qualified person to do this role of any person I could imagine. And then you start asking them questions and they give you very sensible answers that reflect a great deal of wisdom and practical acumen. And frankly, then you just walk around and talk to people and they’re having fun and they’re smiling and they’re working their butts off and this from energize is absolutely cool. And so, that’s another reason. What they’re doing is hard, and so it’s going to require a lot of talent to do it and they have the talent.
Why does Clifford Sosin short stocks?
TL;DR: He looks for bond-like, boring shorts that are about to get hit by a truck. Shorts were less than 2 percent of the portfolio (2019).
Why does he short?
[00:04:32] 
[00:04:44] 
So, when it comes to shorting, to define shorting don’t do’s, I don’t short dreams, I don’t short pyramids schemes, I don’t short highly shorted stocks. What I’m really looking for is something that’s bond-like and boring and going to get hit by a truck. And those are really hard to find. And constantly we’ve done relatively little, but the ones that we’ve done have been reasonably successful. The return on capital is great because it’s all incremental. The return on time has been crummy because the positions are small and the dollars are… out of the matter, small. But again, if you think this leads to one great outcome over the next 10 or 20 years, then it’s worth it.
How big are the shorts?
[00:07:06] 
[00:07:10] 
The other thing is, you really need to size your shorts small, relative to the float of a stock. Because if borrows are hard to get or whatever, you do need the ability to exit. So, it’s important for you to know, you can own 8, 9, 10 percent of the company. You really short why a percentage or two percent of the company at your peril. And so, for those reasons, I think it’s a very risk/reward in this particular short but there is real practical constraints in terms of sizing.
What could Carvana look like in 10 years?
TL;DR: Operating costs probably more than 1,500 dollars lower than a dealership and almost 20,000 cars on the site, he says (2019).
[00:08:07] 
[00:08:18] 
And so, the economics are just totally different. And if you think about the tradeoffs, what Carvana has is that they have a pooled national inventory, versus a local inventory, which is the cars in that lot. Or if you’re searching on like a CarGurus, the cars in that geography. In order to affect that pooled national inventory; what they have is, they have IRCs, which are Inspectionary Recondition Centers around the country. And that’s where they bring cars they’ve purchased to, they then reconditioned them, and they store them there and they’re part of the pooled inventory. They’ve connected them with a hub and spoke logistics system that they can transport cars between the IRCs, and then from the IRC to a local market hub. And then finally, via I assume a car hauler, to the end customer.
The hub and spoke, that is what allows them to pool the inventory. And so if you can look at the two models, Carvana doesn’t have the dealership location cost, and it doesn’t have the salespeople cost. And because of the self-service nature of the web, they are able to… and so the averaging of traffic that you get with larger groups, they have much lower kind of sales support costs for people who have questions or whatever. And also, because the Inspectionary Reconditioning Centers are bigger, they have economies of scale inspecting and reconditioning cars to a standard. So they’re more efficient in all those senses. What they give up, is that they have the added logistics costs of trucking the cars around. However, it turns out that while trucking a car is very expensive if you hire a third-party hauler to do it, the reason for that is that there’s a very low density of car transport between cities. And so, car haulers often run at very low occupancies and have to follow these funny circuitous routes, to try to build a reasonably full load.
Good Investing Plus
Meet the experienced investors
Plus is our application-based community of 200+ investors. Around half run or support a fund, or allocate capital.
- Online: idea exchanges and events every two weeks
- In person: gatherings in the US and Europe
- Omaha: our 150+ member conference in AGM week
Carvana fixes that problem by collapsing…. So if you were to call up someone to ship a car in the U.S from City A to City B, they will usually say it’s going to be sort of… we’ll pick it up in the next two weeks. We’ll get it there some time, two to three weeks after and we’ll charge you between $0.75 and a dollar a month. By having a hub and spoke logistics system, Carvana can basically collapse the cost of shipping a car down to roughly what it really costs to ship a car. And if you have a full truck and you’re driving your car, it ends up being about $0.16 a mile to move a car a mile. And the cars go 50 miles an hour. And so, obviously, there’s a huge difference between $0.16 a mile and 50 miles an hour and like sometime this month and a dollar a mile. So, the logistic system expenses come down a great deal.
And so, what you have now is a system where Carvana is just a much…. So, if you add it all up, Carvana’s operating costs are probably more than $1,500 lower than a dealership on a per part sold basis. And then there’s really the other benefit, which is that because they have a pooled national inventory, they can offer a radically better selection. Carvana has almost 20,000 cars on the site right now. And you know, a local dealership will have 150 cars in the lot and even all the cars in an area will typically average about 15,000. So Carvana already has more cars than the average market has in all the dealerships in the market. Obviously, as Carvana grows, their selection is going to continue to expand and that will make the offer ever more advantaged, versus competitors.
Also, in order to maximize revenue, dealerships have to haggle the price, there’s this long, miserable selling process. Carvana can get rid of all that, there’s a much higher level of convenience, in that you can work from your couch and it can be brought to your home. So if you line up the consumer proposition and price, selection, service…. Price, selection, service, and convenience, Carvana dominates on all the metrics.
And so, for all those reasons, I think ultimately Carvana can become the dominant way that people buy a used car in North America. And that’s probably not going to happen in 10 years. It will probably take longer than that. But if you’re on that path in 10 years, you’re going to be heading a lot higher than what I’ve said. And there is a long discussion we’ve had about the unit economics there. But one of the nice things about the business is that as it gets bigger, it gets more efficient in basically every measure. And right now, it’s loss-making because it’s too small, and that’s part of the reason why it’s so hard to replicate. But as it gets bigger, the unit economics get very attractive and then even more attractive.
Why does Clifford Sosin not own FAANG stocks?
TL;DR: 5 years earlier he would have picked Google or Amazon, but he thought he owned better things (2019).
[00:14:19] 
[00:14:32] 
[00:15:10] 
[00:15:14] 
How big could Carvana become in 15 years?
TL;DR: About 33 million younger used cars a year in the US and Canada. In his scenario Carvana sells 15 million of them a year (2019).
[00:15:57] 
[00:16:05] 
[00:16:06] 
[00:16:07] 
If you then assume the question is, what percentage of cars will Carvana get? And if you make up the assumption that they get a third, then that would be 11 million cars. There’s no reason they could go to a third, but if you look at a lot of other industries that are online, there’s been a lot of inroads, it’s not inconceivable that it could be a third. I’ve joked with people, that if you were to have a neighbor and you were talking to your neighbor and your neighbor says, “What are you doing this weekend?” I’m like, “Oh, I plan to go to five different car dealerships and test drive cars and look at them”. If Carvana was of that size: You’d say, what are you? Crazy? Like, you’re going to pay more, you’re going to have the worse selection and you’re going to have the worse experience. Like why not here? Let me show you on your phone. I can get you exactly the car you want, right now. And so, I find, wonder why I’m stopping at a third. But a third seems like a lot, so we’ll stop there for now.
The next thing that happens is people, buy and sell cars about every 6 years. And one thing that’s really interesting about Carvana, is that if you make buying and selling a car easy, you allow people to get the car they want as opposed to the car that’s available. If you take a lot of the hassle out of it, make it even fun, you’ll lower the actual cost because Carvana offers a discount on changing cars. People should do it more and the exact amount more they’ll do it, is difficult to know. But it seems like they should do a lot more and the experience in other markets is that they do it a lot more.
So, if you kind of put those factors together and say that Carvana could sell 15,000,000 cars a year, in 15 years. Then if do some math about their unit economics or whatever, you get that the company could make $35 billion a year, maybe more, more than $40 billion, depending on your assumption. So, 40 times 20 is $800 billion; the company will generate tens of billions, maybe more between here and there as profit parameters. And so if that were to come to pass, that would be a tremendous investment. So that’s why we own it.
What do self-driving cars mean for Carvana?
TL;DR: Shared fleets drive many empty miles. Ubers drive roughly half their miles empty, so ownership still wins in many cases, he argues (2019).
[00:19:06] 
[00:19:32] 
Exact measures of what that would be, are hard to come by. But you know, if you look at New York City taxis, they drive almost half of their miles empty. Actually, more than half of their miles, empty. Ubers, I have been told, drive roughly half of their miles empty. Long-haul trucking, depending on the company, is often as much as like 15 percent empty. And so, a shared system of this nature would likely have a fairly high percentage of its miles empty. If you think it’s a third, effectively that means you have to drive 1.5 miles for every one mile of occupied travel. And what that does, is you know, a car sitting idle is gathering cost because of depreciation and capital, but a car driving is really gathering cost. Because you know, even if it’s in a very efficient energy, electric vehicle, you’re still using tires, you’re still consuming energy. You have some degree of hazard from accidents and even if people who have got self-driving systems, animals will still drop in front of it, etc.
And so when I’ve tried to work through what the fixed and variable costs of an electric vehicle would be in an owned paradigm. So, I own my own self-driving electric vehicle, versus in a shared paradigm, so I participate in the self-driving electric vehicle fleet. I’ve worked out that the savings for, if you compare buying a car and owning it for its whole life versus participating in a shared fleet over its whole life, I’ve worked out of the savings are modest, or often reflected probably negative. In other words, you probably would have to spend more if you’re an average American, using a shared vehicle fleet, than you would, owning your own vehicle.
Now, that’s an interesting starting point, because it means that like for the average person, there wouldn’t be much in the way of savings to switch, if any to switch, to give up their car.
The next thing to think about is that’s a bit of a false comparison because if you look at what people actually spend per mile of travel, many people have many different cars. And so, comparing the average car owned over its whole life, to kind of a shared vehicle fleet misses the point. If someone is very cost-conscious, they can already own a car that has almost no capital cost. They can buy a 9-year-old Toyota Corolla and pay a few thousand dollars and have a car with a depreciation of capital cost per mile, very, very low. And so for a cost-conscious person, there’s no way, like buying a used car, an older used car, there’s no way that a shared vehicle service would be a more efficient way to travel unless they drive very little and live in a fairly dense area.
On the flip side, for somebody who is buying an expensive new car, the cost savings of using a shared service would be substantial, but that person has already given a revealed preference for a willingness to spend a fair bit more for a different experience. So, the person who buys a BMW 7 Series and is incurring, I don’t know what the number is, like a dollar mile of cost to travel, $2.00 a mile of cost to travel. The fact that there’s a $0.20 a mile option out there in kind of the old Toyota Corolla, probably isn’t relevant. And now, the fact that there’s a $0.50 cent option if you bought a median… here if you want to median car, isn’t really relevant. And the fact that now somewhere there’s going to be a $0.45 shared option, it doesn’t strike me as particularly relevant. And so my sense would be that those factors would mean that, if you do the individual comparison by consumer, it makes the market look a lot smaller than you might think. Ownership still seems to win in a lot of cases.
If you then go, there’s another argument too, which is that a shared vehicle fleet is in many ways importantly, experientially, dynamically disadvantaged versus an owned fleet. The simple, for example, you need to add wait times into effect and the average trip in the United States is something like 10 miles in 20 minutes if I recall. And so even adding, just a couple minutes of wait time…. Let’s say the savings are going to be $0.05 a mile, $0.10 cents a mile to use the shared system. And so we’re talking about saving, $0.50 on a typical journey. You know, giving up 3 minutes of your time to save $0.50, works out to be a pretty crummy wage. Moreover, in a shared system, when people have built models about how wait time should evolve, you have varying abilities in wait times; there are peak times and they’re non-peak times. And there are distributions of wait time, depending on the actual ride. And so, it’s probably necessary to think about… You know, the way consumers would probably think about it is, they probably would think about the peak…. The bad day experience at peak times. And so when you think about it that way, wait times… the average wait times probably understates the cost that consumers would incur, because if you know that sometimes it takes 10 minutes to get a car, then now you need to plan, call a car earlier, even though on average it only takes 3. And if you value your time, you kind of create this dead time.
The other thing is, there is a great deal of variation in the vehicles that people choose to own. Pickup trucks and minivans and big cars and small cars. And shared vehicles fleets have a lot of really great economics and have to have fairly homogeneous vehicles. And so people wouldn’t be able to match with the vehicle they want. They also, these vehicles would be shared by a lot of people. So it’s public space. You’re not going…it’s going to be dirty. You’d have to factor in cleaning costs, and even if you don’t have cleaning costs, the person before you should come and left there and now it’s gross. So the reason that people wouldn’t want to do that…. And also, there’s an honest convenience factor, like leaving your gym bag in your car for your workout after work or running multiple errands and having your stuff in the car between. Having your kid’s car seat installed, leaving your sunglasses, you know this, that and the other. And frankly, those reasons probably get much more important in a self-driving vehicle context, because now, this is essentially a room that you occupy while you travel from point to point. And so, it becomes more of a probably intimate personal space than it would be for a car where you’re kind of occupied driving.
So those are all reasons why people, I think would be willing to pay a premium to have their own vehicle. And I’ve already said this at the start about how it would probably be cheaper. Another couple of factors to consider. If there would be meaningful transition costs to move from owned vehicles to shared…. In other words, for people who own a vehicle, switching to a shared fleet service would be costly, and so, that would, even if it were better equilibrium, you could think about there as being meaningful transition cost. And so, it would have to be sufficiently better to cause people to incur the transition costs to make the adjustment. There are markets right now, if you look at India, labor is very cheap. And so, one could argue that economically, cars are already self-driving. But you don’t see mass participation in shared vehicle services, what you see are individuals who have their own car and driver. And so you know, the other thing is, lastly, if you can model those things, the self-driving vehicle market in that model gets smaller and smaller and smaller, and keep cutting pieces away. And that reduces the quality. These things are much better at much higher densities and so, as I reduce the number of vehicles in the shared fleet model of the world, I would in turn presumably reduce the quality of the shared service, which would have a knock-on effect of lowering the use of it. If you look at other markets, be it second homes, or RVs, or boats. There’s a pretty clear reveal preference amongst people to own things, versus to share them. And so, when I put all that together, I think it is, to the extent this technology gets developed and rolled out et cetera, et cetera. There is a niche that it will occupy, and it will be a nice part of the transportation system and it’s a great way to have a third car, as opposed to… You know, to go with two cars as opposed to three, in the family or one car, as opposed to two. Or for people who drive infrequently or short distances, or there’s a lot of people who actually…. There’s a lot of use cases. But as the mainstay, the way that people travel to and from the places that they go every day. My impression is that car ownership is a better economic model kind of for a society writ large, and that it would win over time. Obviously, if that… even if I am wrong, heterogeneity across density by market and consumers and you know, different places and people and incomes and all of that, should mean that we’re talking about pieces of the industry, not the whole thing. And so, when I put it all together, I come away thinking that, this isn’t a threat I need to worry about much about, over time. One last fun fact is that, outside of major cities in the U.S, if you surveyed people and asked why they used Uber, the answer you get which counts for 90 percent of users, are some very high percentages: Are parking and drinking. Makes sense, right? I’m going to dinner, I’m going to have some drinks, I don’t want to drive or I’m going somewhere, I don’t want to park, right? If you have a self-driving vehicle, parking and drinking aren’t a problem anymore, because your car can park itself and come when you call it, and drinking, obviously, it’s self-driving. So, one might ask the question, would the self-driving business be bigger or smaller, if they were self-driving vehicles that you could own? Because you could make a case that the shared vehicle like Uber, if so many of the use cases are for drinking and parking, if I got rid of those use cases and everyone had a self-driving car, yes, the cost for Uber would go down, but so would kind of the benefits? So anyway, that’s my sense. Obviously, if I’m wildly wrong on that, then Carvana is going to be worth less than I think.
What advice does Clifford Sosin give young investors?
TL;DR: Build a real business instead. 100 people with larger funds could allocate capital about as well, he says (2019).
[00:32:11] 
[00:32:20] 
[00:32:22] 
[00:32:25] 
There are a lot of good things to do, investing isn’t necessarily particularly good for the world. It also isn’t natural for most people. People drive an enormous amount of satisfaction from flow states, from continuous improvement, from getting feedback for the work they do every day. From being part of like an energized team, like tackling a project together and getting to win like all at once. Investing doesn’t have all of that. So, you make a decision, you need to think about Carvana for a while and then you make your bet, and then, it goes up, it goes down, it goes up, it goes down, it goes up…. Five years later, yes, you’ve made a lot of money, but you know, the whole process felt nothing like, building a chain of coffee shops, where like, you have an opening and the sales go up and you’re very happy and you hire a great employee and you’re happy and you have all these little wins along the way. Whereas, like owning a stock, that’s just not what it is experientially. And the process of sitting around and thinking all day and reading and talking to people. It’s just not what people are built to do, which is why I can exist, because I happen to be built more to do it than most people. But on average, people find much greater satisfaction in more real businesses. I think they just find more human satisfaction in doing it. And they also probably serve society better. And so, frankly, given that on average, most investors do poorly, they’ll probably do better. So, you know why not pursue that, and if you really must come my way, it’s really hard to get into the business that I’m in. Most institutions aren’t set up to do it. There are a lot of reasons why they’re not but basically has to do with the principal-agent problem between investors and allocators.
And so if you want to practice true long-term investing like I do, ultimately you’re going to have to start your own shop. There’s just not a lot of places to do it and it’s a long process, it has taken a very long time to build this. And so there’s no king’s road to geometry.
[00:36:07] 
[00:36:12] 
[00:36:14] 

- Carvana was nearly 40% of the portfolio (2021) [00:00:16]
- Shorts were less than 2% of the portfolio [00:07:10]

