Carvana: Illuminating the Path

When do they press the button?

Carvana reported its first-quarter results on May 4th. For context, the company had previously provided very encouraging preliminary first-quarter results in late March that the market seemed to oddly underreact to, which I wrote about in Carvana: The Disconnect. Interestingly, Carvana’s actual first-quarter results came in at or above the high end of the previously announced preliminary results. CVNA closed up 24% the day after earnings and is currently trading about 108% higher than its pre-earnings close.

While that sort of stock price move would be huge in almost every other situation, I consider this move fairly muted. Why? Well, 18 months ago Carvana was a 50%-100% annual grower that was consolidating the U.S. used car industry with a triple-digit stock price and a ~$50 billion market cap. These days it is widely thought of as a shitco that never made economic sense that is destined for bankruptcy with a (until very recently) single-digit stock price and a ~$2 billion market cap. Before earnings, the stock was down 98% from its August 2021 all-time high of $370 per share. Today, despite the 108% move higher this month, the stock is still down 96%. So whether we’re talking about a $7.20 stock or a $15 stock, the equity market value is either $1.4 billion or $2.8 billion. Frankly, there isn’t much difference when considering that surviving this used car recession intact and eventually prospering would put the company’s stock price in a completely different galaxy.

Carvana offers investors a binary outcome. Either the company gets to free cash flow neutral before running out of liquidity resources (and without unacceptable levels of equity dilution), in which case it will be back on a path towards selling two million retail units annually and more over time at attractive unit economics. The stock price would be a clear multi-bagger from the current price. On the other hand, if liquidity resources are exhausted and they are forced to raise equity or swap debt for equity at depressed prices or even file Chapter 11, equity investors would clearly do less well.

An investment with a fairly binary outcome should be priced like an option. If it works, it can grow in value by several multiples. If it doesn’t, it could go to zero. So the investment should be priced based on the perceived probabilities of the different outcomes. To state what should be obvious, the probability of the company surviving intact and returning to growth clearly increased with its first-quarter report. The company is very likely to turn profitable on an adjusted EBITDA basis in the current quarter—which is management’s guidance—and appears to have several levers to continue improving non-GAAP GPU and reducing non-GAAP SG&A in the coming quarters. In the not-so-distant future, management is going to start to pull the growth levers again, notably by increasing advertising and inventory selection off low levels, which will almost certainly increase retail unit volume, which in turn should grow adjusted EBITDA to increasingly offset the large amount of interest expense.

In December, co-founder and CEO Ernie Garcia estimated that ~40% of the retail unit decline was self-inflicted; that is, relating to their decision to reduce advertising, inventory selection, exit certain markets, and the general prioritization of profitability over growth. The other 60% of the volume decline was related to macroeconomic conditions impacting used car affordability, namely high used car prices and high auto loan rates. So I believe bringing back a large portion of that 40% lost volume attributable to its profitability initiatives is largely under the company’s control whenever management decides to ease up on the self-inflicted headwinds. For example, when the time is right, it will increase advertising and inventory selection, both of which are known to improve traffic and sales conversion.

Pushing the Button

Management has outlined its plans in terms of three steps.

Against all expectations just weeks or months ago—consensus adjusted EBITDA for the year was -$453 million going into first-quarter earnings, which I highlighted in Carvana: The Disconnect—Step 1 is now suddenly in the bag. Carvana just reported -$24 million of adjusted EBITDA in the first quarter and has guided to positive adjusted EBITDA in the current quarter. That is a long way from the -$291 million in the fourth quarter. Here’s the context, including my guess for 2Q:

Management is now working on Step 2. That means continuing to drive non-GAAP GPU up and drive non-GAAP SG&A per unit down. Remember the key formula:

(Non-GAAP GPU - Non-GAAP SG&A per unit) x retail units = Adjusted EBITDA

While adjusted EBITDA is far from true cash flow—interest expense and capex is another substantial hurdle to overcome—it is certainly the key building block on the road there. And of course, stock-based compensation is a real economic expense, but it’s also non-cash and not relevant when evaluating liquidity.

What’s most interesting to me is the question, “When does management go to Step 3 and push the growth button?” They say they want to complete Step 2—drive the business to “significant positive unit economics”—first before pressing the button. And if you listen to them, it sounds like it could be another 9-12 months of cost cutting first. Here’s Ernie on the first-quarter call:

If you take that at face value, it seems like they plan to continue optimizing the unit economics for the rest of 2023, more or less, before starting to turn the growth dials up again. That would be a great outcome that would more than likely send the stock soaring at some point.

But at the same time, I wonder whether they can continue optimizing the unit economics while also starting to turn the growth dials up a bit. Here’s why I think that could be possible.

First, management is incentivized to grow adjusted EBITDA as much as possible as fast as possible because their liquidity resources will eventually expire if they don’t. Given their guidance, Carvana is adjusted EBITDA profitable right now, which means non-GAAP GPU exceeds non-GAAP SG&A per unit. For example, in the second quarter, management expects non-GAAP GPU of over $5,000 given the meaningful reductions in average days to sale from 1Q (>120 days) into 2Q (~65 days) and strong loan monetization due to selling down some of their elevated loan backlog. Meanwhile, non-GAAP SG&A per unit is likely below $5,000, assuming ~$390 million of non-GAAP SG&A and ~81k retail units.

That means the incremental unit economics of the next retail unit are even better. On an overall basis and excluding non-GAAP wholesale GPU of about $1,000, non-GAAP GPU is already $4,000 or more per unit, roughly split between retail GPU and other GPU. That’s management’s stated guidance for the current quarter.

And as an overall figure, even that bakes in loads of fixed costs including the severely underutilized reconditioning infrastructure that is run-rating around 300k-325k annual retail units while having 1.4 million units of capacity at full utilization. The next unit only requires the variable costs, which means it should have even higher non-GAAP GPU than the “$4,000 or more” that the overall business is doing.

As for non-GAAP SG&A, incremental fulfillment costs per unit are relatively modest. If you look back at the peak COVID era, management disclosed data that suggested incremental per-unit fulfillment costs were well below $1,000. That was a period when hiring was frozen, meaning the usual hiring for growth was absent from the numbers, but there were no layoffs either. That seems to have provided a decent glimpse at the underlying variable fulfillment costs while the business was in more of a steady state.

Exhibit 2 shows how retail units ramped from April into May and June of 2020 far more than total SG&A ex-advertising ex-D&A increased. You could call that a close proxy for cash SG&A ex-advertising. On an incremental basis, cash SG&A ex-advertising appears to have been between $360 and $625 per incremental retail unit.

It’s also interesting to consider how different Carvana is operating today than it was then. Today, Carvana is running a more regionalized fulfillment network, which has lower fulfillment costs, whereas previously it was a more national fulfillment network. In addition, today more than 40% of retail units sold are being picked up at vending machines or other hubs, more than twice the rate that it has been in the past. Pickups save around $100-$150 per unit versus the cost of home delivery. Both of those factors help explain why miles driven were down 40% year-over-year and down 12% sequentially last quarter. It seems likely that the variable fulfillment costs per unit are, if anything, lower today than they were three years ago.

So if the incremental retail unit has more than $4,000 of non-GAAP GPU and less than $1,000 of non-GAAP SG&A associated with it, that means cash contribution per incremental unit is likely already $3,000 or more. If that’s the case or is even reasonably in the ballpark, then management should want to sell more retail units. Selling a larger number of highly cash contribution positive cars is the only way Carvana will be able to grow adjusted EBITDA to cover, and eventually more than cover, its hefty interest expense burden.

Again, management is suggesting they’ll hold off on pressing the growth button for perhaps another 9-12 months. Based on this reasoning, I can’t see them resisting pushing the growth button for very long. I’d expect them to start to slowly push it sooner rather than later.

This may be happening already. On Monday, Carvana announced a new national advertising campaign. To state the obvious, the only reason one launches a new advertising campaign is to help drive volume. Why wouldn’t they? The incremental unit economics are now in good shape. Here’s Ernie mentioning that:

And here he is explaining why that allows them to grow marketing spending:

In contrast, Ernie is also saying is not the time to turn on growth yet:

How do I reconcile this? My sense is Ernie considers Step 3 of “returning to growth” to truly be returning to normal levels of growth for Carvana—something like 40%-60% annually. I don’t think he’s saying “We can’t grow volumes at all during Step 2.” After all, they’re launching a new ad campaign, which is clearly intended to attract more customers. So I’d expect Step 2 to show continued improvements in non-GAAP GPU and non-GAAP SG&A over the next few quarters and flattish (2Q) and increasing (3Q+) retail volumes over time.

After all, returning to serious growth rates can’t be instantaneous. Carvana is not going to have flat volumes for three quarters and then suddenly grow 40% the next. Returning to growth requires hiring in advance in variable roles like reconditioning and logistics. It will require rebuilding inventory off depressed levels. We will probably see advertising spending picking up. Instead of no retail unit growth for all of Step 2 and then sudden growth, growth seems more likely to gradually pick up before accelerating further when they reach Step 3.

Consensus Estimates

Here’s the current consensus adjusted EBITDA from Bloomberg:

Once again, the sell-side refuses to accept management’s guidance of positive adjusted EBITDA in the second quarter. And while management hasn’t explicitly commented on the trajectory thereafter, it will almost certainly be growing in the quarters to come. Yet the sell-side is still modeling negative adjusted EBITDA for another four quarters. They behaved similarly after the preliminary first-quarter results were released. It is truly bizarre. Again, I think the analysts are reluctant to raise numbers because it would call into question their Sell/Hold ratings, which they prefer to keep for now for a variety of reasons. It will be after the stock doubles or triples or quadruples that they’ll raise numbers and perhaps “get more constructive on the name.”

Model

Here is my updated model showing Carvana’s path to core free cash flow positive:

The second quarter should see a bump in non-GAAP GPU from average days to sale crashing lower sequentially, which meaningfully lowers depreciation per unit, an unusually large amount of loan sales as they sell down the large backlog of loans, partially offset by the non-recurrence of a favorable inventory adjustment in the first quarter. Over time, Other GPU should benefit from things like S&P’s recent credit rating upgrade of Carvana’s securitizations, which the market is reacting to today.

I think there’s still some SG&A to cut but the bulk is now behind us. My best guess is management will likely begin to ramp units carefully for a bit during Step 2 before more seriously pushing the button early next year. And frankly, my 2024 retail unit volume growth numbers of +12-13% are extremely modest in relation to what normal growth should actually look like. Despite that, even those penciled in assumptions suggest $490 million of adjusted EBITDA next year, which offsets the majority of interest expense. And then it more than offsets interest expense in 2025. Clearly, I think core cash burn gets pretty minimal going forward.

Big Picture

Carvana has been priced for death lately but seems very unlikely to actually die. If it doesn’t and there is minimal equity dilution, if any, then I expect shareholders to make multiples from this price over time. Carvana is already the most efficient it has ever been—and getting more so—and will be returning to growth in the foreseeable future, which is a scary thought. And it competes in a enormously fragmented used car market with a differentiated customer value proposition with high NPS, has an infrastructure that is basically impossible to replicate economically, and can grow its current 1% market share by several multiples over time. The market is just starting to catch on but it’s still in the 1st inning, in my opinion.

Oh, and let’s not forget that 45% of the public shares are still short. And an even higher percentage of the true float given insiders and large committed shareholders own meaningful stakes.

Stopping There

There is much more I could write about Carvana that I don’t have time for. If you have any questions, comment below or shoot me an e-mail: impliedexpectations@gmail.com.

Disclosure: Long CVNA

Disclaimer: This post is for entertainment purposes only and is not a recommendation to buy or sell any security. Everything I write could be completely wrong and the stock I’m writing about could go to $0. Rely entirely on your own research and investment judgement.