Description

Long Carvana

I wrote most of this before the Fed’s pivot and Carvana’s massive move up in the last few sessions.  Nonetheless, the business and the stock still remains attractive on a long-term basis. The stock has been re-priced in 2023 to reflect a bankruptcy/dilutive restructuring event that did not end up occurring.  There is significant additional value to be created from scaling the best business model in auto retail.  

Carvana has created the low-cost platform for buying, inspecting, reconditioning, and selling used cars.  It already enjoys significant advantages in unit economics which will continue to grow over time.  This is an ideal market for a low-cost operator to compete in and to win market share across multiple dimensions.  Carvana’s economic advantage translates into significant customer value, including price/value, quality, convenience, selection, and more.  

Carvana today is analogous to American Tower in the early 2000s: both had/have high fixed cost structures, high financial leverage, and significant solvency concerns from the investment community amidst a pullback in growth and worsening macro conditions.  Conceptually, Carvana today is analogous to a one tenant cell tower, but it has a growth opportunity far beyond the incremental tenancy that a fixed network of cell towers can generate over time.  

With the company’s recent focus on operational efficiencies, it is essential to look at the company with fresh eyes – free from legacy opinions and impressions from any prior era.  Additionally, Carvana’s balance sheet has been restructured to give the company ample time and opportunity to implement and execute its significant efficiency improvements which are leading to superior variable costs per unit, enabling smoother and more profitable growth in the future.  

Used Auto Economics

Used cars are commodity-like, trading in pretty tight value ranges in local geographies.  Two 2018 Toyota Camry’s with 30-35k miles and the same trim will be priced similarly in the used market, generally +/- $1,500.  For an iBuying platform at scale, pricing stability and predictability is extremely important, and the used car market has those attributes the vast majority of time.  Used car spreads are a function of (a) inventory acquisition price, (b) the selling price, and (c) the cost to recondition a vehicle and transfer it to a customer.  (a) and (b) are highly dependent on prevailing market trends, while (c) is dependent on the dealer.  We believe Carvana has vertical capabilities that translate into superior performance across all three dimensions.  

Carvana has built a used car retailing platform consisting of:

• Last-mile pickup and delivery on 1- or 2-car haulers
• Long haul transportation between locations
• A significant footprint of inspection and reconditioning centers (IRCs)
• An integrated consumer front-end that exhibits national inventory and allows customers to shop and complete a transaction within 15-20 minutes with no negotiating
• A digital and facilities-based wholesale auction business

This platform, running at ~30% utilization (i.e. will only get better at higher utilization) and <1% market share, already achieves cost per unit advantages vs. the field of tens of thousands of used car dealers in the country. 

Used car transactions have pretty standard components of gross profit:

• Retail GPU (gross profit per unit) is the gross margin on the car itself.  It is the customer’s purchase price less the cost of acquisition, accumulated depreciation during the holding period, and other expenses including inspection, reconditioning, transportation, etc. 
  • Retail GPU tends to be in the $2,000-2,500 range.
• Finance GPU is the bucket of gross profit generated from originating auto loan.  Depending on the dealer’s model, these can be origination fees, gains from selling whole auto loans, gains from selling securitizations, etc.  There also can be costs that fall into this bucket, e.g. Carmax pays fees to third parties to acquire its subprime loan originations.
  • Finance GPU varies depending on business and customer mix, but tends to be another $1,500-2,000.
• Other GPU comprises commissions and fees earned from other items attached to a transaction – GAP waiver insurance, extended warrantees, etc.
  • Other GPU tends to be around $500.

Depending on a dealer’s capabilities and business mix, total GPU per retail unit tends to range from $2,500 to $5,000.  Excluding wholesale GPU, which Carvana includes in its “non-GAAP GPU” metric, Carvana generated normalized GPU of about $5,200-5,300 in the two most recent quarters.  I normalize this for exceptional gains on loan sales, inventory adjustments, and other factors.  These recent GPUs can be decomposed into:

~$2,600 of retail GPU

~$2,200 of Finance & Other GPU

+$500-600 of D&A related to ADESA and other D&A included in COGS.

= $5,200-5,300 GPU

It also generates gross profit from its wholesale operations, but since those are unrelated to retail sales, we should bucket them separately since most other dealers do not have wholesale auction operations.  Using Carvana’s calculation, non-GAAP total GPU is closer to $6,000 per unit. 

Despite carrying a cost base for capacity well above current volumes, Carvana already is generating superior unit economics to other dealers in the market.  The chart below shows where it currently stands in asset utilization.

Source: Carvana

Carvana has idle capacity that can be “spun up” relatively easily by hiring and training additional workers in inspection, reconditioning, and transportation roles.  Most new hires at Carvana do not require prior training.  The standardized workflow, equipment, and processes make it easier for employees to onboard and train easily, and can ramp to being fully productive within a short period of time.  Carvana’s model does not require fully trained mechanics to conduct every function in the IRC workflow, the product of which is a more efficient and cheaper inspection and reconditioning process. 

Why Does Carvana Have Better Unit Economics?

This is the most important question to determine if Carvana can be a differentiated market leader in an industry that has commodity-like pricing dynamics.  Carvana’s vertically-integrated platform allows it to operate the entire value chain with owners economics rather than relying on third-party vendors to help with logistics, transportation, inspection, reconditioning, loan origination, and more.  In fact, a significant element of Carvana’s transformation over the last 12-18 months has involved fully in-sourcing all its critical functions, eliminating the use of “band-aid” third-party vendors it had been using for transportation, logistics, inspection, and reconditioning during its growth-oriented phase.  The vertically integrated economics of in-sourcing allow Carvana to “produce” a used car at a significant cost advantage that I estimate to be $1,200-2,600 per vehicle, depending on circumstances.  Additionally, its size and scale generate better pricing on multiple fronts, and its ability to acquire ~80% of inventory directly from consumers creates structural cost advantages for its inventory acquisition teams. 

The chart below shows my estimation of the differences in costs for each major component in the used car production chain. 

Of course, these costs can vary by unique vehicle, but taken together as a platform, Carvana has significant advantages.  Carvana uses large, centralized IRCs to “mass produce” used cars with better economies of scale than smaller players. 

I believe its purchasing power and Sell to Carvana business unlock another >$500 of cost advantages on top of its production cost advantages. 

If Carvana can source and recondition cars more efficiently than others because of its “mass production” scale, superior processes, and vertical in-sourcing, the fundamental question for Carvana is if the last- and middle-mile logistics and transportation links in the operational chain are so expensive to operate that the model is uncompetitive vs. brick-and-mortar dealers. 

Let’s look at two examples to illustrate the differences between transactions:

• Carvana acquires a car from an auction in Los Angeles, reconditions it in Tolleson, AZ, and then sells it in Albuquerque, NM. 
• 360 miles of long haul from Los Angeles to Tolleson. $90 of transportation at $0.25/mile.
• $1,000 of reconditioning costs
• 429 miles of long haul to Albuquerque, NM.  $107 of long haul.
• $150 of local delivery costs
• Total delivery and transportation costs of about $350. 
• Relative to the cost advantages outlined above, $350 of additional transportation and delivery costs are acceptable. 
• Currently, Carvana would typically charge a delivery fee for such a transaction, recouping some portion of this. 

Total gross total variable costs of $1,697, with net variable costs $1,550 or thereabouts. 

• Carvana acquires a car in Los Angeles, reconditions it at ADESA Los Angeles, and then sells it in Los Angeles. 
• Up to $150 of local pickup costs (if attached to an existing delivery or pickup route, would be less)
• $1,250 of reconditioning costs (higher because ADESA doesn’t yet have Carvana’s IRC scale)
• Up to $150 of local delivery costs
• Again, up to about $300 of total transportation and delivery costs.  Typically, Carvana would not charge for local delivery.

Total gross variable costs of $1,550, net also $1,550.

Both scenarios involve roughly the same net costs, but when adjusting for advertising and other customer acquisition costs, the company would prefer to scale the second type. 

The transportation and logistics costs relative to Carvana’s variable cost advantages position Carvana to be more profitable than competitors in most of the country because of its proximity to supply and demand centers – particularly after integrating ADESA.  The map below shows Carvana’s pervasive infrastructure near important population centers.  Green icons are Carvana locations, while yellow and orange ones are ADESA locations. 

Source: Carvana, ADESA, internal estimates

Carvana’s vertical capabilities and broad reach create a variable cost advantage which allows Carvana to invest in growth and the customer value proposition to win share aggressively in a stable end market.  It can turn many knobs to optimize different attributes to gain share, including:

1. Acquiring inventory at slightly higher prices = more customer value
2. Selling inventory at slightly lower prices = more customer value
3. Investing in reconditioning to establish a reputation for quality and consistency = more customer value
4. Reducing or eliminating delivery fees = more customer value
5. Reducing interest rates = more customer value
6. Reducing delivery times = more customer value
7. Increasing inventory = more growth and more customer value
8. Increasing advertising and marketing spend = more growth

A faster/better/cheaper company operating with structural advantages in a large commodity-like industry can generate scaled economies shared which should translate into significant market share gains over time.  As Carvana scales, it will further leverage its fixed costs and generate capacity for additional value to be shared with customers and/or to be invested into growth-oriented areas (inventory growth, advertising, etc.).  Carvana’s superior value proposition and customer experience should allow it to grow rapidly and “get filled first,” within the limits of its labor and asset capacity. 

Prior to 2023, Carvana had been building its platform while scaling volume simultaneously, resulting in many inefficient band-aid solutions to continue growing and creating significant obfuscation of the company’s underlying operational leverage.  The company also over-invested in growth initiatives and in customer value to stimulate significant volume growth and share gains.  After taking a breather on growth (partially due to market forces, partially due to changing operational focus) for the last 18 months, the platform benefits that Carvana has built are now more evident than ever.  Its investments in proprietary software and systems and systematic standardization of processes, workflow, and more makes Carvana similar to Amazon with its fulfillment, logistics, and delivery platform.  It has invested focus, time, and resources on building the most scalable used auto platform in the industry, and in doing so it’s been preparing for the next leg of growth. 

Carvana’s hard work on implementing operating efficiencies leads me to believe that at each volume interval, Carvana will be more profitable than previously expected.  Operating through 2022 and 2023 forced the company to become significantly more systematic and standardized across all elements of its operating platform which will make future growth easier and more profitable.

In the future, there are numerous avenues for Carvana to extract value from the market using 1P and 3P inventory of all types, serving the market as the scaled delivery, logistics, transportation, and reconditioning platform to add value to end customers, fleet owners, dealers, and OEMs over time.  This is analogous to Amazon creating a scaled fulfillment platform for 1P inventory, and then opening it up as a service to 3P sellers. 

Balance Sheet and Liquidity

Most readers will assume Carvana is on a delayed path to insolvency because of its -99% stock decline from 2021 to 2022, because of its debt restructuring in mid-2023, and because of where its bonds are currently trading.  While it surely won’t be easy to convince skeptics, let’s walk through the balance sheet and liquidity in some detail. 

First, it’s worth walking through how cash cycles through the company. 

1. Dealers use cash on hand and Floor Plan facilities to acquire inventory.  They utilize negotiated short-term financing capacity from counterparties like Ally Financial to cycle their cash.  Carvana currently has $1.5 billion of Floor Plan capacity, of which $117 million is used.  With about $1.1 billion of inventory at quarter end, Carvana could have drawn on about $970 million of cash if it wanted to.  Of course, with an 8.5% base rate on the floor plan, it’s sub-optimal to do that if it can instead use its own cash on hand to buy inventory.
2. Carvana also originates loans and holds finance receivables which consumes cash until those loans are sold.  It sells those loans opportunistically via different channels to optimize for liquidity and channel profitability.  When it sells its loans, it tends to generate about a blended 7% gain. 
3. The company also holds debt at the Holdco level.  Let’s dive into that and the debt exchange that was agreed upon in July 2023 and executed in September 2023.

Holdco debt exchange

Prior to the debt exchange, Carvana had $5.725 billion of bonds outstanding across five maturities ranging from 2025 to 2030.  It had real estate and other secured assets that could have been used to issue additional debt senior to the existing bondholders, which proved to be a strong negotiating chip.  Ultimately, a debt exchanged was consummated which accomplished the following:

1. Exchanged $5.520 billion of senior unsecured notes due 2025 through 2030 for new senior secured notes due 2028 through 2031.
2. Reduced face value to $4.193 billion, which can grow back to $5,371 billion by 2025 if the company elects to pay PIK interest.
3. Reduces cash interest expense by $456 million per year for the two years subsequent to the debt exchange.
4. Pushed out near-term maturities significantly, reducing 2025 and 2027 maturities by 88%.

In effect, Carvana and its lenders made a deal to trade its senior secured borrowing capacity for a cash interest holiday for two years.  The PIK interest option allows creditors to let the face value accumulate to 97% of the exchanged notes, which would simultaneously give Carvana more runway to generate operating cash flow.  It gave the company and its creditors much more flexibility to continue executing the operational transformation that has already shown significant progress, resulting in industry-leading unit economics.  With the model proving itself out and the company on the cusp of returning to growth mode, it would be surprising if Carvana didn’t call these notes early via cash on hand and/or refinancing at significantly lower rates. 

We believe Carvana will continue to generate adjusted EBITDA going forward and will manage the business for improving EBITDA as volume scales, i.e. its investments in customer value and/or growth will not overwhelm the contribution from new units.  There may be a temporary setback in EBITDA/unit when the company transitions from efficiency mode to growth mode and some costs are recognized before the associated full revenue and gross profit benefits are realized, but we expect there to be a gradual increase in EBITDA per unit as fixed costs are leveraged.  There should be a nice counterbalancing effect of investing in growth initiatives/customer value, offset by fixed cost leverage.  Fixed overhead expenses peaked around $2,000/unit and were ~$1,800 last quarter.  As volume ramps, this line should generate significant operating leverage.  Additional efficiency gains in variable functions should translate into growing EBITDA per unit as volume ramps. 

Comparing American Tower in 2002 to Carvana in 2023

Carvana’s fact pattern echoes with American Tower’s near-death experience in the early 2000s.  It’s worth studying the similarities since both companies share many attributes. 

In 2002, AMT was on the verge of bankruptcy.  Its stock fell from $56 in early 2000 to a low of $0.60 in October of 2002, a -99% decline.  Today, cell towers are recognized as one of the world’s greatest businesses.  Each tower is a wonderful local monopoly with great unit economics that benefits from a tailwind of mobile data growth at nearly zero marginal costs.  But in the early 2000s, AMT’s balance sheet and an abrupt slowdown in growth and change in the macro environment nearly crushed the company. 

As we can see below, the company generated significant growth in revenue and EBITDA from 1996 to 2001.  Its fortunes turned in late 2001 and 2002 when carriers started to pull back on wireless capex, leading American Tower to reduce significantly its tower development activities and resulting in total revenue (but not site rental revenue) to slow dramatically.  This was a costly hiccup in growth for a company with high operating leverage and high financial leverage (about 8x), which was used to finance the construction and acquisition of its assets.  These factors, plus a significant decline in telecom-related assets, triggered a significant drawdown in AMT’s stock price. 

AMT had spent significant capital building its portfolio of towers and was not generating free cash flow, i.e. it was not yet self-sufficient.  Interestingly, if we isolate the site rental business, it was easy to see the quality of the business in 2002 when network expansion slowed.  Incremental site rental gross margins were 85% in 2002 vs. 2001, demonstrating the leverage in the financial model when expansionary investments slowed. 

Like AMT, Carvana pursued a significant growth strategy in its early days and has used significant debt and equity capital to fund the development of its platform and its early gains in market share.  Similar to AMT, when we can see the financial leverage inherent in the model when growth slowed and the cost structure adjusted to a new environment in 2023. 

Carvana is in a scenario that is highly analogous to the cell tower industry 20 years ago. 

• It has high operating leverage
• It has high financial leverage
• Its industry faces some of the worst conditions in decades, causing customers to pull back

The breakeven point for capacity-based businesses often communicates the quality of those businesses.  For airlines, breakeven points are at high load factors, generally above 80%.  For cruise lines, they are in the 35-40% range.  For cell towers, it is 20-25% (i.e. the first tenant).  For Carvana, we can infer from recent results that its breakeven point is in the 20-25% range.  There should be tremendous profitability down the line as utilization increases. 

Valuation and Conclusion

There is a recursive feedback loop in the business such that a profitable Carvana will generate lower costs of capital, more growth, and more free cash flow that will deleverage its balance sheet and feed back into lower cost of capital, more growth, etc.  In other words, assuming our analysis of Carvana’s superior unit economics holds, the future balance sheet will be completely different from what we see today.

Carvana required significant capex and operating losses to build its verticalized asset base and its strong brand in the market.  Its balance sheet is backward-looking, i.e. it was the cost to build the apparatus and to become the recognized e-commerce leader in auto retailing.  What we care about is forward-looking, i.e. it is what can be achieved now that the apparatus has been built. 

If we look at the interplay between fixed and variable costs as Carvana grows, I believe the company should be able to achieve profitable growth such that EBITDA per retail unit will exceed $2,000 per retail unit at 1.5 million units and >$2,500 per retail unit at 3m units.  As such, Carvana should be able to generate >$3 billion of EBITDA on its current asset base, with minimal incremental capex, and >$7.5 billion of EBITDA on its current asset base after investing $1 billion of incremental capex to expand its ADESA facilities, as disclosed during the acquisition process.

Ultimately, we believe Carvana can generate over $10 of FCF/share at 1.5 million units and roughly $30 of FCF/share at 3 million units.  As such, we believe Carvana is trading at roughly 5x medium-term free cash flow and less than 2x long-term free cash flow.  Additionally, once the model is proven to be profitable and Carvana’s platform is overpowering smaller competitors, there is no reason for Carvana to stop at 3 million units, which is only about 7.5% of total market share in the US used car market. 

The company is on the cusp of accomplishing enough operational efficiencies to return to growth.  High operating leverage companies that are under-earning due to low utilization rates should trade at “growthy” multiples, and I expect Carvana to follow suit once it shows that it can grow profitably.  Once it returns to growth, with the lowest variable costs per unit in the industry, Carvana should be able to optimize its investments in customer value and/or growth expenses to generate significant profitable growth in the years ahead and completely reconstitute its balance sheet.  We believe this is an AMT-like situation that is hiding in plain sight because owning Carvana has been a fire-able offense for the last two years.  This is a great opportunity for those willing to look at Carvana with fresh eyes.

Finally, Carvana is analogous to AMT except that it has the ability to keep expanding its footprint and capacity over time, whereas the tower land grab in the US and NIMBYism has severely limited TAM expansion for the tower industry which is a key barrier that enabled the local monopoly dynamics to evolve over time.  Carvana is not operating a network of local monopolies; rather, ultimately it should operate a powerful platform that has nationwide monopoly-like power, with positive feedback loops throughout the value chain to consumers and for shareholders which further enhance its market leadership, scale advantages, and operating capabilities. 

Catalyst

Company returning to growth mode profitably while generating strong unit economics and growing EBITDA/unit.  Should be visible within 6-12 months.