In Defense of Made-Up Metrics

Plus! Scaling the Non-Scalable; VR and Platform Ownership; What is a Bank?; The Non-Recession Recession, in One Graph; Going Direct; Diff Jobs

In Defense of Made-Up Metrics

A common complaint about public companies, especially growth companies, is that they hate talking about net income but love talking about "Adjusted EBITDA," or even wilder metrics than that like WeWork's "Community-Adjusted EBITDA," a metric that stripped out not just interest, taxes, depreciation, and amortization (the "ITDA" in EBITDA) but also the cost of marketing, corporate overhead, and the upfront cost of designing and building out new buildings. It's a pretty big departure from generally accepted accounting principles; on a GAAP basis, WeWork was a massively unprofitable business, but when you "community adjust" most of their costs away, it looks pretty good.

From the outside, this looks like an incredibly aggressive form of accounting. The article linked above quotes a bond researcher as having said: "I've never seen the phrase 'community adjusted Ebitda' in my life." But if that bond researcher had looked at many restaurants and retailers, the metric itself would be familiar: "Community-adjusted EBITDA" is just a (ludicrous) euphemism for what's otherwise called contribution profit, four-wall profitability, or another measure of unit economics. For any business that scales by adding discrete units with similar economics—locations, subscribers, oil wells, bench resources, delivery drivers, etc.—it makes sense to break the economics down this way, and to think of the driver of profits as:

  1. The long-term evolution of unit economics for whatever unit the economics focus on. A well-run restaurant chain should be able to gradually trim food waste, improve labor efficiency, optimize locations, tweak opening hours, etc., in order to drive higher margins at established locations. If nothing else, regulars are a good source of predictable margin, and it takes a long time to establish them.
  2. The operating leverage as relatively fixed corporate costs get spread out over more units. Those costs do rise over time; a streaming media company with 100m subscribers will have more people working in HR, finance, and other back-office roles than a streaming company with 1m subscribers. But the relationship is much less linear, especially if some of the fixed internal expenses are simultaneously investments in better margins. (A tiny streaming company probably shouldn't invest much in optimizing bandwidth; nobody says "their content library is absolute garbage, but it loads fast and the video quality is impeccable." But at larger scale, extra bits do take a meaningful bite out of margins.)

There is, of course, the cynical reason managers prefer to report these kinds of metrics: they look better. EBITDA itself was originally created as a way to talk about the cash economics of the business independent of how it was financed—if a company has high depreciation charges for assets that don't need to be replaced after they depreciate to zero, it can lose money based on conventional accounting while still generating cash flow. And if those reported losses make the stock cheap, the optimal decision for management is to finance the company more with debt and use some of the cash to buy back stock.[1] But investors also have some anchoring bias around multiples; if you start your career looking at P/E ratios and then companies switch to focusing on EBITDA, everything will look cheap.

The move from EBITDA to "Adjusted EBITDA" came a bit later. There are some adjustments that do genuinely make sense; if a company has one-time costs or one-time gains, it makes sense to strip those out of its economics. But many companies go further than that, and exclude stock-based compensation from their adjusted EBITDA calculation.

This, too, can be defended, at least slightly: stock-based comp can be lumpy (though it's certainly possible to exclude one-time adjustments while keeping run-of-the-mill stock-based comp in the numbers). And they're a non-cash expense. Looking at non-cash expenses is good for basically one reason: it tells you whether or not a company will need to raise additional funding before it gets to breakeven, and offers some sense of the timing. A company that's GAAP-unprofitable but adjusted EBITDA profitable is still a company that is, in some sense, losing money—but it's also a company whose losses don't need to be immediately funded by shareholders. Since stock prices are determined by supply and demand, just knowing the company doesn’t need cash, i.e. knowing that there doesn't need to be a sudden increase in supply any time soon is helpful.[2]

In practice, professional investors will come up with their own metric; if you were an analyst, and you pitched a stock based uncritically on its multiple to some metric management has invented, you could expect a good portfolio manager to make you defend every single adjustment the company made.[3] But it's also quite common to take management's metrics as a starting point, especially for understanding how the cost structure responds to growth, and to make further adjustments from there. And it's generally easier to start by valuing the business and then to adjust for capital structure as the last step.[4] This is good for two reasons. First, it's a good way to separate the somewhat unrelated questions of what the business is worth and of which financial claimants get which piece. And second, it sometimes reveals that the company would be worth a lot more if it took on more debt, or if it replaced some of its existing debt with equity.

Unfortunately, there are two kinds of companies that focus more than anyone else on custom metrics that nobody else uses: the worst companies, and the best. For the worst companies, it's a simple calculation: the GAAP numbers look bad, so any alternative scorecard will look better. For some businesses, adjusted EBITDA just means "this would be a pretty solid business if we didn't have to pay people." But at the other end of the spectrum, companies with a unique model will manage themselves according to metrics that highlight that uniqueness. Salesforce put a great deal of effort early in its life as a public company into explaining to investors why annual recurring revenue didn't mean the same thing as revenue. What started out looking like a vanity metric became an industry standard, in part because it was a metric chosen by a company that set up the playbook for the rest of the industry.

This doesn't mean that the metrics companies share necessarily correspond to how they judge themselves internally, though. Meta doesn't like to give a time spent metric for Facebook, for example, even though this is a very important data point; there's too much competitive intelligence at risk.[5] The metrics these companies release are often close to the ones that matter, or are ways of sharing the outputs of inputs that they judge internally.

Using unconventional metrics is a map and territory problem. The right map depends on the purpose; the most useful map of the continental United States will vary depending on whether you're flying or hiking. Generally accepted accounting principles are also an imperfect map. They will eventually correspond to the cash a company can deliver to investors, but only at the end of that company's life. The growth of intangibles has changed the meaning of operating expenses and capital expenditures—the money Palantir or Servicenow spends on sales and marketing is analogous to the money that US Steel spent on building new steel mills, but it doesn't get the same treatment on paper. And even the nature of capital expenditures has changed; when the WSJ highlighted a rise in accounting adjustments recently ($), one example they chose was cloud companies extending the depreciable life of their hardware. But we should expect it to be harder to predict the useful life of a server than that of machine tools in a factory; computer hardware is more feasible to monitor than other kinds of hardware, and the sample size of nearly-identical equipment is larger, so we should expect there to be efficiency gains in owning it.

When management presents a non-GAAP view of profits, or they point to some new metric that they say matters more than revenue to them, they aren't telling you everything. But they're certainly telling you something.

Disclosure: Long META.

  1. Doing this repeatedly over a long period is one reason, though not the only reason, that John Malone is so rich. ↩︎

  2. Of course, stock-based compensation is its own form of supply. But there isn't a 1:1 relationship between new shares issued as equity compensation and new shares that get dumped on the market. This creates a fun nonlinearity, since employees' propensity to sell their shares as they vest is partly a function of optimism about the company. This creates a self-fulfilling dynamic; higher morale increases productivity and also gets rid of an ongoing supply of shares. If a given company's stock tends to go up, employees hold on to more shares. When they're feeling pessimistic, they sell as soon as they vest. And a great predictor of morale, especially at a business whose compensation is weighted towards equity, is the last six months of market performance. No wonder these stocks can be so volatile; one of the biggest investor constituencies is structurally likely to sell at the low, and when persistent good news restricts the supply of stock. ↩︎

  3. The traditional way to do this is to make lots of jokes about how often "non-recurring items" keep recurring. ↩︎

  4. The main exception here is a case where a company's capital structure is partly a function of its growth. Think of a BNPL business that keeps some loans on the books and puts others into structured products, a cruise line that gets customer deposits six months before the ship sails, Starbucks' app deposits, or a bank that acquires cheap deposit funding by building branches. In this case, there are a few different ways to handle it. The most useful, but most dangerous, is to mark things to market: value a bank branch based on the net present value of all the interest the bank saves from getting more zero-interest checking accounts and low-interest savings accounts. Another more useful but harder to manage option is to loosen the rule of valuing the business first and applying a capital structure second. Instead, you'd treat the financing that's directly tied to the business as part of operations, rather than financing. The meta rule of thumb might be that regardless of what the balance sheet says, if an investment bank can't help you refinance it, it's not the kind of debt you account for as part of the capital structure rather than as part of operations. ↩︎

  5. The negative view of this is that they don't want to tell investors how much growth comes from ad load and ad pricing versus usage, except in the most general qualitative terms—or when they're talking about how a product like Reels started out purely dilutive to monetization but then started producing meaningful revenue from ads. But this is unlikely; using ad load to hit revenue targets when usage doesn't measure up is just a Martingale strategy, since it means responding to lower usage by making the product worse. A maximally cynical wealth-maximizer wouldn't do that, since its net present value is lower than that of a graceful decline. ↩︎

A Word From Our Sponsors

Tegus is your go-to research destination for bold investing. We curate expert insights, analysis, and financial data to give you a powerful perspective for your investment decisions. With lightning-fast access to over 60,000 transcripts across 20,000 companies, you’ll discover a wealth of unique insights to fuel your fundamental research.

Gain perspectives, synthesize information, and model outcomes—all on Tegus.

For free trial access to Tegus, click here.


Scaling the Non-Scalable

Bloomberg has an interesting writeup of a new fund that tries to find small-scale strategies: they're aiming for thirteen teams managing $1bn (so $77m per team; compare this to Millennium's $51.8bn in assets and 290 teams, or about $178m/team—and presumably Millennium operates with more leverage). There's a tricky efficient frontier for this category of strategies: if the opportunity is too small, it's hard to find someone who can execute the strategy but who needs outside capital. If the opportunity is big, it's also competitive, and the fund running that strategy is just another multistrategy fund, but without the multi-decade track record of the bigger players. More paradoxical still is the question of how to find portfolio managers for these strategies: any attempt to list all the niche non-scalable strategies is going to select for the strategies that sound better than they are.

But there are at least two justifications for the model:

  1. Some strategies require variable amounts of capital. A portfolio of strategies that produce good returns but are capacity-constrained past $10m is really a portfolio of options on whichever one of those strategies will suddenly experience an increase in opportunities such that its capacity is more like $100m. (SPACs present some obvious and not-so-obvious trading strategies, for example, and SPAC IPO value went from $3.5bn in 2016 to $162.5bn in 2021. Hard for a single-strategy firm to figure out the optimal level of assets under management to within an order of magnitude in that environment!)
  2. Being the only person who runs a particular trading strategy is a lonely place to be. Talking shop means leaking alpha. This suits some people just fine, but in a few cases it gets old. "Hedge fund" has always been a protean category—"a compensation scheme masquerading as an asset class" is a running joke for a reason—and thinking of a hedge fund as a social club where it's safe to talk about big wins and huge drawdowns may not be too far off the mark.

(Via Abnormal Returns. And The Diff has previously covered hard-to-scale strategies and their utility as both a way to fund more scalable ones and as training for new investors.)

VR and Platform Ownership

As has been widely reported, Apple is planning on announcing their new VR headset today. VR companies expect this to be a positive catalyst for their own businesses ($, FT). Whether or not that's true depends on which iPhone analogy you prefer: when the app store launched, it created a vast new market, both for more ubiquitous versions of existing products and for entirely new categories. More recently, though, Apple has been happy to weaken the business model of ad-supported iOS businesses and to compete directly with subscription-based ones. So the relevant question is whether this launch looks like the iPhone launch in 2007 or like an Apple product launch in 2023. (Or, more to the point: will it be designed to look like the launch of the iPhone, but with a much faster evolution towards Apple capturing more of the upside?)

If this thesis works and Apple validates virtual reality as a general category, there will be one big irony: the reason Facebook's parent company is Meta and not Facebook is that the business now known as Meta made a very public pivot towards the metaverse, and one reason for that was that they were just a bit too late to have a big impact on mobile, which has turned out to be a costly outcome. For Meta, the best possible outcome is for Apple's new announcement to be unimpressive, and to make the rest of the world more doubtful about the metaverse thesis; the only thing worse than being right-but-early is being right about how big a deal a market is, but wrong about how to capture it.

What is a Bank?

The WSJ has a fun story about how there are some banks that are still able to gather deposits: the online-mostly or online-only ones ($). Banks always face a tradeoff between two sources of funding: wholesale funding at a rate and availability determined mostly by immediate market considerations, and retail funding from depositors that responds to things like branch locations and relationships rather than Fed decisions and the bank's financial position. When the market for deposits as a whole gets more information-sensitive, it acts more like the wholesale market. And the online-only banks are already used to acquiring their customers from lead-gen sites that rank banks by interest rate. It's an interesting irony of the banking crisis that banks ran into problems by buying assets that, in at least one sense, had zero risk (even if your position in long-term treasuries is underwater, you will get 100 cents on the dollar eventually). And one long-term result of this is that the assets that get impaired, the physical branch network, have nothing to do with the losses that led to that impairment, either.

The Non-Recession Recession, in One Graph

When there's a contentious ongoing debate over any topic, especially in economics, one way to break down the question is to not just ask why it's hard to come to an agreement, but why the question is a live issue in the first place. Right now, one such question is, depending on exactly who's asking, "Are we on the brink of a recession?" or "Are we already in one?" There are definitely parts of the economy that are stressed right now, but aggregate data has been surprisingly strong. So one reason for the divergent trend is that over the last twelve months, people earning $125k or more went from having the strongest growth relative to other income brackets to having year-over-year declines in household income while lower earners are still doing better. Higher earnings for lower earners tend to produce more growth and more inflation, since they have a higher marginal propensity to spend. So it's possible to have a situation where people who are overrepresented in online discussions of the economy feel like it's a recession, while people who are overrepresented in the economy itself feel just fine.

Going Direct

American Airlines is laying off some of its corporate travel staff, offering smaller discounts, and generally trying to make the corporate travel experience closer to that of regular flight booking ($, WSJ). It's a bold move: there's an immediate payoff, but that's because of one trend that's temporary and one trend that might still mean-revert over time. The temporary trend is that there's a shortage of planes and pilots, which improves pricing power for airlines, especially the largest. There's no specific reason to expect this to go away soon, but no industry stays in a permanent condition of over- or under-supply. The other trend American is betting on is the behavioral convergence between business and leisure flying, driven partly by remote work. That change is more durable, but given how many companies are asking employees to return to the office, it, too, could be a case where there's a long process of mostly unwinding the shift, and it just happens to take more time.

Diff Jobs

Companies in the Diff network are actively looking for talent. A sampling of current open roles:

Even if you don't see an exact match for your skills and interests right now, we're happy to talk early so we can let you know if a good opportunity comes up.

If you’re at a company that's looking for talent, we should talk! Diff Jobs works with companies across fintech, hard tech, consumer software, enterprise software, and other areas—any company where finding unusually effective people is a top priority.