We are All Buffettologists Now

Welcome to the weekly free edition of The Diff. Abbreviated issue this week; I'm traveling, but will be back to the regular schedule Tuesday. (As all parents know, a return to work is a refreshing vacation from the relentless, fast-paced, ever-challenging world of traveling with small children.)

We are All Buffettologists Now

Milton Friedman once famously said "In one sense, we are all Keynesians now" and much less famously followed it up with "in another, nobody is any longer a Keynesian." This quote has come up a lot recently due to some very inside-baseball arguments about Milton Friedman's legacy.1 Keynes is still a relevant economist, and you have to be fluent in him to discuss the economy even if you think his ideas were wrong, or ridiculous. We're not quite at the point where everything in economics is either an extension of Keynes' work or an attempt to refute it, but he remains an influence.

There's something similar with Buffett. Like Keynes, he's among the most-studied and most-quoted practitioners in his field. (In Buffett's case, this has a lot to do with the help he got from Carol Loomis; the Buffett partnership letters don't have the same polish as the Berkshire shareholder letters, although the returns quoted will still set your heart racing.) Buffett has had a long career, and has documented it well and had it documented at length by others (Of Permanent Value weighs in at 1,178 pages, and The Snowball is 832). Because a lot of the process involves solo reading of publicly available financials, a lot of people who get interested in finance start by trying to figure out what Buffett would do. Most of us won't read about Jim Simons and try to implement quantitative algorithms, but buying decent companies at low prices—or, following earlier parts of his career, buying garbage companies at ludicrously cheap prices—is something many people implement.

There are races to anoint the "next Warren Buffett," to reverse-engineer his strategies, to declare his strategies passé based on the last decade or so of returns, or to declare them timeliness based on the cumulative results.

Breaking down how Buffett thinks about investing is a somewhat tautological exercise, but still worth doing. The core tenets are:

1. A stock is worth what a private owner would pay for the company, and that is the net present value of future cash flows.
2. A great business is one that can continuously deploy capital with above-average returns, so the business is not just cheap for the money upfront, but grows in value over time.
3. It's essential to be self-aware about which businesses you can really understand, and which are, for whatever reason, too hard.

Point 2 is worth focusing on. Usually, it's phrased in an accounting profits sense: if a manufacturer adds another facility, it gets some return, and if that return is high because the manufacturer has a trusted brand name, a better cost structure, or some other durable advantage, then it grows at a high rate. If a bank is well-run, it can reinvest its capital, use that to expand its balance sheet, and keep on growing. But many growth companies today don't grow by spending on things that show up as assets on a balance sheet—they grow by spending R&D on adding features, and marketing on promoting their products to more people. The return on R&D is very hard to measure at an individual level, and even in the aggregate it's tough to accurately attribute returns to inventors versus managers, marketers, and everyone else who is needed to scale a business. And even when the return on marketing is easy to measure, it’s generally treated as an expense, not an asset—you could treat all the money Disney has spent making Mickey Mouse famous as an asset that slowly depreciates and gets topped up with more marketing dollars, but in general putting intangibles on the balance sheet doesn’t resolve any relevant issues, and just changes the terminology around the debate.

But those unit economics matter a lot, and they were important to Buffett. For example, when talking about Coca-Cola:

Wal-Mart’s selling Sam’s Cola. And Wal-Mart is a very, very potent force. One thing that’s helpful is that they were selling it as cheap as $4 a case here. And I don’t believe that’s sustainable. That’s 16 2/3 cents a can.

It’s been a while since I looked at aluminum—and it’s down. But I think the can is close to a six-cent item by itself. The can is far more expensive than the ingredients... Distribution costs, trucking, stocking and all that sort of thing have to be fairly similar. In a 12-ounce can, there’s 1.3 ounces of sugar—which at the domestic price, would be around 1 3/4 cents per can. And that’s got to be the same whether it’s Sam’s Cola or Coca-Cola.

The Coca-Cola Company sells about 700 million 8-ounce servings—largely of Coca-Cola, but also of other soft drinks—worldwide every day. If you take 700 million and multiply it by 365 days, you come up with 250 billion or so 8-ounce servings of Coke or its products in the world each year.

The Coca-Cola Company made about $2 1/2 billion pretax last year. That’s one penny per serving.

Beautiful! It's a great feeling to know a company so well that when you hear the sound of a can being opened, you can walk through the math like this and figure out roughly what the impact is on the company's bottom line.

Buffett has a similar anecdote about another investor who made his money in water utilities, and was also obsessed with the economics of every single customer interaction with the company. “He could tell the effect on American Water Works’ earnings if somebody took a bath in Hackensack, New Jersey.”

This kind of deep knowledge is necessary to articulate why a company should be able to earn superior returns on capital. The average company doesn't, and the average company that does doesn't do it for very long. Understanding the entire supply chain, its sensitivities, where there's pricing power, and where there isn't is powerful. It's the difference between blindly extrapolating from a few good years and coming up with a strong basis for expecting similar results over a few good decades.

And it's the key driver of many eye-popping valuations of growth-stage private companies today. The focus is still on unit economics, but the spending is on Facebook ads or sales reps, not on aluminum and sugar. When a company raises at a $200m valuation and then raises again a year later at $2bn, it's partly because of general market enthusiasm, and partly because they've made a case that their customer lifetime value is predictable, and provided strong evidence that the number of customers is big enough to justify that valuation. At that point, it's a land grab: someone else who produces similar unit economics for a similar product could grab those customers—if they're still up for grabs.

Paying 100x revenue for a company with negative margins is only possible if you either ignore valuation completely or take it seriously and literally: the company is worth the net present value of all the customers it can acquire, and once the product works and the model is steady, the limiting factor is cash. Scaling works very differently for software companies than it did for consumer packaged goods brands, charge cards, insurance, and other areas where Buffett made great returns. Scale happens fast, and if future capital raising is taken into account, the payback on good investments is incredibly swift. This is a compressed version of the same long-term growth story that can be told about Coca-Cola, Moody's, the Washington Post, or other long-term Buffett holdings; instead of years of compounding at an above-average rate, it's months of compounding at an absolutely insane rate until the business reaches unassailable escape velocity.

Another element of durable returns is a monopoly, or at least monopoly-like traits. As supply chains get more complicated and abstracted, there's more room for monopolistic layers in them. So the same basic Buffett approach now has more targets of opportunity. Some of this is driven by factors like high switching costs, which show up a lot in enterprise software and a bit less often in consumer goods (though nobody likes to downgrade to a cheaper brand, unless they find that it's cheaper and better—part of the Costco magic ($) is ensuring that every time customers switch to a cheaper Costco house brand, it's an improvement). Network effects are another factor, and they're certainly not something that gets name-checked very often in Berkshire's annual letters. But if you squint, Coca-Cola is a network effects business. People are loyal to brands, and one way to get them loyal to your brand is to advertise it everywhere and make sure it's available everywhere. Universal distribution raises the ROI of a splashy ad campaigns, and vice-versa.

So, understanding unit economics is important to private company investors and to one of the best public company value investors of all time. But that's not the only part of the process. Buffett also preaches self-awareness. And here's where things get interesting.

A common experience I have when talking to tech investors and founders is their deep incredulity that so many people live in the past. "Can you believe it? He called a restaurant to order delivery." "She bought a car that burns gasoline!" And, over and over and over again: "They're still using fax machines! They email each other spreadsheets and copy and paste the numbers into other spreadsheets."

A lot of those behaviors are suboptimal, but that's because people don't all optimize for the same things. If you're defining your circle of competence, as Buffett advises, you need to have a persistently good argument for why the status quo makes sense to the people who practice it. It's not truly safe until "why isn’t there a universally adopted API for this?” stops being a rhetorical question. So if you're deep in the software business, as an investor or as a developer, a number of legacy industries drift out of your circle of competence because modeling how they work gets harder since they have so many human-intensive kludges, and so many load-bearing inefficiencies. In other words, growth-stage investors are practicing the Buffett circle-of-competence metagame when they focus on exactly the industries Buffett has famously mostly avoided.2

This doesn't mean that early-stage and growth investors have adopted Buffett's views wholesale. There's a wider range of outcomes in private tech companies than in blue-chip public stocks, so it's hard to attain a margin of safety. And, even if the valuations are ultimately justified based on detailed business logic, the prices early-stage investors are paying do look pretty stretched. Some of those unit economics calculations will turn out fine, and some will end up looking very bad. At that level of uncertainty, you simply can’t make investments with the expectation that any one of them will avoid serious capital losses. On the other hand, the same compressed time scales for growth de-risk certain things: paradoxically, these investors don’t have to worry about what the future holds so much because they’ll get relevant answers fast, and because one of the big ways the future will be different will be because of the company they’re investing in. Investors in big growth companies have to work about tail events and black swans; moving a few stages earlier means betting on those same tail events and black swans.

It's a testament to Buffett's career success that even the people who call him washed up and behind the times are ultimately using a kind of business and investment logic that he popularized, and demonstrated with great results. It's a bit like Scott Alexander's argument that CBT stopped working so well because its ideas have become part of broader culture. If you do a good enough job, and write persuasively about how you did it, a significant fraction of your acolytes will have no idea they're building on your results.