Newsletters Can Move Markets
Plus! "Real Men Have Fabs"; Rollups; Platform Control; Outperformance; Chips and China; Double Zero
Welcome to the Friday free edition of The Diff! Subscribers-only posts this week:
Reshaping Competition looks at the Biden executive order from last week. In industries where the average company is well-run, reducing barriers to competition can lead to more concentration, because the best company has an easier time growing into its natural market share.
Weber and Traeger: two grill manufacturers filed to go public within a week of each other. These companies illustrate the shift from one-time transactions to continuous ones, and their relative performance highlights the importance of unit economics.
Nextdoor: Old Model, New Networks: social networks have network effects (it's right there in the name!), but as it turns out, there are many real-world networks and it's hard to optimize one service to cover all of them without creating uncomfortable overlaps.
Thanks for reading The Diff. If you like what you've been reading, please subscribe!
Welcome to the weekly free edition of The Diff! This newsletter goes out to 23,207 subscribers, up 111 since last week. In this issue:
Newsletters Can Move Markets
"Real Men Have Fabs"
Chips and China
Newsletters Can Move Markets
In June of 2020, a group of advertisers announced that they were withdrawing from Facebook in response to some of the company's moderation decisions. A few smaller advertisers started it, but by June 26th, the boycott had momentum: Unilever cut its estimated $11.8 million Facebook ads budget to zero, and Coca-Cola joined in. Facebook's stock had closed the day before at $235, but dropped to $216 intraday and opened the next Monday at $210. By the end of the day, Facebook had recovered to $227, and by July 1st it was trading back where it had been before the boycott was big news.
There were many reasons for this recovery. During an event like this, the main questions investors are asking are:
How bad can this get, and
How long can this go?
Big advertisers leaving made a stronger case for "worse" and "longer," so, naturally, the stock went down. Anyone with ad spend estimates could start marking down FB revenue: a bit less from Unilever, a bit less from Coca-Cola, some odds of losing more revenue from other spenders, multiplied by their spend.
It's not perfectly clear what drove the recovery, but a very good candidate is an email from Stratechery, which pointed out that Facebook was absolutely swarming with long-tail direct-response advertisers, and that any time Unilever won an ad auction there was probably a Shopify store or a casual game right behind them. So losing an $11.8m customer didn't mean losing $11.8m; it probably meant losing a single-digit percentage of that. And it means getting proportionately less revenue from advertisers who are choosing from a variety of options, and more revenue from companies that couldn’t exist without Facebook. As far as Facebook's financials were concerned, a big, visible boycott was a nonevent. (Facebook revenue, as it happens, beat consensus by 7.6% that quarter.)
FB is a large-cap stock, and it's rare for one person's opinion to move a single stock that much, so this might be a special case. And there were plenty of other people opining about the situation, with varying levels of information. It's not a certainty that Stratechery moved the price that much, though the timing lines up nicely.
And that's a good indicator of an important trend in finance: newsletters are increasingly able to move stocks. Some of them, like Non-GAAP Thoughts and The Bear Cave, seem to have a visible impact right away—not every time, and in both cases the theses in the newsletters are usually about moves much larger than the ones the stocks immediately see—but prices regularly react when both of these newsletters publish. It's an interesting phenomenon for individual investment researchers to be able to publish market-moving research, and it's worth dwelling on. Once a newsletter has a reputation, and its readers have collectively high assets under management, it becomes much more likely that they'll affect a stock, but even a fairly obscure publication can have an impact. This new newsletter made a persuasive case for a small-cap stock—weak company, tough industry, but very cheap and with near-term catalysts, and the next day the stock was up 10.4%. (Disclosure: by "persuasive," I also mean "persuasive to me," and I bought a little. It's now down, unfortunately.)
Many of the new crop of financial newsletters are pseudonymous, and some seem to be written by financial professionals who are moonlighting, between jobs, or perhaps switching to a business model they enjoy more. Others are written by interested outsiders, or people trying to break into the industry. (It worked for me!) And this kind of thing has been happening for a long time. In high school, I stumbled across a writeup of a ridiculously cheap Mexican chicken company, and bought some. I might have trusted the writeup less if I'd known it was written by a former neurology resident with almost zero professional financial experience, but all I had to go on was the quality of the writeup.
A trend towards individual researchers, some of them pseudonymous, writing things that move stocks smells suspiciously similar to the other big online-driven market-moving trend of WallStreetBets traders pumping up stocks. But it's not the same thing. There are certainly some people on WSB who are doing a fair amount of research before they make their pitches, but sometimes this research is performative. This post, for example, is great: it's a look at the returns from buying companies that top the Fortune Best Places to Work list. But some of the research is pretty shoddy: SPAC projections taken at face value, comps to companies like Tesla that are simply incomparable, conspiracies about psychological warfare teams at market makers.1 Reddit, Twitter, and chatrooms naturally select for virality more than validity; a thesis works if it's upvotable and repeatable. In fact, this dynamic puts a limit on just how much research you can share—the longer it takes someone to read a post, the further down the "new" page it drops before they upvote it, so shorter posts win. Newsletters have some selection for virality, but the inbox is sacred; if you waste people's time, or abuse their trust, unsubscribing is easy.
They are both phenomena that reflect cheaper and faster information flows, and lower transaction costs—courtesy of Robinhood's zero-fee structure for retail investors, and thanks to high-frequency trading for institutional investors. These are a tax on people who overtrade (paying 1/10th as much in total transaction costs and trading 100x as much without a profitable strategy is not a great deal), but they're a subsidy for investors who are right more often than they're wrong. Meanwhile, traditional financial research has been squeezed by fee pressure, and that has been especially visible for research coverage of small companies.
The default economic model for providing financial research is to cover a list of companies and continuously provide estimates, valuations, and justifications for both. This model works reasonably well for companies where there's always something to say; covering a company like Apple, Facebook, or Taiwan Semiconductor Manufacturing can easily be a full-time job. But for smaller companies, it's hard to financially justify any coverage with that setup, and the usual model doesn't support writing about a company once a year, or rotating through a list of interesting companies and then dropping them when they get boring. Covering such companies can work with a solo operation that doesn't have overhead, and good coverage is viral within the small community that cares about a particular mid- or small-cap stock.
(This community can include investors whose day job has them focus on big companies, but who invest their personal funds in smaller ones in the same industry. Being able to do this is a sort of fringe benefit for working on the buy side: you generate higher dollar returns by figuring out big companies, and higher percentage returns when you look at smaller ones. Since the big funds are not dollar-constrained, but employees, relatively speaking, are, it's a good division of labor.)
There is a market, of sorts, for research into penny stocks, but that market is much more geared towards manipulating prices and fanning speculative enthusiasm. It has existed for a long time, but is closer to gambling than investing; Matt Levine has argued, persuasively, that the readers of penny stock newsletters are in on the joke. They know the stock they're buying probably hasn't cured cancer or invented a Tesla-killer, but they might make money trading it against someone more credulous, or someone who opened their email too slowly.
Like many markets, the market for investing ideas has a sort of on-exchange spot market as well as an OTC market that's an order of magnitude bigger. You can source investment ideas from reading or watching the news.2 And of course you can read sell-side research, although plenty of other people are reading it. But the best ideas, of course, come from doing tons of independent research. And those ideas get exchanged, through formal idea dinners arranged by banks, and informal ones that consist of a few people getting together and talking shop. There are plenty of investors out there whose portfolio is about 25% original ideas and 75% borrowed ideas—all paid for by swapping other original ideas. This market—also like an over-the-counter market—is rife with conflicts of interest: everyone is pitching trades they've already made. This sounds like a recipe for disaster, but it usually works out, for two reasons. First, it would be very hard to come up with a pitch so compelling that it could move a stock straight to its price target with just a few conversations, so most of the value is in someone owing you a favor, not in the price impact of trades they make. And second, people are acutely sensitive to this risk, perhaps over-sensitive, and tend to spot it right away.3 (One exception to this: any time price movement is a fundamental catalyst. A currency peg, or a stablecoin, could in principle be destroyed by a single idea dinner.)
There are two economic models that drive subscribing to a newsletter, and make it a bit like picking a stock. If it’s undeservedly unpopular, you get an edge by reading it. If it’s too popular, you want to read it and bet against it; you’re peeking at your opponent’s hand. There are institutional investors who subscribe to research they know is wildly overrated specifically so they know the thought process of whoever is on the opposite side of the trade, and this model can work with newsletters, too. Ironically, this means that the worst position is to be good but fairly-rated—but there can be value there, too. Part of the analyst's role is to provide good frameworks, not just good estimates; there's value in understanding a company well even if it doesn't present an immediate trade. (Especially since institutional investors have access to lots of real-time datasets that other investors can’t afford or don’t know about; knowing that a company's strategy hinges on a particular new launch is very valuable to a fund with good data on how that launch is doing.)
If more small-cap analysts are writing on Substack instead of sharing their research informally or enduring the herculean ordeal of convincing a portfolio manager to at least consider an illiquid position, will this create abuses? Definitely. It's inevitable that if you set up some way to exchange social capital for dollars, people who don't value social capital much will liquidate theirs as fast as they can. But right now the market for newsletters is inefficient enough that someone who could make money picking small-cap stocks, and pitching them in a newsletter, can probably make more money working at a big fund. The people who publish small-cap equity research today are the ones who really like it, so the default attitude should be trust. And modern newsletters have inherited the blogger tradition of linking back to sources and entertaining vigorous debates in the comments section. This will end up being a strong signal of quality: if you write a personal essay on your Substack, it's perfectly reasonable to remove comments so you don't have to deal with trolls, and of course you don't need to cite sources; if you write a careful analysis of the arbitrage opportunities entailed by a complicated restructuring, but don't link to any sources and refuse to accept comments, readers will assume it's garbage.
This is an exciting time, both for students of information economics and for those of us who delight in finding microcap stocks that are priced like they're flaming garbage but are, when you dig down deep enough, merely garbage. (Or for anyone brave enough to short frauds, meme stocks, companies that trade as if every Covid trend will last indefinitely, etc.) Small companies are often small for a reason, but on occasion there are hidden gems. And now there's a better economic model for the people who find them.
A Word From Our Sponsors
Here's a dirty secret: part of equity research consists of being one of the world's best-paid data-entry professionals. It's a pain—and a rite of passage—to build a financial model by painstakingly transcribing information from 10-Qs, 10-Ks, presentations, and transcripts. Or, at least, it was: Daloopa uses machine learning and human validation to automatically parse financial statements and other disclosures, creating a continuously-updated, detailed, and accurate model.
If you've ever fired up Excel at 8pm and realized you'll be doing ctrl-c alt-tab alt-e-es-v until well past midnight, you owe it to yourself to check this out.
"Real Men Have Fabs"
When the semiconductor industry first started splitting into companies that fabricated chips and companies that designed them but didn't own their own manufacturing, the outspoken CEO of AMD responded to the trend by saying "Real men have fabs." (In Fabless, an oral history of this transition, the line is quoted six separate times—the only thing worse than not being quotable is being very quotable and very wrong!) AMD ended up spinning off its fab operation as GlobalFoundries, and Intel is now in talks to acquire them for $30bn ($, WSJ). The long-term trend in chip fabrication has been towards greater and greater scale, at the level of manufacturing plants and companies. This represents a bet by Intel that the company-level scaling is important, and that they'll earn more from size than they'll risk from the inherent conflict of making their own chips and also making other companies'.
I've written before about Amazon merchant rollups, and the financial theory behind them: diversification is a financial free lunch, so all else being equal a merchant is worth more to a diversified company than it is as a standalone business. Openstore has just raised $30m to run a similar rollup strategy for Shopify stores, with a difference: "At some point, OpenStore plans to launch a consumer-facing product featuring all of its Shopify merchants." For an Amazon rollup, this wouldn't make much sense, since Amazon has so many locked in customers. But Shopify has focused on building tools rather than controlling distribution; a company that uses the tools and gets good at distribution can improve the economics significantly. This ends up looking like a benign version of Amazon's white-labeling strategy: instead of creating a competing in-house product, Openstore buys the seller.
A few months ago I wrote about how Valve Software had started with games and backed into game distribution instead. Yesterday, Valve announced a handheld gaming device, priced between $399 and $649, running a variant of Linux but with libraries allowing it to run Windows-compatible games. Software companies rarely get into the hardware business to profit from it directly, but proprietary hardware can be very complementary to a software business. For Valve, one way to look at the goal is that maximizing available gaming time per day improves their value prop—and increasing the pace at which players finish games is a good way to continue selling new ones. Another way to look at it is that this product is backwards-compatible with existing games, but future games may be designed specifically for this form factor and controller, so Valve's hardware can make their open software platform a bit more closed.
A few years ago, the stereotype about venture capital was that it was a way to pay 2 & 20 for a Nasdaq 100 index fund with a ten-year lockup. Unfair, but, at least in terms of median performance, arguably true. Returns have improved substantially since then; the FT says that venture funds started between 2009 and 2017 beat the S&P by 10 points annually ($, FT). It's still true that performance is skewed to a small number of big winners, but as the total sample size gets bigger, the number of big winners expands, too. And now that there's a whole vocabulary to describe all sorts of different rounds a company can raise ("pre-seed" was probably the point at which round designations stopped meaning anything), it's more plausible for any given fund to have some kind of participation in the biggest winners. If there's more room to produce unicorns, particularly in enterprise software where there's a long tail of tasks that are a) important for companies, and b) currently very labor-intensive for particularly pricey varieties of labor, this changes the entire payoff structure. When only a handful of companies contribute to good returns for VC, fear of missing out is actually the healthiest emotion a venture investor can experience. But when there are lots of companies that can potentially return a fund, it makes more sense to be picky.
Chips and China
Taiwan Semiconductor Manufacturing Company is arguably the greatest force for world peace in existence; it provides essential inputs to China's most important exports, and is, from a CCP perspective, inconveniently located in land that China claims but cannot govern. But this also means, annoyingly, that certain kinds of catch-up growth are worrisome for the world since they reduce the cost of conflict between China and Taiwan; China's chip industry set production records in June, up 44% since last year. The country is still behind the cutting edge; China expects high-volume 14nm chips by the end of next year, while Samsung and Apple were selling devices with 14nm chips in late 2014 and early 2015. But it's still worth watching; catch-up growth happens faster than Moore's Law. And supply chain ties can ensure peace: part of the argument for the EU's ancestor, the European Coal and Steel Community, was that supply chains criss-crossing Germany and France would make war between the two countries impossible. And the drop in war frequency in Europe since then has been unprecedented, although there are other explanatory factors. Still, it's an idea worth taking seriously: mutual economic dependency constrains countries, which is another way of saying that every time a country finds a new way to be self-sufficient, it has new options to choose its own destiny.
The Bank of Japan plans to encourage banks to lend to climate projects by offering them 0% interest loans and reducing the share of their central bank deposits with negative rates ($, FT). The paradox of near-zero interest rates is that in financial terms, they make long-term, high-certainty investment projects close to free—but they can only exist when investors don't want to pursue long-term investments and are willing to earn zero or negative returns instead. Tweaking incentives for short-term interest rates can have an effect, but actions that influence expected returns and risk have a bigger one.
To be clear, this is not specific to retail investors: plenty of institutional investors develop a conspiratorial mindset. It seems especially common among short sellers, which makes sense, because shorting a fraud means developing an elaborate conspiracy theory and then finding out it's true. But there's a difference between a theory that fits facts better than the conventional story and a totally unnecessary theory for why a movie theater chain trading at 70x sales might decline in price from time to time.
I don't know of anyone who's put together a good long-term track record based on watching CNBC, but CNBC-on-mute is still surprisingly common on trading floors.
I once heard an analyst say he'd just gotten back from an idea dinner: "The guy from [Generic Geographic Feature] Capital was pitching XYZ, he says it's a double from here." The trader he was talking to scoffed, pointing out that [Generic Geographic Feature] Capital had owned XYZ for the past year and it was down 30% over that time. Even if someone thought the pitch was good, it burned social capital.