Subscribe
Share
Share
Embed
The Diff is a newsletter exploring the technologies, companies, and trends that are making the future high-variance. Posts range from in-depth company profiles, applied financial theory, strategy breakdowns and macroeconomics.
##
Historically, the hierarchy of important investor constituencies looked like this:
1. Large, long-term, long-only investors—a group that includes the big fund complexes like Fidelity and Capital Group, but also founders and their descendants, evergreen venture capital funds, strategic part-owners, and activist investors.
2. Shorter-term long-short funds. If they're only going to hold a position for a month or two on average, they're just a very inefficient way to turn persuasion into many share-years of ownership relative to someone who'd buy and hold for longer.
3. Retail investors are hard to reach, and they add overhead because of how much Broadridge charges for mailing out investor materials. This is costly enough that there's a high-IRR but very hard-to-scale trade that takes advantage of companies buying out small investors' stakes at a premium!, so it's not quite risk-free.)
That historical role makes sense. It's just another case of economies of scale. If you're setting up a group meeting at a conference, and you can choose between someone who already owns 5m shares and might buy a few million more, someone else who's periodically long or short half a million, and a third person who might buy 25 shares in their Robinhood account, the decision is obvious. Or at least, it was obvious. But retail investors are pushing stocks around again, and they’re either doing well enough on some meme names or just committed enough to gambling that they have serious purchasing power in the aggregate.
And there's some helpful synergy between market populism and the political kind. Online retail investor communities lean right, in comparison to many other online groups that are default center-left or so. Meanwhile, regulatory outcomes for companies are a) a lot less certain than they used to be, and b) a lot more tied to whether or not Donald Trump feels personally flattered by the CEO's behavior, and to a lesser extent the company's actual decisions. The high-stakes race to build artificial superintelligence is one thing, but the guest list at Mar-a-Lago is sometimes a bigger deal.
This creates some synergy. A combination of investment commitment, photo opp, and gift of an object made partly out of gold is good politics, for the moment, but also good investor relations. And for companies that want the financing to take advantage of whatever new moneymaking opportunities their White House visits produce, getting a new base of investors is very helpful—retail can buy the additional supply from a secondary offering, and retail investors often introduce enough volatility that a convertible bond is a good idea, too.
This isn't a permanent end state for the market, because the presence of so many retail investors creates attractive prices that will eventually mean-revert, and also leads to a growing share of the market consisting of companies that are explicitly building their strategy around attracting retail interest. The most obvious of these is the crypto treasury strategy, which is really showing its age. What's interesting about this strategy is that it was plausibly a way to attract institutional flows when it started: there were lots of people who had a mandate to buy software stocks, or who could only invest in US-listed equities, but who really wanted to take a flyer on Bitcoin—or who felt that a quasi-currency with a fixed inflation rate would do well in a negative-real-rates environment. So, they might be willing to pay a premium to own the stock. Now, that's not at all the argument, nor is it the target audience for the pitch. Crypto treasury strategies have started to go through their own version of memecoin supply tsunamis, where people's willingness to bet on one theme create an endless supply of riffs on that theme (if you don't like Dogecoin, maybe you'll try Shiba Inu, or dogwifhat‐which actually exists, and has an $875 million market cap.)
Retail investors tend to be less valuation-sensitive than other market participants—though more so than index funds, which helpfully demonstrates that you don't strictly have to care about value to outperform the majority of investment professionals. And that means that when they get attracted to the market, the market's going to meet them halfway by increasing the supply of the companies retail wants to own. Which, in a way, is a highly capitalistic implementation of democratic socialism: everyone voting on how companies behave (by giving them a low enough cost of capital to invest in otherwise infeasible projects), and with society tolerating lower aggregate returns on capital in exchange for a market that better expresses the will of the people.
## Elsewhere
### Credit Spreads
Normally, the way to think about the gap between corporate bond yields and treasury bond yields is that it's a measure of how nervous investors are about the private sector. But this time, US credit spreads are the tightest since 1998 because there's just a tiny bit more tail risk around the US dollar, but dollar weakness is good for companies that borrow in dollars. Investment-grade bond investors have to think about tail risks because the companies they're lending to are too stable for any other risk to be big enough to lead to a sustained loss. And, for a bit, there's a new tail risk that treasury holders need to be paid a few extra basis points to bear.
### Retail and Institutions
Millennium Management is a vast collection of mostly-independent hedge fund fiefdoms that produces steady returns from the aggregate output of lots of individually lumpy-return strategies. Goldman Sachs is part of a deal to buy 10-15% of it, and market the stake to individual investors who will pay management fees and carry in order to own it. It's an amazing instance of financial recursion, to pay a slight discount on very-active-management fees for the most sedate strategy imaginable, buying and holding a single asset. Millennium's model evolved to justify its fee structure, and that's created a business exciting enough that access to it deserves an aggressive fee structure, too.
### Scale
The biggest private equity firms are feeling more optimistic than their smaller peers, partly because they're involved in so many more layers of the capital structure. If there's action in credit, they're providing credit; if equity valuations are out of whack, they'll borrow and buy. So their real bet is on volatility over longer time scales: if we're in a period where there are lots of risks to worry about (trade, AI, a recession), then firms in the business of deciding which risk to take, and running diversified portfolios of those risks, will have an advantage.
### Non-Tariff Barriers
One of the holdups in US/EU trade negotiations is EU-wide rules governing platforms, which disproportionately impact US-based companies. For a long time, EU tech regulation has felt fairly zero-sum; a lot of it makes sense if there's a finite pie of economic value, which can be captured in the form of ad revenue, taxes, or user privacy. In that model, a shift towards stricter consumer privacy rules is just transferring some wealth from ad platforms back to their users. On the tax side, the EU is probably right, in that the tax rate that would actually cause a big platform to leave their market is quite high. Part of what holds them back is that for plenty of European exporters, the tariff that would make selling to the US untenable is also quite high, so if the US views EU rules as taking advantage of American businesses rather than taking advantage of tech businesses, the US will probably retaliate. (This is also a case where tech's Trumpy turn in late 2024 may well pay off.)
### Technology Adoption
The NYT has a fun piece on how hard it is to get CEOs to adopt AI tools. One of the problems is that LLM chatbots are little minions who do lots of the random tasks you need done but don't have time for, so the more senior someone is in an organization, the more they're already used to being able to do this. What they have to adapt to is that they can now get an answer that's probably less accurate, but much faster, and much cheaper. (And that they don't have to feel guilty about throwing the LLM a weird random research assignment at 9pm on a Friday.) LLMs are great for slowing age-driven executive turnover, because they're unusually good at things that get harder with age—remembering something on the tip of your tongue, or getting back context when stepping away from a task. But they'll also drive some executive turnover if too many people who could get the most out of them are stuck at a local optimum.