The Diff - On Currencies

In this issue:

• On Currencies—Currencies have powerful network effects, but despite this there's a strong case for many countries to have their own rather than using somebody else's. This case is so strong that private companies do it, too; a loyalty program or an in-game currency is just a special case of a command economy with strong capital controls (and shares some of the problems with that system, too).
• Taiwan and the US—Building more advanced chips outside Taiwan.
• Massaging the Numbers—China's reduced economic transparency.
• First Loss—It's hard to optimally capitalize a bank.
• Pods—A stripped-down hedge fund model doesn't quite work.
• Fake Offers—Bluffing is expensive.

This issue of The Diff is brought to you by our sponsors, Brex.

The Diff September 9th 2024 12:48 • Click to play audio

On Currencies

When a product has network effects, there's usually one big winner, whether that's in office software in the 90s, search in the early 2000s, or social a bit later. Winning isn't inevitable; AIM and MSN Messenger were dueling for supremacy in the 90s but are irrelevant today, and sometimes the relevant network flips—operating systems matter less than they used to when most products are accessible through browsers and many desktop versions are just a mini-browser with the app wrapped around it ($, Diff), but that's generally the way to bet.

Currency is a special case. Money could be defined as a product that consists almost entirely of network effects—you need at least two parties to agree that something constitutes money in order to transact, and their willingness to agree is a function of how many other people view something as money. The biggest currencies have powerful network effects, in both trade and finance—when two countries that don't have widely-used currencies transact, they tend to denominate their trade in dollars, because they'll both be able to spend them. Since a large number of investors denominate their returns in dollars, it's easier to sell dollar-denominated assets.^[1]

The historical reason countries issued their own currencies, rather than defaulting to some universal one, was mostly a functionof technological limitations: if units of currency are moved around physically, and the fastest means of transportation is a ship or a horse, they're going to be produced locally. And state capacity was limited enough that countries mostly couldn't issue fiat currencies, and were instead minting precious metals into coins. Take away those limitations, and there's a case for everyone to just use the dollar, or to invent some other currency as the main one. But this isn't what most countries do (there are some who peg their currency to the dollar, either oil exporters who try to reduce the damaging effects of a rising currency by accumulating savings or, historically, developing countries that didn't have the credibility to maintain the value of their currency, but that also wanted either the cachet of issuing a currency with their name on it or the option to devalue it if they couldn't maintain the currency peg). And yet, plenty of other places issue their own currency. They have good reasons to do this: using the US dollar means having everything benchmarked against US interest rates, and those rates are set based on US interests. It's inconvenient to have locals struggle to pay floating-rate debt because US central bankers are worried about inflation, or to experience an inflationary episode locally because the US is running big deficits.

If you broaden the definition of currency even more, the world abounds with micro-currencies: loyalty points, in-game currencies, cloud computing credits, API call limits—a wide variety of companies try to convert government-issued currencies into a store of value that they control, and then use a sort of corporate monetary policy to maximize their value.

These currencies fit a model that was more common in the mid-twentieth century, and that persists in a few places today: a currency that can be converted into other currencies only with the permission of the issuer. Some countries use currency controls as a way to manipulate their currency, either by intervening directly or by encouraging other entities to do so. (For a fun example of this, here's a paper on how Taiwan probably keeps its currency artificially cheap by encouraging life insurers to buy dollar-denominated assets.) It's a way to reduce the impact of speculation on a country's economy, which is particularly important for small, open economies where tiny changes in global portfolio allocations can create massive inflows. And it gives governments a measure of control over what gets purchased: Nigeria, for example, has a different exchange rate specifically for pilgrimages to Mecca. Limiting convertibility can be a way for countries to engage in protectionism without technically violating tariff rules; during its growth era, Japan had a reputation for making it very easy to source dollars to buy capital equipment, and much harder to get dollars in order to buy consumption goods.

When private companies do this, they have far more control, and that raises the question of why they'd bother with a parallel currency at all. Delta and United could just track dollars instead of miles, Marriott could talk about the dollar value of points, Roblox purchases could be denominated in sawbucks rather than Robux, etc. Some of them do this, as with cloud computing credits.

One reason to denominate them differently is that people think about them differently. There's a form of behavioral double-dipping where, if you're getting x% back on purchases, you feel like you saved x% at the point of purchase and then, when you redeem them, you feel like you got something for free once again.^[2] In casual gaming, there's a concerted effort to make the mental math of converting real money into in-game currency as mentally-taxing as possible—in Clash of Clans's in-game currency, gems, spending $10 gets you a 20% better deal than spending $5, but spending $20 or $50 at a clip only improves your gems-per-dollar ratio by 4% each step, but going for $100 is an 8% improvement. Presumably, the actual behavior the game has identified is that some players will switch from spending $1 at a time to $5 or $10, but past a certain point they just buy in the largest increment offered.^[3]

For most of these games, currency conversion is a one-way process: turn dollars into gems, turn gems into fun, and that's it. But in some cases, there's a two-way market: Roblox users can convert dollars into Robux, and Robux back into dollars. They can actually use different exchange rates similar to the way Nigeria does, to encourage some activities and curtail others. So Roblox offers a more generous revenue split to creators of premium games, when they're paid for with US dollars rather than in-game currency and bought on desktop. Capital controls exist to be loosened in cases when the capital movement is the kind that doesn't need to be controlled.

This also applies quite well to travel loyalty programs. The flights and hotels where points get redeemed are not the same ones where they're earned; a lot of New York to Chicago travel generates miles that are redeemed for flights to somewhere more exotic, and this introduces odd complementarities—a hotel chain with a big presence in markets that skew to business travel will tend to get higher returns than average from adding a new property in Hawaii.

Loyalty points have another perk: if a credit card company or airline unilaterally announced that they were bailing in customers' accounts, and would be cutting everyone's cash balance by 10%, their customers would be outrageed. But if they change the exchange rate for points, it's a different kind of annoying—customers will grudgingly acknowledge that the value of Marriott Bonvoy points is, in fact, set by the Marriott corporation, and that they're not a guarantee of any set kind of spending power.

While companies try to control the exchange rate of these points, mostly by constraining exchanges, the more of an advantage that gives the company the more of an opportunity it creates for black market operators. This market is opaque by design, and it's risky—their terms of service generally give them wide lattitude to zero out someone's account if they believe it's being used for this, and given how rewarding careful data analysis is to the design of these programs, they're reasonably well-equipped to spot anomalous patterns. Nonetheless, if there's some currency that leads to cheap flights and hotels, there will be demand for that currency.

Corporate quasi-currencies basically let a company opt some of its customers into a miniature command economy. Like a regular command economy, the part that's hardest to dictate is consumer demand, but, unlike a classic command economy, the people in command have enough information to track and exploit shifts in consumer demand.

These currencies also help companies avoid one of the paradoxes of currencies' network effects: the dollar is a poweful asset of the US, but it's not exactly a tool that can be used. The more popular and widely-accepted a currency is, the less control that issuer has over how it's used—a small country's central bank can more or less unilaterally decide to hike rates enough to halt inflation, or to cut them until there's an economic rebound. But the dollar is run on something closer to a consensus-based model than a top-down one: since it's used by so many countries, it's a reflection of their policies as well as the US's. More closed economies don't have this issue, and the popularity of this model among private businesses is a decent demonstration that it's possible to manage a currency in useful ways that produce profits for the people doing the managing and are valued by that currency's users. But it's also hard to scale—as many countries have discovered, capital controls can work for a while, but any sufficiently popular currency ends up beyond the full control of its issuer.

A Word From Our Sponsors

Banking that takes every dollar further.

Runway is life for startups, so why do most banks chip away at it with fees, minimums, and delays? Founders deserve better — that’s why Brex built a banking solution that helps startups take every dollar further. With Brex, you get the best of checking, treasury, and FDIC protection in one account. Send and receive money fast worldwide. Earn industry-leading yield from your first dollar — while being able to access your funds anytime. And protect your cash with up to $6M in FDIC coverage through program banks. Ready to join the 1 in 3 US startups using Brex and take every dollar further?

Elsewhere

Taiwan and the US

TSMC's Arizona plant has early yields similar to what they've achieved in the US. Chip fabrication is one of those industries that looks like it should be a commodity—buy some fancy capital equipment, feed in raw materials, get out a mass-produced product—and turns out to require ridiculous amounts of tacit knowledge to the point that Intel's mantra for a while was Copy Exactly! First principles are well and good in designing, but when it comes to making particular atoms behave in the right ways, apparently experience trumps all. TSMC has either been able to sort their workers into the most experience-critical tasks or to get US workers up to speed on the way they operate. Either way, it's a promising sign; being able to build advanced chips in more jurisdictions makes the world a more stable place.

Massaging the Numbers

A bit of inflationary policy, for a short period, is almost guaranteed to lead to a brief improvement in infalation-adjusted outcomes—we're all wired to view a 5% raise as evidence that we're working hard and a 5% increase in the cost of groceries as evidence of price gouging, so the average response to inflation is to spend a bit more and save a bit less, which contributes to growth. The costs show up later.

This kind of model also applies to reducing the availability of economic data: in the short term, it means hiding some problematic bad news, but in the long run it leads to the assumption that all the news is bad. This is a cycle China is goign through right now, as the country reduces the availability of macro data and censors some more downbeat commentary on it ($, Economist). On any given day, a lack of gloomy op-eds about the economy is good for consumer and investor sentiment. But after a while, it gives people the impression that the economy is bad, and that when nobody is allowed to talk about it nobody will be allowed to fix it.

First Loss

In other China news, Chinese banks issue "loss-absorbing" bonds, which are designed to be written off in the event that banks need more capital—but rating agencies assume that the banks will get financial support from the government rather than actually zero these bondholders ($, FT). The trouble with setting up capital structures for banks is that 1) you want them to have a large buffer between how much capital they absolutely need and how much they have or have access to, and 2) anything that involves using that buffer makes the bank in question, and the banking system as a whole, look less trustworthy. It's easy to say that banks should have a special kind of bond that they can write down to zero to preserve capital, but actually writing it down is an admission that the banks need to preserve capital, which is never what depositors want to hear.

Pods

An attempt to build a multi-manager fund without the portfolio managers has shut down due to the difficulty of fundraising. The original pitch was compelling: the usual structure of a multi-manager fund is that different teams pick stocks (or other assets) while managing a portfolio that nets out risk exposures. But that netting-out is a very quant-flavored process that's basically trying to get the best absolute return possible given a set of fundamental theses. It was tempting to run this model more cheaply by replacing the pricey discretionary managers with algorithms, while the analysts kept doing what they'd always been doing. This seems like an idea that should work, or is at least worthy trying, but it raises the question of why other funds are hiring so many portfolio managers in the first place. Those funds already run alpha capture ($, Diff) strategies, where they use systematic signals to enhance the returns from individual trades. If they're already offering this kind of return, and getting positive returns from their traditional structure, too, it's hard to compete by unbundling.

Fake Offers

In the early days of the US Steel / Nippon Steel drama, one of the bit players was Esmark, an obscure privately-held steel company which made an offer last summer and then nixed it the week after. They've now been fined for making an offer and falsely claiming that they had the cash on hand to fund it. It was somewhat implausible that Esmark happened to have ten billion dollars sitting around, and their offer would have made a similar splash had they said that they didn't have funding yet but would be looking into raising it. It's hard to understand exactly what they were going for at this point, other than making it slightly more expensive for a competitor to do a deal that they didn't want to do themselves.

Diff Jobs

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

• A fast growing start-up enabling SMB focused software businesses to offer accounting functionality to customers with an API call is hiring a founding full-stack engineer. (SF)
• A new data startup that's seeing early traction bringing transparency to a $1.5tr asset class is looking for a frontend-focused generalist engineer. 1-3 years of experience, or compelling internships and side projects. (Hybrid, West Palm Beach/Miami, FL).
• Growing team at a multi-strategy firm is looking for senior quantitative researchers with direct experience or exposure to a relevant field (statistics, optimization, ML) 250k-400k base + bonus (NYC)
• Growing edtech company with product/market fit is looking for an evangelist who is passionate about gifted and talented education and can tell data-driven stories online. (Remote)
• A tech-bio startup creating the world’s largest database of protein-molecule interactions is looking for a machine learning engineer with experience training LLMs on large scale distributed systems (Remote)

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.