Building Parallel Financial Systems

Building Parallel Financial Systems

One of the stranger events of the Cold War was that time the USSR helped to bootstrap a free market enabling speculators around the world to make bigger levered bets without being held back by government regulation. The postwar economic system rested on the Bretton Woods Agreements (previously written up on this newly-unlocked Diff post from last year). At a high level, the system fixed most countries' exchange rates against the dollar, and fixed the price of the dollar against gold. As any student of economic geometry knows, fixed exchange rates require participants to either give up their ability to set monetary policy or restrict the flow of currency, and the Bretton Woods system largely stuck with restricting flows.

It wasn't perfect. US banks happily complied with government regulations covering US dollars, French banks followed the rules for francs, British banks for pounds, etc. But some dollars found their way into non-US parts of the financial system. This presented local banking regulators with a dilemma of their own:

1. They could play nicely with the US, and treat those dollars as if they were covered by the American financial system.
2. They could let their banks have a little fun, using US dollars (the world's most popular currency) for pretty much anything the US government wouldn't allow other financial institutions to do.

Shading their view of option #1 was the fact that the US in the 50s and 60s alternated between needing to constrain its behavior to maintain the value of the dollar—part of the tradeoff between US ground troops and nuclear weapons in Germany was that soldiers leak dollars—and using the dollar's dominance as a tool to enforce policy demands. The Suez Crisis of 1956 was partly resolved when the US refused to bail Britain out of a balance of payments crisis unless Britain withdrew its military from Egypt. (This paper is a good and detailed analysis.) Affecting option #2 was the fact that the US economy's postwar boom kicked in faster than Europe's, and the European banking sector took an especially long time to recover. Decolonization meant that banks that used to finance trade deals around the globe were now stuck in their home markets, with fewer captive customers and less activity from their remaining customers. Letting European banks play around with US dollars was a shot in the arm for a long-suffering domestic industry. These banks wouldn't be competing on a level playing field—they'd actually have an advantage. US banks were limited in how much interest they could pay on deposits. In the offshore dollar system, they could pay whatever they wanted.

One of the large dollar depositors in European banking systems was none other than the USSR. The Soviet Union needed overseas banks to expedite trade; it needed dollars to buy imports (rubles were not especially useful even within the USSR, and worthless except for novelty value everywhere else). For a while, Soviet Russia had deposits at American banks, but they repatriated this money to Europe so their assets wouldn't get frozen. And then those dollars started floating around Europe, moving to whatever the highest-yielding opportunities were.1 The dollar system turned into two parallel dollar systems, one designed entirely around preserving a global system of fixed exchange rates, and one—much more dynamic, with a strange mix of participants2—that was creating hyper-capitalism in otherwise socialist-leaning Europe.

The Eurodollar story is a strange one, but it's a useful way to frame some modern innovations. For many companies, the single best way to raise capital is to sell stock on an American exchange. America has the world's largest equity market, and best-capitalized retail and institutional investors. A company in a strategic industry with state support in China might be able to access more money, but it comes at the cost of being under the direct supervision of the state. But one reason American capital markets are so trusted is that they're so regulated; retail investors are limited in what they can do, particularly when it comes to highly levered bets (on things other than houses and college degrees). Tapping in to the US capital markets means following some very strict rules in exchange for getting more funding on better terms than just about anywhere else. It's a good deal for mature companies, and mature industries, but for businesses that thrive on agility, US capital markets are just too slow.

FTX is an example of what can be built outside the US financial system. If you visit FTX.com and you're in the US, one of the first things you're greeted with is a nice message in all caps telling you that under no circumstances should you even try to sign up for an account and trade there.

Which is really too bad. FTX has crypto derivatives; it has levered inverse tokens, so instead of shorting Bitcoin futures, or borrowing Bitcoin and managing your margin and funding, you can just buy a token that does the opposite of what Bitcoin does, or the-opposite-times-three. (BEAR, the 3x inverse Bitcoin token, originally traded at $25 in August of 2019 and now changes hands for 0.00091.) It has a profusion of other crypto derivative assets.

More interestingly, FTX builds crypto derivatives on top of real world assets, allowing users to do things like trade futures on the shares of pre-public Coinbase and Airbnb. If you were in a big hurry to get a position in Coinbase, you could have bought it—well, bet on it—at $295 in December. When lumber futures were getting exciting, FTX launched their own version with about two hours of work.

The lumber futures product is worth digging into: it's not the same thing as the Random Length Lumber contract traded on the CME. The CME's contract is a contract for delivery of 260 cubic meters of wood; FTX's product is a futures contract that settles based on the producer price index of lumber and wood products. These certainly correlate, but they're not the same thing.

And this gets at a deeper distinction.

Cryptocurrencies are a unique asset class because they're not tied to real-world assets except through not-100%-crypto mediators. The Bitcoin blockchain can be fully distributed because anyone can run a hashing function and confirm that a block is valid. And the blockchain has no native concept of what transactions mean. A Bitcoin transfer can be the result of a speculative purchase of BTC, a purchase of physical goods, or just someone shifting money from one wallet to another; the protocol doesn't, and can't, know. This is unique. Every other financial entity in the world has some kind of preceding referent, some sort of preexisting legal structure and real-world phenomenon it’s tied to. An oil futures contract is based on a certain amount of a certain kind of oil (and implicitly assumes an infrastructure for transporting it, grading it, and storing it). When Alfred Winslow Jones started his long/short hedge fund, he was doing something new, but he was using existing legal infrastructure (a limited partnership) and existing financial infrastructure (brokers, stocks). This means that every other asset is born fitting into some legal structure, while crypto was born in an extralegal gray area; only the ways people used it gave it some kind of legal status. In that sense, crypto is analogous to something like a thought, or a mathematical formula. Thoughts can’t be banned or even regulated, but as soon as they lead to real-world actions, those can be regulated. And a formula can’t be banned, though sharing it physically or digitally can be restricted.

The fact that these assets never refer to anything unless a counterparty makes it so gives them incredible flexibility, and completely alters the mindset of the people who trade them and build financial services on them. A crypto product's value is entirely based on the consensus of the outside world. This makes it distinct from other products. As FTX's founder points out in an interview, there are real-world bounds for other asset prices: if Apple gets too cheap, Berkshire Hathaway will just buy the rest of it; if lumber futures get too expensive, people will build sawmills. But DOGE can't be overpriced or underpriced. It's just priced. These detached asset values mean that derivatives producers aren't thinking about hedging real-world risks, but thinking about efficient ways to create more interesting bets.

A scalable source of interesting bets is a nightmare to any financial regulator, with perhaps one exception: the sort of implicit financial regulator who takes a fee on every transaction. (In FTX's case: 2 basis points for price makers and 7 bps for price takers, but more on that in a moment.) An exchange that is less constrained by legal rules is still constrained by economic rules. Here are two that FTX has to follow:

1. An exchange whose customers are all broke will not do much business.
2. An exchange can't do very much business with sophisticated customers if those customers think the exchange has a chance of going broke.

FTX deals with these risks in a few ways. The company was founded by the operators of Alameda Research, a large crypto market-maker. This gave the exchange capital, an initial network effect, and a keen sense of how smart traders think about crypto counterparty risk. A crypto market-maker is partly in the business of actual market-making, and partly in the business of taking credit risk—of knowing that a price aberration on one exchange reflects an actual market inefficiency, rather than the market's accurate view that getting money out of the exchange will be hard. But treating every superficial inefficiency as evidence of counterparty risk is also a bad idea; sometimes, markets are imperfect, and sometimes, counterparty risk can be measured. FTX itself is very open about how it handles liquidations, which are a major source of counterparty risk: if an exchange has customers who are insolvent, then the exchange has to do some combination of dipping into its own capital and taking capital from other participants, neither of which is ideal. Since Alameda can supply emergency liquidity, it can prevent some cascading margin calls (and since trading volume rises and market makers clean up during market drops, they're actually better positioned to do this exactly when they need to. The whole enterprise is one big hedge.)

Financial engineering doesn't just extend to what's traded on FTX and how it supplies liquidity. It also describes part of how the business is funded. FTX has its own token, with two interesting traits: first, people who hold the token in their accounts get a fee discount, and can use the token for collateral when they trade other assets on FTX. This a pretty straightforward financial transaction, in a way: spend money upfront, get money over time. But it ends up capitalizing FTX in the process, by pulling future commissions forward. In a way, the FTX token pays a variable dividend to holders by saving them money on commissions. FTX also commits to using a third of its commission revenue to buy back and destroy the tokens, and that makes FTX's token look a lot like a modern equity: a company that commits to using its free cash flow to buy back stock is continuously returning capital to shareholders and continuously supporting its own stock price.

All of this creates a virtuous cycle: if FTX's volumes rise, or people expect them to, the FTX token's "buyback" is worth more, so the price goes up. A higher price gives FTX holders more collateral, which they can use to trade more, which raises the "dividend" value of the token.

Exchanges rely on network effects, and it's hard to get a good network effect in finance while deliberately cutting off the world's biggest capital market. But it turns out that growth hacks and virality can apply within a financial system, too, and at some point having maximum flexibility with a more limited market is more valuable than having access to a large market where everything trends towards being commoditized or banned. The eurodollar market eventually grew enough that it bled into the regular dollar market; they weren't separate, but the US system had to converge with the more flexible and robust non-US dollar system. Something similar could easily happen in crypto: if all the offshore liquidity is on one platform, and that platform can launch new financial products faster than other exchanges, it will end up setting the rules for a market that exists because the usual rules don't apply.

Further reading: Moneyland is a good, albeit angry, look at offshore finance. Moneyland is the second book I've read that points out that the plot of the book Goldfinger was actually pretty well thought-out given the monetary situation at the time, and Auric's plan probably would have made him millions of dollars while crippling the British economy. "No, Mr. Bond. I expect you to face a crippling balance-of-payments crisis." This paper has background on the origins of the eurodollar market. In Supermoney, "Adam Smith" has a fun chapter about partnering with a Swiss bank in the 60s. It doesn't go great, although it’s fun for a while. Thanks to these folks for recommending good reads.

On FTX, this Odd Lots interview is great, and the FTX whitepaper is a nice overview of how they work.

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.

Elsewhere

Tethers and Dollars

Tether is ostensibly a stablecoin worth $1 per coin, and backed by $1 per coin. The questions of its backing and its worth have been contentious for years, which has led to various accusations and a lawsuit Tether settled for $18.5m along with the promise to report on what their reserves consisted of. The first of these reports has been shared with crypto news site The Block, and it reveals that—Tether is not backed 1:1 with dollars! 76% of its assets are "cash and cash equivalents," mostly commercial paper. Commercial paper is close enough to cash for most purposes, although strictly speaking Tether is trying to earn a tiny spread by taking a very tiny amount of liquidity risk. (If half of their total assets are in commercial paper that matures in a month, and they need to redeem more than half of their tokens for dollars within a month, they'll have to sell some, which might be trivial but would at a minimum lead to delays, making a Tether worth less than a dollar.)

But what's especially notable is that 12.6% of Tether's assets are in "secured loans," (counterparty unknown), 10% in bonds and precious metals, and 1.6% in "other investments (including digital tokens)." A portfolio that's mostly cash, but also includes these assets, is really not a cash equivalent, and is certainly not 1:1 backed with dollars. It is, in fact, a very low-leverage, but very unregulated, sort of bank: it's funded by demand deposits but invests its assets in some products that have longer maturity. Typically, this activity is closely regulated because it can easily blow up; it's a set of promises backed by approximate expectations. On the other hand, Tether is weirdly durable, and has survived much more damaging data releases than this.

And many other Tether theories don't make much sense in light of this backing report. One argument about Tether was that what had really happened was 1) Tether issued some tokens, 2) it invested the proceeds in crypto, 3) crypto went down, and 4) rather than admit more insolvency, Tether printed more Tethers to make crypto go back up, hoping that it could achieve solvency before it ran out of liquidity. The problem with that theory is that at some point in Bitcoin's long history of ups and downs, Tether really should have run out of backing. Bitcoin lost half of its value in two days last year, but Tether still trades at a dollar. Meanwhile, running a small-time, basis-point-shaving operation like calling something 1:1 backed with cash when it's really 0.75:1 backed with cash and 1:1 backed with a mix of assets just can't compare to the profits from essentially printing money to speculate in more volatile assets.

So this report doesn't resolve much, but makes every existing theory about Tether less plausible.

The Hotel Shortage

Hotel occupancy is down 17% from the same period in 2019, roughly in line with recent trends. One driver:

Demand was up week over week, but an increase in supply from both reopenings and new properties pulled national occupancy down. Major markets, such as New York City and San Francisco, are showing the most movement with properties coming back online.

In my piece yesterday on inflation ($), I noted that the travel sector was the biggest source of inflation in April's report, and that some of this was because hotels weren't resetting their wage expectations to the new market-clearing rate. We might see a transition in travel inflation over 2021: early in the year, higher prices because some hotels are still closed, or not operating at full capacity; later in the year, moderate inflation because hotels operating at full capacity will pay slightly more for labor, and will pass some of those costs on to customers.

The Credit Reopening

Banks have relaxed consumer lending standards in the last two quarters ($, WSJ), after tightening them early in the pandemic. One of the standout features of the current housing boom is that it's less borrowing-driven than previous ones; many of the people moving to new houses used to rent in expensive cities, and are buying bigger-but-cheaper houses elsewhere. For banks, the usual goal is to lend as generously as possible when the economy is about to grow—their marginal borrowers will get more creditworthy—and then to tighten up before a slowdown. In the aggregate, the opposite happens almost by definition: absent huge government intervention (of the kind we've seen in the last year) it's hard for high growth to happen without an expansion of credit, and that typically means lower lending standards. If banks are getting more willing to lend, it makes the current economic cycle look a little more normal: driven by the private sector seeing good opportunities to put money to work rather than by the public sector seeing a need for some way to stimulate demand.

Ransomware Acceleration

The big cybersecurity story of the last week was, of course, 1) a ransomware gang shutting down a major energy pipeline, partly by accident, and 2) that same gang promising to be more socially-responsible in the future. But it's hard to maintain ESG best practices in an industry where everyone is anonymous, so Ireland's healthcare system is mostly shut down due to a ransomware attack ($, FT).

I've argued before that cyber risk is following a similar curve to military technologies: at first, the new developments are subsidized by state actors (or, as in Russia, by somewhat independent actors who are constrained by the need to keep the government happy, and who are in return protected by the government since they help cultivate offensive capabilities). Over time, techniques that were pioneered to make strategically useful attacks get used in the private sector instead. One important question is: why have attacks accelerated so much recently? The general trends that make them more viable—cryptocurrency as a way to pay ransoms, more Internet-connected devices—have been happening gradually, but the attacks rose more suddenly. One possibility is that it's a bit of the Baader–Meinhof phenomenon on both sides: individual attacks are more newsworthy because the general idea of cybersecurity is in the news more, and each successful ransomware attack encourages more attackers to look for the next big vulnerability.

Churn, Margins, and Habits

Online grocers are recreating the old norm of scheduled deliveries instead of immediate, on-demand access. There are a few drivers here:

• Grocery purchases are more likely to be recurring.
• Scheduled delivery is much cheaper to arrange than on-demand, since the delivery company can figure out its routes and driver needs in advance.
• Reducing churn is the flipside of creating habits: if someone uses a product or service without thinking about it, they'll pay for it the same way. On-demand delivery encourages a sort of reactive habit formation, associating a need for some product with the use of a particular company to buy it. But active habit formation might be more durable; once people are planning around a service, the rest of their life makes that service a bit stickier.