Welcome to the free weekly issue of Net Interest. For more insights – this week including a special on banking in the 1970s – and full access to the searchable archive, as well as frankly just to support my work, please consider signing up as a paid subscriber.

Over the past couple of years, like many others, I've watched as technologists have had a go at recreating the banking system. Smart people, they combined principles from economics with computer science to replicate ‘traditional finance’ structures in the crypto ecosystem. We’ve discussed some of their exploits here before: I went along for the ride in My Adventures in CryptoLand and, in Reinventing the Financial System, I explored some of the parallels.

The difficulty is that financial systems tend towards instability. That’s because finance is built on confidence, and confidence can be ephemeral. In traditional finance, a set of tools has accumulated over the years to support confidence, yet in the new ecosystem, similar tools can be lacking. That’s perhaps no surprise. Protections in traditional finance evolved over centuries of trial of error. Following the Napoleonic Wars in the second half of the 1810s, the UK economy was characterised by intense financial instability; it took until the Bank Charter Act of 1844 to instil some calm – and that was just the start. 

By contrast, the parallel crypto financial system has grown very rapidly. In many cases, its pioneers didn’t even realise they were navigating a path already trodden.

“In the very beginning of the project, I remember we didn’t even realise… that we were essentially just building a protocol that did the same things as fractional reserve banking, did something very similar to how a banking balance sheet works and we were just implementing that as a blockchain protocol. We thought we were doing something completely, totally different from how money usually worked in the traditional sense.” [source]

The failure to attach strong protections went on full display this past month when one crypto project, Terra, spectacularly exploded. The project was a novel stablecoin – a type of privately issued money that underpins the new financial architecture being built around crypto. Unlike other crypto assets, stablecoins are designed to be, well, stable. Most are linked to the US Dollar and are meant to track it one-for-one. For most of its (albeit short) life, 1 TerraUSD was worth 1 US Dollar… until one day, it wasn’t. Earlier this month, Terra’s tracking mechanism collapsed, taking down the value of a TerraUSD with it; today 1 TerraUSD is worth a measly 9 cents. Just prior to its collapse, the value of TerraUSD in circulation was $30 billion, most of it now gone. 

In the words of Mark Twain, “It ain’t what you don’t know that gets you into trouble. It’s what you know for sure that just ain’t so.”

So what went wrong? And what can two hundred years of monetary history teach us about financial stability? To answer that, we need to dig deeper into the mechanics of stablecoins.

Algorithmic Stablecoins

In the modern economy, most money takes the form of bank deposits. So to create a parallel financial system, you first need to create parallel bank deposits. Crypto entrepreneurs do this via stablecoins. 

There are various ways to tie the value of a stablecoin to that of a widely recognised asset like the US Dollar. One is to stash away equivalent US Dollars in reserve: for every new stablecoin that you issue, you have equivalent US Dollars in the back room to support it. Another is to stash away other assets, but take in more than you need so that there’s a buffer to cushion fluctuations between the value of that collateral and the US Dollar. So rather than having US Dollars in the back room in a 1:1 ratio, you have something else in a ratio 1.5:1, say. If the value of that collateral falls, you have time to call in some more and if it is not forthcoming, you simply redeem the stablecoin. 

Terra didn’t go for either of those approaches; it went for a third – market intervention. It did this using another coin it created – Luna – which conferred ownership rights over the system it was creating. If, due to rising demand, 1 TerraUSD rose in value to 1.01 USD, its protocols would allow Luna holders to swap 1 USD of Luna for 1 TerraUSD, handing them a 0.01 USD profit. If, due to falling demand, 1 TerraUSD fell in value to 0.99 USD, the protocols would allow TerraUSD holders to swap 1 TerraUSD for 1 USD worth of Luna. Given that you can buy 1 TerraUSD for 0.99 USD and exchange for 1 USD worth of Luna, you profit 0.01 USD.

If it sounds a bit complicated that’s because it is, and you were well placed to have stayed away. The alchemy works fine as long as Luna has some value. Such value can derive from transaction fees the system collects as it grows, from subsidies directed to Luna holders, or from aggressive marketing. Recognising that the whole edifice rests so squarely on Luna’s perceived value, Terra mobilised all three strategies, not least the third – the aggressive marketing. Its founder, Do Kwon, wasn’t shy on Twitter:

Nine months after that exchange, confidence in Luna evaporated and with it, Terra. The precise catalyst is inconsequential: some holders began to reassess the value of Luna, and it caught on. A doom loop ensued as the organisation was forced to issue more Luna to protect the value of TerraUSD even as the price of Luna fell. Frances Coppola was right: self-correcting mechanisms do not work when panicking humans are stampeding for the exit. 

Pegs rarely survive shaky foundations and the risk-adjusted rewards of betting against them can be vast given the asymmetric payoff. In the 1990s, George Soros made billions for investors betting against unstable currency pegs. The key uncertainty is timing. Soros’ success came as much from getting the timing and the sizing of his trades right as the direction. He placed his initial bet against sterling in August 1992 and ramped it up days into its peg breaking in September. One crypto fund manager said about the Terra situation, “This is why we didn’t start shorting this thing until actually pretty recently… I don’t want to say exactly when, but it was sometime this month and it wasn’t actually earlier than that, that we were short. And you know, I think that with something like this, being early is almost as bad as being wrong.” ¹

In a cruel twist of fate, one macro hedge fund manager who should have been more sceptical about the durability of asset pegs is Mike Novogratz. As founder and CEO of Galaxy, he was heavily invested in Terra: the firm was an early investor in Terra and he even had a Luna tattoo inked on his arm. Yet in January 2015, Novogratz was burned when the Swiss central bank abandoned the Swiss Franc’s peg to the Euro. His fund at investment group Fortress was eventually liquidated later that year. “The hedge fund business is Darwinian,” he said on an investor call. 

Crypto-Backed Stablecoins 

Before the era of central banking, private banks used to issue their own money. The period is not remembered fondly in the US, where policymakers have unfavourably compared today’s stablecoins with the “wildcat banking” that was eventually curtailed in 1863. “The period in the nineteenth century when there was active competition among issuers of private paper banknotes in the United States is now notorious for inefficiency, fraud, and instability in the payments system,” said Federal Reserve Governor Lael Brainard a year ago.

Elsewhere, however, the model was more successful. In Scotland, banks freely competed in note issuance. The notes were denominated in pounds but were the obligations of the individual banks, and they competed vigorously to try to get people to take their notes rather than those of a rival bank. ²

The Scottish model is most directly channelled in the crypto ecosystem via Maker, a crypto “bank” which issues the Dai stablecoin. We discussed its model in detail in Reinventing the Financial System but in summary, it works like this: An investor comes to Maker for a loan, armed with some collateral they are happy to keep locked in a vault – typically a crypto asset like Ethereum. For every $100 worth of crypto assets, Maker will lend $66, the gap necessary to provide a buffer of protection against a possible fall in the value of the asset. Maker accepts the collateral and advances a loan, which it does by issuing its Dai money.

Among crypto purists this is a popular model. It allows for the issuance of a stablecoin linked to the US Dollar without the need to tie up actual US Dollars. The problem is that it does tie up other collateral and this puts a brake on growth. Among non-purists – or at least those less wedded to the ideal of decentralisation – a problem in the model is its lack of regulation. Although it looks like a bank, Maker is not regulated as a bank. One of the protections that has evolved over the years in the traditional banking system is deposit insurance; without a bank charter, Maker doesn’t offer deposit insurance. The onus is therefore on depositors to understand the risk they are taking on as a creditor. 

Luckily, balance sheet comparisons are available (h/t @SebVentures). But just as few depositors have the patience to examine Bank of America’s balance sheet (bottom left) few want to look at a stablecoin issuer’s (although Terra’s – bottom right – is a sight to behold). An alternative, simpler structure, is to invest in a stablecoin backed entirely by fiat currency – one that looks more like a money market fund than a bank. 

Fiat-Backed Stablecoins

When the Scottish banking system was formally brought under the control of the Bank of England in 1845, Scottish banks retained the right to issue their own banknotes but they were required by law to set aside assets that are worth at least the value of all of their banknotes in circulation. To this day, the Bank of England employs a small team of staff within its Notes Directorate to monitor compliance. The team conducts physical checks of the assets ring-fenced to back the notes – which can include £1 million Bank of England notes known as Giants and £100 million Bank of England notes known as Titans – and analyses daily data reported by the authorised banks.

The model is close to a fiat-backed stablecoin, the largest in circulation being Tether. Again, there is no regulatory oversight here and so depositors need to do their own work to ascertain the value of their security but balance sheets within this category are typically easier to read. They are prepared not by a small team from the Bank of England, but by a team of accountants – and they typically come with a lag. Just this week, Tether disclosed its consolidated reserves report for March 2022.

As of that date, Tether had $82.2 billion digital dollars (UDST) in circulation. The organisation has been criticised in the past for holding too much risky commercial paper against those deposits. By March, it had taken down its commercial paper holdings to 24% from 31% of assets, although its holdings of money market funds – which may themselves have exposure to commercial paper – rose to 8% from 3%. The safest assets – US Treasury bills and cash – still represent only around a half of Tether’s assets (53% up from 49%) and more opaque assets, like secured loans, corporate bonds, funds, precious metals and “other investments (including digital tokens)” contribute around 14% of assets (down from 16%). 

If you’re unsure about the value of a financial firm’s assets, then equity provides a cushion for depositors. That’s one of the reasons why banks have such stringent equity capital requirements. But in Tether’s case, there’s only $162 million of equity backing the whole construct, equivalent to 0.20% of assets. So if the value of Tether’s asset portfolio dips more than 0.20%, there’s a solvency deficit in there that needs filling. 

It’s no surprise, then, that depositors are getting nervous (although it’s unclear why it’s taken so long). Tether has suffered over $9 billion of redemptions since last week when it briefly broke its peg, equivalent to over 10% of its issuance. And Curve’s 3pool, which acts as a kind of interdealer broker between stablecoins, signals continued redemption pressure. ³

The beneficiary in all this could be USDC, the stablecoin operated by Circle Internet Finance which has filed to go public via a SPAC merger. As other coins have lost share, it has gained share. Unlike Tether, it holds assets in cash and short term treasury bills. According to its attestation report, “the total fair value of US dollar denominated assets held in segregated accounts are at least equal to the USDC in Circulation at the Report Date.” Such attestation doesn’t come as regularly as TrueUSD’s, another stablecoin, which provides reports in real time. Nor is Circle as actively supervised as Paxos Trust or Gemini Trust, which have subjected themselves to oversight by the New York Department of Financial Services (NYDFS). But compared with those others it has scale, and in the networked money business, that is important – it is one of the reasons Tether flourished for so long despite its failings. 

One of the challenges all stablecoin issuers face is how to make money as private enterprises. Banks make money by taking an interest rate spread; money market funds via fees. An ongoing paradox in financial services is why so much investment capital is trained on an industry that just isn’t that profitable. US bank sector return on equity was in a range 5-10% for most of the 2010s and money market funds have had to waive fees in the low interest rate environment. But as rates rise, an opportunity emerges. 

Last year, Circle made $28 million in interest income on USDC. After income sharing with partners and transaction costs, its USDC income was only $16 million despite having issued $42 billion of the coin by year end. The reason was plainly low rates. Around a quarter of its USDC reserves were held in cash earning just 0.16% over the year, and three-quarters were held in US treasury bills, earning 0.05%. As part of its filing procedures, Circle has published its profit forecasts for coming years. The company may have anticipated higher issuance in 2022 than is now looking likely (its projections are for $110 billion in circulation by year end) but rates are also higher, so its target of $275 million of net USDC revenue may not be far off. 

This is a critical time for stablecoins in their challenge to become a sensible alternative mechanism for payments. The Terra implosion has left a residue over the entire space and they were anyway on a path towards tighter regulation. But as higher rates create a route to profitability, the low-risk survivors could be well placed.