Perhaps the most shocking thing about the collapse of terraUSD this month was how many people saw it coming. Most crypto investors blame the stablecoin’s failure on a targeted attack by speculators, akin to how George Soros ‘broke’ the pound in 1992. While the assault was a surprise, to many in the industry the collapse was inevitable.
TerraUSD, commonly referred to as UST, and its sister token luna were linked by an arbitrage mechanism designed to ensure UST would hold a steady price of one dollar. Between May 9 and May 13, though, UST plunged to less than 15 cents and luna to around a hundredth of a cent.
“It was probably the most successful speculative attack on a cryptocurrency ever,” says Yin Luo, vice-chairman of quantitative research, economics and portfolio strategy at Wolfe Research.
Crypto insiders estimate as much as $1 billion was deployed to destablise UST. Social media buzzed with rumours that Citadel and BlackRock were behind the attack, though both deny any involvement.
According to Igor Igamberdiev, a researcher at crypto industry publication The Block, proprietary trading firm Jump Trading spent nearly $700 million trying to defend UST, to no avail.
Luna Foundation Guard, a not-for-profit entity set up to support the decentralised ecosystem, blew through its $3 billion reserve fund in a desperate attempt to check the fall.
Yet multiple experts say UST’s collapse could have — indeed that it would have — happened anyway due to the construction of the terra ecosystem, which ensured that inflation in luna would at some point inevitably become overwhelming.
“It was a flawed design,” says Alexander Lipton, a global head of quantitative research and development at the Abu Dhabi Investment Authority and co-author of a well-regarded textbook on blockchain and distributed ledgers. “Algorithmic stabilisation is an impossibility and an oxymoron and people pursuing it need to improve their understanding of finance.”
Lipton, who says his views are his own, predicted in lectures back in 2018 that stablecoins such as UST would run into trouble.
If it was not a concerted attack that brought terra down, eventually something would have derailed the projectAnatoly Crachilov, Nickel Digital Asset Management
Terra’s stabilisation mechanism relied on perfect arbitrage. If UST fell below a dollar, traders were incentivised to buy the coins and swap them automatically on terra’s blockchain for a dollar’s worth of luna. They would then sell luna for dollars and buy more UST until the peg was restored. If UST’s price rose above a dollar, the process would work in reverse, with arbitragers buying a dollar’s worth of luna and swapping it for one UST, and so on.
But when UST ran into trouble, a flaw in the mechanics unleashed a feedback loop of colossal inflation. To fund the UST-luna swaps, the system automatically minted new luna and “burnt” UST. This caused luna’s price to fall so fast as to panic holders and shatter the arbitrage logic that was supposed to stand the system up.
This defect in UST’s design was well understood by more sophisticated players. “The collapse came as no surprise,” says Anatoly Crachilov, co-founder and CEO of Nickel Digital Asset Management, a $300 million hedge fund running arbitrage and multimanager strategies in cryptocurrencies.
“It was only a matter of time until market conditions were unfavourable enough for luna to slip into a death spiral caused by hyperinflation,” he says. “If it was not a concerted attack that brought terra down, eventually something would have derailed the project.”
Nickel sold out of its UST positions on January 27, more than two months before the crash.
There were other problems with the terra project. UST’s creators assumed markets were instantaneous, quants say, when in reality the arbitrage trade on which the system depended took hours to execute during the height of the panic. As it turned out, the holders of UST were overwhelmingly yield-chasing speculators who would quit the currency at the first sign of trouble.
Ultimately, UST and luna’s self-referencing mechanism was like a fantastical perpetual motion machine, Lipton says, dryly noting the French Academie des Sciences banned submissions from scientists laying claim to such machines back in the eighteenth century. The fact that UST was able to draw in $18 billion of capital at its peak, he says, is “mind boggling”.
War on terra
When it came, terra’s unravelling played out in less than a week. The first signs of trouble appeared over the weekend of May 7. UST dropped around 2% before recovering to $0.995 and luna fell from $80 to $60.
In the following days, UST fluctuated wildly. It fell to 40 cents on May 10 before recovering to 90 cents, then collapsed again to 20 cents on May 11 before bouncing back, briefly, to 80 cents on the morning of May 12.
By the afternoon of May 13, UST’s value had plunged to about 14 cents. The terra blockchain was halted on May 12, then again on May 13, after which no on-chain swaps could be carried out.
The attackers doubled the pressure by selling additional UST on BinanceCarlos Betancourt, BKCoin Capital
Several market participants point to an $85 million UST to USD Coin swap on the decentralised crypto exchange Curve on May 7 as the shove that first tipped the system off its footing.
USD Coin is a collateralised stablecoin backed by cash and government debt.
The trade triggered talk of trouble on social media, says Carlos Betancourt, founding principal at crypto hedge fund BKCoin Capital. The rumours in turn propelled $2 billion in withdrawals of UST from the Anchor protocol, an on-chain borrowing and lending platform where much of the UST in circulation was parked. Betancourt calls this “the initial sign of a possible de-pegging event”.
That prompted Luna Foundation Guard to activate its defence mechanism, buying Bitcoin with UST to reduce supply in the market and lift the price. In what is widely seen as a misstep, LFG had already committed to removing $150 million of UST from the main stablecoin liquidity pool on Curve on May 7 to supply a new pool. The move was meant to improve liquidity going forward but did the opposite in the immediate term.
That allowed an “unidentified attacker” to use about $350 million of UST to drain the main Curve pool, Betancourt says. “At this point… users of Anchor started hitting the exits.”
On the evening of May 9, UST dropped to 97 cents. “The attackers doubled the pressure by selling additional UST on Binance,” Betencourt says. Some estimate as much as $700 million UST was dumped over a three-day period. Whether this was part of a co-ordinated effort, how many attackers were at work and who they were remains unknown.
This was precisely the sequence of events in which terra’s founders (see box: Founding principles) had proposed arbitragers would step in. In practice, as investors began heading for the exits, the luna mint went into overdrive.
As UST’s value fell, the arbitrage mechanism necessitated the creation of ballooning quantities of its sister token. That meant luna holders were getting diluted, causing its value to crater. Selling pressure led to a one-sided market, further weighing on the price.
This is a well-known phenomenon in traditional finance, dubbed a ‘death spiral’, that is commonly seen in companies that issue convertible bonds. Bondholders convert to stock when a company runs into trouble, which dilutes the stock and causes more bondholders to convert, leading to further dilution and more conversions.
The supply of luna shot up from 343 million tokens on May 9 to nearly 169 billion on May 13 and 6.5 trillion by May 14. From May 8 to May 12, luna’s price fell from $66 to under a cent. On the night of May 10 alone, it lost more than 75% of its value. “I feel bad for anyone who slept through that,” says Wolfe Research’s Luo.
Given the time it takes to transfer USTs from exchanges to the terra wallet, that was a proper riskAndrea Barbon, University of St Gallen
The hyperinflation was so extreme it outstripped even that of Germany’s Weimar Republic in the 1920s.
“The intensity of hyperinflation undermined the stabilisation mechanism and any hope of arbitragers stepping in,” says Crachilov at Nickel. “The price of luna became monodirectional – straight to zero – with the market cap of the luna ecosystem dropping well below the minimum required amount to meaningfully try to return UST to its peg.”
When UST was at around 40 cents on May 12, arbitragers would have lost money if luna fell more than 60% during a trade. “It seems a lot,” says Andrea Barbon, assistant professor of finance at the University of St Gallen in Switzerland. “But given the time it takes to transfer USTs from exchanges to the terra wallet, that was a proper risk.”
There are anecdotal reports of UST transfers from exchanges to wallets getting stuck during this period. When Barbon experimented with moving a small sum of UST from a Binance exchange account to a terra wallet on May 11, he waited five hours. Eventually, the withdrawal was rejected because the network was too busy.
It is unclear if arbitragers were active during this time. But any efforts to stabilise UST would have added to the minting of new luna, meaning that once hyperinflation had set in, arbitrage efforts would only have made the situation worse.
In the months preceding its collapse, much of the money flowing into UST came from investors seeking to park holdings in a blockchain borrowing and lending application run on the terra blockchain – essentially an on-chain bank – called the Anchor protocol.
Anchor paid 19.5% interest on UST deposits and provided crypto loans to borrowers. Those borrowing on the platform were required to post two-times collateral in luna or ethereum. There was no lock-up.
Anchor was meant to earn staking yield on collateral, interest on loans and some fees. The volume of deposits, though, grew to far exceed borrowing, meaning Anchor was effectively running at a loss. On May 6 Anchor deposits were more than $14 billion UST versus $3 billion UST of borrowing.
Many of the structural problems with the terra project were apparent ahead of time, says Nikita Fadeev, who runs a crypto fund at $3.5 billion hedge fund Fasanara Capital.
He estimates that over $10 billion of UST deposited into the high-yielding Anchor protocol could be termed “mercenary” money from investors that would “de-risk at the first sight of worsening conditions”.
More sophisticated investors had been tracking metrics such as redemptions from Anchor, how much UST was being minted and burnt, and activity in the biggest UST wallets. Most of all, though, they were watching the capitalisation ratio of UST to luna for signs of impending freefall.
If you could reverse the trend, you’d make a killingNikita Fadeev, Fasanara Capital
When that ratio approached one, confidence snapped.
“The point had been that the market cap of luna should be a lot bigger than UST’s and that gives more confidence in the system,” Fadeev says.
During the year to May, luna’s capitalisation had been about two to three times that of UST. As investors crowded into the latter to harvest interest on Anchor deposits, its capitalisation climbed from $10 billion to $18 billion, pushing the ratio to less than two.
By May 8 the ratio was close to one. That meant any further fall in the price of luna would make it impossible to convert all outstanding UST into a dollar’s worth of luna. “At that point it was clear the system that should keep UST stable could not work anymore,” Barbon says.
The fund Fadeev runs at Fasanara pulled its money once the system’s “health metrics worsened”, he says, “and it became apparent that luna and UST were at the point of no return”.
Even at this stage, theory says the collapse could have been reversed. The Block reported plans for a possible $1 billion bailout whereby investors would buy luna at a deep discount and be locked in for two years.
But such a plan is unlikely to have worked, crypto specialists say. “Most likely it would serve as exit liquidity to those looking to get out,” Fadeev explains. “If you could reverse the trend, you’d make a killing. But to do that with a high likelihood of success would require much more capital than was available at that time.”
Terra’s founders had not ignored the possibility of hyperinflation completely.
In a 2019 white paper setting out their idea, terra’s founders noted that dilution from “unmitigated” minting of luna would “present a problem” for those holding the token. But the white paper says nothing of the likely inflation rate of luna in a panic, or the risks of such a collapse in value overwhelming the arbitrage incentives.
The $3 billion Luna Foundation Guard reserve fund, meanwhile, was intended to step in and buy UST should it lose its peg.
But the fund proved too small and concentrated. It mostly held bitcoin, which was also falling fast when UST and luna hit trouble. When LFG sold bitcoin to try to prop up UST, it pushed the price of bitcoin lower, reducing the value of the remaining fund.
Reflecting, Lipton says the system’s key failing was the assumption of instantaneous markets— that investors would be able to immediately convert UST to luna, and luna to dollars, and vice versa. In practice, doing so takes time.
Markets jump, Lipton points out. An entire body of quant research exists to model this phenomenon. Terra essentially ignored it.
If something is impossible in the real world, it will be impossible in the blockchain worldAlexander Lipton, Abu Dhabi Investment Authority
Most crypto protocols have inbuilt token issuance mechanisms that result in a certain rate of inflation. This incentivises miners to validate blocks on the chain, Crachilov says. Bitcoin’s annual inflation rate is roughly 1.8%; ethereum’s is about 4.5%.
Broadly, for actively growing crypto assets, investors are willing to accept an annual inflation rate of up to 5%, he says. “If luna’s inflation had been around that number the market would have shown greater resilience. Beyond that point, the market is naturally concerned with the dilutive effect of new coin issuance.”
Luo says terra was proposing to achieve a known impossibility. “It tried to maintain free capital flows, a fixed exchange rate with the dollar and automated printing and burning of cryptocurrency to prevent exchange rate fluctuations, which could be equated to sovereign monetary policy,” he says.
In economics this is known as the “impossible trinity”. Established theory says the three cannot occur together. If a currency floats freely without capital controls, any peg must rely on monetary policy. In terra’s case, the system should have been burning luna at the point when its flawed design required the opposite.
George Soros proved this point when he forced the pound out of the European Exchange Rate Mechanism in 1992 after the Bank of England failed to maintain its peg to European currencies. The bank had tried hiking interest rates from 10% to 15% in a single day.
Terra’s developers could have built better defences, says Campbell Harvey, professor of finance at the Fuqua School of Business at Duke University, where he teaches a course on cryptoventures, and author of DeFi and the Future of Finance. An obvious improvement would have been to hold a bigger pool of more diversified collateral. This was “no black swan” event, he says.
Not dead yet
Where does this leave stablecoins and cryptocurrencies more broadly?
For a while, the industry will be smarting. “When you have a case where something has gone from being worth $40 billion to zero within a month, obviously it scares people,” says Fadeev.
On May 12, the collateralised stablecoin tether — a core transaction currency in crypto markets with a market capitalisation of over $70 billion — also broke its peg and briefly traded as low as 98 cents. Tether has been widely criticised for failing to disclose details of its reserves.
Tether recovered, but opinion is split on whether the episode constitutes a test that the wider cryptocurrency market has passed, or a blow that might leave lasting damage.
Even the staunchest critics of terra’s project are loathe to write off algorithmic stablecoins entirely.
Am I going to write off stablecoins that are algorithmic or uncollateralised? NoCampbell Harvey, Fuqua School of Business
Crypto stablecoins are a “great innovation”, Harvey says. They meet a need in the banking system to cheaply and quickly transfer money globally and follow in a long tradition of mechanisms in the same vein. In the 1850s banks routinely issued US dollar notes, he points out, which is why banknotes are called banknotes.
“Stablecoins address a problem in our financial system and hold great promise. They’re not all the same. Am I going to write off stablecoins that are algorithmic or uncollateralised? No.”
To work, though, such coins will require insurance features to mitigate the risk of a terra-style death spiral, Harvey says. “When you’re partially collateralised there is a risk that you’re not going to be made whole for the dollar. It comes down to the other mechanisms within the actual protocol and whether there’s enough risk management. Then people potentially would be OK with a partially collateralised stablecoin. Never say never.”
Notably, during the recent turmoil, fully collateralised stablecoins such as USDC, DAI and BinanceUSD rose in price as buyers swapped into those coins.
Caution will be the watchword, though. “If you want to invest in a crypto project you need to dig down, read the white papers, understand the mechanism behind it, listen to the critics,” says Barbon.
When Lipton was responsible for approving the design of new financial products as the co-head of the global quantitative group at Bank of America Merrill Lynch a decade ago, he would ask for proposals in a half-page memo. “If they could not explain the idea in half a page, I’d kill it,” he says. “And when I received the half-pager, it would be clear in many cases that the idea should be killed even faster.”
The same rigour is needed now in crypto, he says. “Finance in a blockchain world works on the same principles as finance in the real world. If something is impossible in the real world, it will be impossible in the blockchain world. People need to switch off this suspension of disbelief.”
“I hope that going forward they will try to do a little more reading and thinking before jumping in.”
Additional reporting by Faye Kilburn and Mauro Cesa.
Stanford graduate Do Kwon and his associates formed Terraform Labs in 2018 to create a family of stablecoins pegged against different national currencies.
Their idea was to build a system that would rely on an algorithm for its stability rather than holding reserves as fiat currencies and other collateralised stablecoins do. The libertarian aim was that no central authority – such a central bank to deploy reserves – would be needed. Control over the currency would be truly decentralised. The group launched UST in September 2020.
Terraform Labs did not respond to a request for comment for this article.
As flawed as the terra project might look in hindsight, big investors were backing it. Venture capital firms in a 2021 funding round staked the company $150 million. The Luna Foundation Guard included on its governing council Kanav Kariya, president of Jump Trading’s crypto arm.
Mike Novogratz, the boss of $2.5 billion investment firm Galaxy Digital, which touts itself as the “bridge between the institutional and crypto worlds”, was a backer of Terraform and even had “luna” tattooed on his arm.
At its peak, UST’s market capitalisation reached $18 billion. Luna’s in early April was more than $40 billion.
Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.
You are currently unable to print this content. Please contact [email protected] to find out more.
You are currently unable to copy this content. Please contact [email protected] to find out more.
Copyright Infopro Digital Limited. All rights reserved.
You may share this content using our article tools. Printing this content is for the sole use of the Authorised User (named subscriber), as outlined in our terms and conditions - https://www.infopro-insight.com/terms-conditions/insight-subscriptions/
If you would like to purchase additional rights please email [email protected]
Copyright Infopro Digital Limited. All rights reserved.
You may share this content using our article tools. Copying this content is for the sole use of the Authorised User (named subscriber), as outlined in our terms and conditions - https://www.infopro-insight.com/terms-conditions/insight-subscriptions/
If you would like to purchase additional rights please email [email protected]
Former Fed adviser welcomes long-advocated Treasuries clearing mandateReceive this by email