The Collapse of Terra Classic
This article is the third of a four-part series by YaaS Analytics discussing algorithmic stablecoins, with a focus on the UST crash and the resulting fallout. The series consists of four articles:
3. The Collapse of Terra Classic
4. The Aftermath of the Terra Crash
State of the Market Before the Crash
Prior to the crash, the crypto market was beginning to stabilize after a couple of bearish months. Bitcoin was trading at $36,000 and Ethereum at $2900 . On May 7th directly before the crash began, the Luna Foundation Guard’s reserves consisted of:
- 80,394 BTC
- 39,914 BNB
- 26,281,671 USDT
- 1,973,554 AVAX
- 697,344 UST
- 1,691,261 LUNA
This made the LFG the 7th largest single holder of BTC. These reserves had been most recently bolstered two days prior on May 5th, via a $1b OTC swap with crypto broker Genesis in exchange for $1b worth of UST, as well as a $500m purchase of BTC from the now insolvent crypto hedge fund Three Arrows Capital, also paid in UST. LUNA was trading at $86, down from its all-time high of $119 one month prior, while UST was holding its peg steady at $1 with a total market cap of $18.6b. As well, the total value locked in DeFi protocols was $200b at this time, down 20.6% from its all-time high of $252b in late December, but 9.5% higher than its YTD low of $181b in January.
Not the First Time
The most recent depegging event was not the first time that UST lost its peg. In fact, UST had previously lost its peg almost exactly one year ago in May 2021, when the price dropped more than 5 cents below its peg. The price began to depeg in the first week of April, when it dropped to $0.9944. UST continued to fluctuate between $1 and $0.995 for the next month and a half, before dropping to $0.9838 on May 20th, and then further dropping to $0.9457 on May 22. The price quickly recovered, and by the start of June, UST had regained its peg.
Market conditions at the time of this depeg were very similar to those at the time of the 2022 depeg. Within the weeks leading up to depeg, the crypto market was experiencing extreme volatility, with the price of BTC dropping 40.4% in the span of 2 weeks, falling from $58.2k on May 8th to $34.7k by May 22nd. Similarly, ETH had dropped even more dramatically, falling 49.6% from $4.17k on May 10th to 2.1k on May 22nd. As the price of LUNA began to fall, rapid liquidations on Anchor caused it to fall further. However, at this point the market cap of UST was only around $3b, and there were not nearly as many whales invested in the protocol. As well, many of the investors in the protocol were genuine believers and not simply driven by the attractive high interest rate offered by the anchor protocol. This made a ‘bank run’ less likely, thus a death spiral never initiated, and UST was able to naturally regain its peg by the start of July.
Notably, this depegging event marked the first time the total market cap of LUNA fell below that of UST. In the history of UST, this has only happened twice; during the May 2021 depeg and the May 2022 depeg. This meant that during these times, UST was not fully backed; all the LUNA in the ecosystem could be sold, and it would not be enough to cover the outstanding UST.
This ratio is notable because it illustrates how much LUNA can be burned in response to price volatility from UST. Soon after UST regained its peg in late June, this ratio rose back up, eventually reaching ~5, meaning that there was about $5 worth of LUNA in circulation for every $1 of UST. This remained the case until the November 2021 burning event.
November 2021 Burning
On November 9 2021, the Terra community voted to burn 89m LUNA tokens, roughly equivalent to $4.5b worth of LUNA, over the following weeks. This move, proposed by Terra co-founder Do Kwon, was designed to increase the supply of UST in the ecosystem, as well as to increase the value of LUNA by decreasing the supply. In response to the announcement, the price of LUNA shot up from $50 to $54. While this move did indeed lead to a higher valuation of LUNA, it also caused the ratio of LUNA market cap to UST market cap to fall. The November burning caused this ratio to fall from about 5:1 to almost 2:1 within a matter of months, drastically altering the supply relationship between UST and LUNA and lowering the volatility cushion inherent in UST.
The Beginning of the End
On May 7th at 9:44pm, Terraform Labs withdrew 150m UST from the Curve 3pool in order to get ready to fund the Curve 4 pool, which was to go live on Ethereum the week after. Prior to this withdrawal, there was about $650m UST in the 3pool. Terra, along with another growing algorithmic stablecoin FRAX, aimed to make the 4pool the most liquid trading pool on Curve. Terra had this goal in mind because of the way through which Curve Finance managed their pools; projects would compete to provide the deepest liquidity to their pools, with rewards being distributed in Curve’s native token, $CRV. The more CRV tokens a project holds, the more leverage they have to vote on how token rewards are distributed to the pools. This gives projects a huge incentive to add liquidity since the winning pool would generate higher yield on their stablecoins thus attracting more liquidity. Additionally, having a deep liquidity pool on Curve is especially important for an algorithmic stablecoin such as UST since it provides the deep liquidity necessary to maintain the token’s dollar peg. The removal of a large amount of UST from the Curve 3pool on May 7th made it more susceptible to disruption from a ‘whale’. Thirteen minutes later at 9:57pm, an anonymous wallet sold 85m UST into the Curve 3pool, causing an imbalance in the pool and causing UST to start to depeg. The wallet behind this swap has come to be known as “Wallet A”, and it is suspected that this swap was part of a planned attack on UST’s peg. It has become highly suspect as Wallet A, in addition to the 85m Curve transaction, also withdrew around 200m UST from Anchor on May 7th, and sold 108m UST on Binance causing UST to depeg on that exchange as well. Within one hour of this 85m UST swap on Curve, another trader swapped a total of 100m UST for USDC over the course of four trades further disrupting the pool’s balance. Terraform Labs responded by withdrawing another 100m UST from the pool in an attempt to rebalance the pool, but by this point the damage was done, and these large trades in quick succession had started to irreversibly break the UST peg.
Vive La Resistance
UST however did not go down without a fight, and multiple entities committed to repairing its peg. Chainalysis reported that three unidentified traders swapped a combined $480m USDT for UST across three trades on the Curve 3 pool. The first trader bought 276m UST on May 7 at 10:55pm, the second bought 108m UST on May 8th at 2:25am, and the third and final trader bought 96m UST on May 9th at 5:02am. Using data pulled from Dune Analytics from May 7th to May 24th, the graph below displays the peaks and valleys as attempts were made to rebalance the Curve 3pool but eventually the pool was drained of USDC/USDT/DAI as people sought to exit their UST positions.
The price continued to fluctuate slightly below its peg on May 8th, at which point the Luna Foundation Guard committed to loaning $750m BTC to market makers in order to defend UST’s peg and pledged another $750m in UST to buy back the BTC after the peg recovered.
On May 9th, the Luna Foundation Guard transferred 52,189 BTC to a counterparty in exchange for 1,515,689,462 UST. On May 10th, Terraform Labs sold a further 33,206 BTC on behalf of the LFG in exchange for an aggregate 1,164,018,521 UST. These trades briefly recovered UST’s peg, but by this point a death spiral had begun to initiate as fear and panic pervaded the market, and by late May 9th UST had once again lost its peg, this time permanently.
The Lunapocalypse
On May 9th at 11pm UST dropped to a then all-time low of $0.7693. Around this same time, LUNA was trading at $31.84, down from its pre-crash price of $87.4 days prior. The prices of both tokens never recovered; on May 10th UST experienced a brief resurgence to $0.9318 before dropping to $0.3063 on May 11th, with LUNA trading at $0.65 by 11pm on the same day. By May 13th UST was trading at $0.154 and LUNA was trading at $0.0001 and by June 1st UST was consistently trading below $0.02, while LUNA was trading at less than $0.00001, an effective drop of 100% from its pre-crash price.
The Perfect Storm
The vast majority of all UST in circulation was on the Terra blockchain, with about 18.7b, followed by 2b on Ethereum. Of the UST on Terra, over 72% was locked in Anchor. As such, Anchor was clearly the most influential player in UST’s ecosystem with the majority of UST holders depositing on Anchor to take advantage of the generous earnings of almost 20%. At the beginning of 2022, there was already a lot of negative sentiment in the community around Anchor and the lack of liquidity in the protocol’s reserves. True algorithmic stablecoins are a faith-based peg that relies on true believers of the system — what happens if your supporters get scared? May 7th brought about conditions that may be considered the perfect storm, with crypto markets beginning a downward trend, lower liquidity in the UST pools on Curve, Binance being hit with a large quantity of UST, and a few large withdrawals off Anchor totaling almost 2 billion. Withdrawals off Anchor continued at a steady rate and the TVL on Anchor went from $14b to $9b in less than 48 hours. In the span of one week, TVL dropped over 98%, from $16.75b on May 6 to $0.314b on May 13. Within one month of the crash, TVL in Anchor was sitting at $4.25m.
Looking at transactions on anchor occurring between May 7 and May 8 — around the time UST began to depeg, we categorize these transactions by amount, with the following classifications:
- Whales: > $5m
- Large: $1m — $5m
- Medium: $100k — $1m
- Small: $10k — $100k
- Very Small: < $10k
From the total hourly redemptions graphed below, there are two big spikes in total redemptions, the first occurring on May 7 at 10pm and the second, larger spike occurring on May 8 at 1am. These spikes correspond to $480.5m and $723.7m in redemptions respectively. Notably, the composition of the transactions making up these spikes differs greatly, with whale transactions comprising only 47.2% of the first spike compared to 82.1% of the second. This means that the second spike marked a huge rush for whales to get their money off of Anchor. Typically these larger wallets are more likely to have had built-in alarm systems that were able to react to the depegging event in order to get out quickly.
The Rush to Exit
There were a couple main options through which people could exit their UST positions; through the Terra protocol (basically this acted as an algorithmic central bank where 1 UST could always be swapped for $1 of LUNA) or at the open market such as decentralized exchanges (eg. Curve pools, Terraswap) and centralized exchanges (eg. Binance, OKCoin). With the mass exodus off Anchor beginning and UST down to 98.5 cents, fear took hold and the number of users trying to exit their positions grew rapidly.
Exiting at the Protocol
As discussed in previous articles, UST is a seigniorage style algorithmic stablecoin, with a sister token LUNA that is used to absorb price volatility from UST. It does so by expanding or contracting the supply of UST by providing arbitrage opportunities to users. When the price of UST drops below $1, holders are able to sell their UST back to the protocol in exchange for $1 worth of LUNA. The protocol then burns the repurchased UST, lowering the supply and thus increasing the price. The converse is true if the price of UST goes above $1. Users can then buy newly minted UST from the protocol in exchange for $1 worth of LUNA. This then expands the supply of UST on the market, and thus lowers the price back down to its peg.
Due to the way the protocol functions, since the price of UST was dropping, rather than sell on the open market, UST holders would try to become whole at the protocol through LUNA. There was a rush to sell their newly minted LUNA since LUNA’s price was already crashing. Additionally, there was a cap on how much UST can be burned daily (thus LUNA minted) at the Terra protocol which increased the already intense sell pressure for UST. The Terra blockchain could also only handle so much activity and ‘panic selling’ ensued as the Terra blockchain became clogged. The number of UST burn requests at the protocol skyrocketed to more than 50,000 on May 11th, compared to an average of 1000 on a normal day.
Using data pulled from Flipside Crypto, graphed below is the number of LUNA minted and UST burned at the protocol between April 1 and May 31. Here we note the predictably massive spike in mints and burns around the time of the crash. On May 11, Do Kwon revealed plans to implement community proposal ‘1164’, which proposed increasing the minting capacity of LUNA from $293m to over $1.2b, allowing for LUNA holders to absorb the volatility from UST more quickly. However, by increasing this minting capacity, the price of LUNA would become susceptible to dropping even further. This announcement allowed copious amounts of LUNA to be minted causing further inflation and LUNA’s price to drop even further. As a result, on May 12 Terra was forced to halt their blockchain at a block height of 7603700. This halt lasted for 2 hours, and was soon followed by LUNA hitting an all-time minting peak of 5.88t tokens minted on May 13.
The Osiris of This Meltdown: Curve Pools
We know that the initial events that set the depeg into motion (Terra’s 150mm UST withdrawal, Wallet A’s 85mm UST selloff) occurred on Curve. In order to move UST to the Curve pool, users would be required to bridge their UST to Ethereum using one of the Terra/Ethereum bridges. The most popular was through Wormhole. The chart below displays the number of users bridging UST to Ethereum through Wormhole each day. This number peaked at 56k on May 11, followed by 55k on May 12, before steeply dropping to just 6.5k on May 13 as the supply on curve pools were drained.
At the time of writing, the UST + USDC/USDT/DAI balance in the 3pool stands at $474k, down 99.96% from its pre-crash total of $1.2b.
Binance
UST’s depeg occurred across multiple exchanges. Wallet A’s 85mm dump of UST on Curve caused UST’s price to begin to teeter on that platform, but Wallet A didn’t stop there; they proceeded to offload a further 108mm UST on Binance, causing the price to depeg on Binance as well. This naturally led to capital flight on Binance as users sought to exit their UST positions. Binance offered several UST pairs on their platform such as UST for BTC, UST for BUSD and UST for USDT. Using market data pulled from Binance, we examine one of the most popular of these swaps; UST for USDT. Graphed below is the number of swaps per hour from the start of the depeg on May 7 until Binance froze UST trades on May 13. Here we see that there were very few swaps for most of May 7 with a brief spike of 18.5k trades from 10–11pm, before dying back down throughout May 8. There were relatively few trades for the majority of May 9, before a huge spike from 5.6k trades between 6 and 7pm to 47.5k trades from 7pm to 8pm. Swaps continued to rise during May 10 and into May 11, where the number of hourly swaps hit its all time high of 216k between 6 and 7am. This indicates that it was the morning of May 11 where alarm bells really began to sound off on the UST crash leading to a frenzy of swaps.
Most interestingly, plotting the average number of UST swapped per transaction per hour gives us a graph that appears almost like an inverted version of the first graph, with by far the largest transactions (in terms of token quantity) occurring before May 10, and especially concentrated around the evening of May 7. This indicates that bigger players were the first the liquidate their UST positions, doing so well before the average user. The spike in average UST swapped per transaction seen on the 7th is likely a result of the 108m dump done by Wallet A. Indeed, the average UST tokens swapped per transaction before May 10 is 10220.39, compared to 1809.07 after; meaning that transactions before May 10 were on average 5.64 times as large as those taking place on or after May 10.
Should They Have Known?
Terra was very aware that they neeeded to collateralize UST at least partially in order to support the peg amid volatile market conditions, thus the creation of the Luna Foundation Guard (LFG). They were also aware they required deeper on-chain liquidity to help stabilize the peg thereby announcing the creation of the Curve 4pool. Furthermore, Terra had plans of reducing the high interest yield on Anchor as it was continually draining the reserves since the actual interest rates from the protocol’s borrow side was not enough to support the depositors. Whether it was a planned attack on a vulnerable ecosystem or too little support too late, one thing is for sure — all ‘true’ algorithmic stablecoin interations, (those backed only by smart contracts and algorithms), have failed to date due to their inherent vulnerability in volatile market conditions.
Of the 18.67b UST circulating prior to the crash, 72% of holders were invested in the Anchor protocol. Thus the main driver of UST adoption clearly was to farm yield off the astronomical rates provided by Anchor, rather than a belief in the future of the currency and protocol. These users are less likely to stick with a protocol through periods of adversity and as such were far more likely to jump ship as soon as UST’s price began to wobble. As these users exited their positions, they caused the price of UST to drop further, perpetuating the fear surrounding the protocol and motivating even genuine believers to withdraw their money as it became less and less likely that UST would recover. When mass sell-off ensued it clogged exchanges, congested the blockchain and thus causing the sell pressure to skyrocket until a ‘death spiral’ began.
Almost 60b was wiped out in a span of days causing ripple effects throughout the entire crypto ecosystem and sparking important conversations about controls and regulations in the industry. In our next article we discuss these effects and the aftermath of the UST crash.
By Jack McKay and Julie Paterson
