Chapter 12 of the book Reckless: The Story Of Cryptocurrency Interest Rates is published below. The full book is available on Amazon. The book was written before the bankruptcy of FTX and therefore does not include coverage of this event. However, the book does provide useful commentary in the run up to the failure of FTX, which provides context for the eventual calamity.
From around 2018 to 2020, US Dollar stablecoins took over as the main settlement currency and trading pair coin in the cryptocurrency trading space, replacing Bitcoin. These stablecoins also took over as the main bridge between the cryptocurrency trading platforms. As a result of this, demand to use the Bitcoin blockchain declined and Bitcoin transaction fees therefore also declined from their 2017 peak. This was an era of unprecedented growth and success for blockchain based US Dollar stablecoins. There are several reasons this success:
- Stability – The primary reason is the obvious one, stability. Proprietary trading firms and market makers would rather have their working capital in US Dollars than a more volatile currency like Bitcoin.
- Trading Pairs – The trading platforms used the stablecoin US Dollar Tether (USDT) as the main trading pair against the crypto-tokens, rather than Bitcoin. This was simply more user friendly. For example, traders think about the price of Ethereum in US Dollars. They want to bet the price of Ethereum climbs from US$2,000 to US$4,000. Traders don’t think in terms of the Ethereum price being 0.07 Bitcoin.
- Regulatory arbitrage – While stablecoin issuers have censored some transactions and do have this ability, by and large, most of the time, major stablecoins like USDT and USDC have the same censorship resistance characteristics of Bitcoin. All one needs to do is generate a public private key pair and one can receive and spend USDT, in the same way as Bitcoin. Therefore, getting an exchange or trading firm set up to use USDT is easy, complex and expensive banking and regulatory relationships can be avoided. Of course, it’s possible that eventually regulators crack down and one day this changes, forcing some people back into Bitcoin, but for now USDT is attractive. For most traders USDT is better than Bitcoin, at least until it isn’t.
- Multi-chain – The stablecoins can exist on multiple blockchains. This means that USDT can be sent around for low fees on Tron and one can use USDT to interact with smart contracts on Ethereum. Of course, custodial Bitcoin can also exist on multiple chains, in the same way as USDT. However, custodial Bitcoin may not be especially attractive and removes many of Bitcoin’s key advantages.
The two largest and dominant stablecoins are USDT and USDC, both of which have a centralised custodian storing the underlying US Dollars in the banking system or treasury market. As mentioned earlier in this book, USDT is managed by Bitfinex. USDC, on the other hand, is run by a company called Circle. At the time of writing, there are US$68 billion and US$46 billion of USDT and USDC outstanding respectively. USDT has a much higher trading volume on the centralised exchanges, while USDC is more popular on smart contracts in Ethereum. USDT is therefore one of the most important tokens in the cryptocurrency space and USDT vs Bitcoin is the most liquid Bitcoin trading pair.
USDT originally launched as Realcoin in July 2014. The coin was founded by Brock Pierce, a former Disney child actor and early cryptocurrency enthusiast. The coin was rebranded to US Dollar Tether (USDT) in November 2014. USDT initially existed only on the Bitcoin blockchain, on a token layer called Omni. The Omni layer is a protocol which interprets extra meaning to otherwise surplus Bitcoin transaction data, for example the creation or transfer of USDT. From 2017 onwards, USDT was also available as a coin on the Ethereum blockchain and on the Litecoin blockchain. Due to the lower transaction fees on Ethereum at the time, USDT on Ethereum began to gain traction. Technically speaking, supporting USDT on Ethereum was the same work for wallets and exchanges as supporting the many other coins on Ethereum, because they used the same Ethereum token standard. For primarily this reason, Ethereum became the dominant venue for USDT, even as transaction fees increased. In contrast, USDT was the only major coin on Omni. More recently (from 2019), USDT also exists on the Tron network, another protocol competing with Ethereum.
USDT has not always performed perfectly with respect to the US Dollar peg. One of the main reasons for the deviations in price is uncertainty with respect to whether the coins are fully backed by US Dollars in the banking system or if there is a significant shortfall. Some examples of the most significant price deviations over history are provided below:
- 2016 Bitfinex hack – As discussed earlier in this book, Bitfinex suffered a severe hack in August 2016. This caused the price of USDT to fall below US$1 for some periods.
- 2017 Bitfinex Wells Fargo lawsuit – In April 2017, shortly after the shortfall relating to the 2016 hack had been dealt with, Bitfinex suffered delays with respect to US Dollar withdrawals and deposits. Bitfinex announced that “all incoming wires to Bitfinex will be blocked and refused by [the] Taiwan[ese] banks”. Bitfinex then filed a lawsuit against US bank Wells Fargo, because the bank “suspended US Dollar wire transfer operations needed to remit to [Bitfinex] customers”. However, this lawsuit appeared to be a public relations move, designed to make customers realise the company was working hard to resolve these issues, actually winning the case seemed unlikely. At around the same time, a Twitter account called @Bitfinexed launched. This account, which now has over 83,000 followers, promoted the idea that USDT was not appropriately backed for years. At the time, Bitfinex was working very hard behind the scenes trying to establish and maintain banking relationships, something which was clearly very challenging, given the nature of USDT. With all this concern about solvency and banking partnerships, USDT traded at low at $0.91 in the period.
- October 2018 solvency concerns – Continued concerns surrounding the backing of USDT sent the price down to US$0.85 on the Kraken exchange in October 2018. Temporary freezes of the USDT wallets of Binance and KuCoin worried some market participants.
- April 2019 Crypto Capital Corp US$850 million loss – On 25 April 2019 The New York Attorney General sued Bitfinex and published a series of documents. The documents revealed that Bitfinex may have been using Tether’s USDT reserves and appeared to lose US$850 million in loans to a Panamanian payment processing firm called Crypto Capital Corp. It was not clear who this counterparty was and Bitfinex claimed the funds were seized by a government, rather than being lost or stolen. Now it became clear that perhaps USDT did not have sufficient reserves and that Bitfinex may have put out misleading statements assuring customers that USDT was fully backed. Bitfinex went on to issue a new token, LEO, to plug the hole.
- May 2022 market contagion – Finally, on 12 May 2022, USDT traded as low at US$0.948, as fear spread in the market about the solvency of various platforms.
It is critical to realise that not anyone can redeem and create USDT. In November 2017, Tether announced that creations and redemptions were no longer available to everyone on the official Tether website.
We invite you to use the services of any one of a dozen global exchanges to acquire or dispose of Tethers for either USD or other cryptocurrencies. Such exchanges and other qualified corporate customers can contact Tether directly to arrange for creation and redemption.
However, it is worth noting that even before November 2017, it is not clear if the creation and redemption process was available to the public. What changed from this point was that it became official, and only approved organisations could create and redeem USDT. The organisations are likely to have been cryptocurrency exchanges and large proprietary trading firms. It is also probably true that even within these approved organisations, they may not have all been in exactly the same position with regards to redemptions and creations. Some firms may have been able to redeem faster, due to their banking relationships or close relationships with Bitfinex or Tether. For example, if a trading firm has an account at the same bank at which the USDT reserves are held, the redemption process may be faster. Given the difficulty and challenges for Tether in establishing the right banking partners, Tether had an ever changing and complex web of banks all over the world. In 2018, a rumour was circulated which said Bitfinex had 14 staff whose full time job was opening bank accounts. Some of the proprietary trading firms were playing a similar game, chasing Tether all over the world and constantly opening bank accounts. If they had a good relationship with Bitfinex and advanced notice, perhaps they could open new bank accounts with the same banks at the same time.
The implications of the failing US Dollar peg and the somewhat messy USDT redemption process are quite significant. It meant that when USDT traded at a discount, which as explained above happened quite often, those who could redeem USDT for US Dollars were often able to make extremely large profits in short periods of time. These funds could, for example, buy USDT for US$0.90 and redeem USDT for US$1. This trade could be repeated again and again. Therefore, it’s possible that some organisations even had an incentive to spread rumours about the incomplete backing of USDT, to make handsome low risk profits. There is always the risk that USDT is actually insolvent and that holders could have a significant haircut, however if you were able to redeem quickly, perhaps in a day or two, the risk was minimal.
In addition to the centralised custodial stablecoins above, there is also the concept of a decentralised stablecoin, a crypto-token designed to follow the value of the US Dollar, without a centralised custodian. This is said to be one of the holy grails of financial technology. A stable US Dollar coin one can use online, without any counterparty risk and where transactions could never be blocked.
The first of these so-called decentralised stablecoins we will discuss is BitUSD, a stablecoin on the BitShares platform which launched in July 2014. BitShares was a delegated proof of stake (DPOS) platform launched in 2014 by Daniel Larimer, who went on to create the coin EOS several years later. This can be described as an algorithmic stablecoin, in that there were no US Dollars in the traditional banking system backing up the coin. One can argue that the coin was crypto-collateralised, in that there was a volatile cryptocurrency, BitShares, which one could redeem BitUSD for.
To create BitUSD, one needed to post BitShares as collateral. Initially there was a 150% maintenance margin, otherwise the BitUSD position could be liquidated. The price of BitShares vs BitUSD was determined on a decentralised exchange inside the BitShares system.
There did not appear to be a specific or strong price stability mechanism in the BitUSD system. One could redeem and create BitUSD, however the price this transfer occurred at was determined by the BitUSD vs BitShares price on a distributed exchange, which was not linked to “real” US Dollars. In a way the price references itself, it was therefore somewhat circular in logic. The argument put forward by some of the coins proponents was “Why would it trade at any other price?”
It implements automatic margin calls, such that if the price moves against someone who is effectively short, it forces them to cover and buy it back in the market and that creates a peg. The market peg works on the premise that all market participants buy and sell based on what they think market participants will be buying and selling in the future. The only rational choice is to assume that it’s going to trade based on the peg in the future. If you don’t believe that then you have to decide on which way it’s going to go, up or down. And if you don’t have a way of saying you abstain from the market. If you don’t think it works you sell the shares and get out, as the system is going to fail in the first place. So its a self reinforcing market peg, that causes the asset to always have the purchasing power of the dollar. It’s the same way dollars are created in the regular banking system. Dollars are lent into existence backed by collateral, in the case of the current banking system the collateral is your house. In the case of our system its shares in the DAC itself.
It should also be noted that BitUSD lacked a price oracle which could have linked the system to real world US Dollar price, one of the most controversial aspects of its design. However, any price oracle system is challenging to implement and may introduce several weaknesses and avenues for manipulation.
The volume of BitUSD in existence was a lot lower than many had hoped for, in some periods there was only around US$40,000 in issuance. At the same time liquidity was very low and the price stability was weak. The main architect of BitUSD went on to propose a new stablecoin called SteemUSD in 2017, this time including a price oracle system. While BitUSD was an interesting early experiment, it did not achieve what was hoped for.
In a way its design was naive and never could have scaled. With an algorithmic stablecoin, there is the possibility of turbulence in periods of volatile financial conditions. At the same time, sophisticated trading firms can try to short the stablecoin and end the peg, potentially producing significant profits.
The largest, most significant and most interesting algorithmic stablecoin is Dai, part of the Maker system on the Ethereum blockchain. This coin launched in December 2017. Its design was reasonably similar to BitUSD, except it had Ethereum as collateral and included external price feeds from price oracles. Again, there was a 150% minimum collateral requirement and holders of Dai could redeem for US$1 worth of Ethereum. These price feeds linked the real US Dollar vs Ethereum exchange rate to Dai. Maker is the governance token of the system and Maker coin holders can vote on policies with regards to the stabilisation mechanisms. While Dai had many of the weaknesses of BitUSD, its design was far superior and the stability mechanism proved more effective. Today there are over US$6 billion of Dai in existence.
March 2020 Crash
In March 2020, financial markets were coming to terms with the level of disruption the policy response to COVID-19 was about to inflict on the global economy. As a result, asset prices started to rapidly decline in a crash. Bitcoin collapsed by around 50% to US$3,900. Ethereum experienced an even sharper decline, falling 64% to just US$90.
These sharp declines were exacerbated by cascading liquidations on the crypto-exchanges. The most significant of these was BitMEX, where a large number of traders with long positions were liquidated. The BitMEX trading engine then needed to close out these positions and took them over, as a result this drove the price down even further, causing more liquidations. Due to the size of these liquidation engine sell orders, the price of the perpetual swap contract on BitMEX traded well below the prevailing spot market price. On 12th March 2020, around US$753 million of long positions were liquidated on BitMEX. On the perpetual swap contract Bitcoin reached a low of US$3,596, compared to a low of around US$3,915 in the spot market. This resulted in an attractive funding rate for those who went long the Bitcoin swap. The funding rate on BitMEX was 37.5 basis points for six eight hour periods in a row. This equates to Bitcoin longs being offered around 410% per annum. However, in the market turmoil of the crash, traders were scrabbling for liquidity and were not attracted fast enough by this attractive rate, to close the price gap.
On 13th March 2020, BitMEX suffered two DDoS attacks, ten hours apart and during these periods the trading engine was unable to execute significant orders and the Bitcoin price recovered. At the time BitMEX thought it was a hardware issue. It is likely that the attacker discovered a vulnerability beforehand and then executed the attack when it would have the most significant market impact. What may have actually caused the downtime was a bug in the customer chat system, the trollbox, which was exploited by sending it a query which clogged up all the computational resources of the server. This somehow prevented messages from being sent to the trading engine and caused the trading engine to fail.
In general, the OTC market and crypto-collateralised lending firms like BlockFi and Ledn appeared to handle the situation reasonably well. It was a rocky few days and many people were liquidated, but the liquidation processes appeared to function correctly. Many providers did not liquidate positions based on the spot price low of US$3,900 and allowed customer positions to remain open, waiting for the market to recover. BlockFi said the following:
We have continued to maintain perfect performance across all of our lending activities with zero losses in the lending book. Last Thursday (3/12) evening Eastern Time, there were particularly violent downward price movements in the cryptocurrency market resulting in very limited liquidity. This was handled strategically by our team and risk management system and we did not liquidate USD loan client collateral below a price of ~$4,500, despite the market reaching lows of ~$3,800. As a result of the team’s prudent actions during this period, our clients’ capital was saved and we also liquidated a smaller percentage (<10%) of our overall USD loan book vs. other market participants. Our system has processed the largest number and volume of daily deposits and withdrawals in BlockFi’s history. Withdrawal processing is operating on our standard cycle and we remain ready to meet our clients’ liquidity needs.
In contrast to the OTC market players and lenders, Dai was not able to handle the sharp decline in the Ethereum price particularly smoothly. When the crash occurred, transaction volume on Ethereum spiked, causing severe network congestion. Part of the congestion was caused by liquidations of Dai positions. The feeds from the price oracles were also delayed by the network congestion. The price of Dai was no longer stable. On 13th March 2020 Dai traded as high as US$1.12 and as low as US$0.96.
As a result of this chaos, the Dai community realised that Ethereum was not suitable as the only form of collateral for Dai. If only Ethereum was used, it appeared inevitable that the stability system would eventually break. On 16th March 2020 there was a proposal to add USDC as a form of collateral. This was adopted and from this point forwards a US Dollar backed custodial stablecoin was used as collateral for a so-called algorithmic stablecoin.
One can argue that using USDC to back Dai makes Dai pointless. Afterall, if the US authorities want to cease the USDC collateral they can, which will then result in the failure of Dai. However, there are some important differences. Dai now has all or nothing censorship, while USDC can be censored at will. USDC has a mechanism on Ethereum where individual accounts can be frozen and this has occurred several times, typically at the request of law enforcement. However, Dai has no such individual account freeze feature. Instead, if law enforcement wants to freeze or censor anyone using Dai, they potentially must censor everyone, billions of dollars. It is possible we eventually reach the same end result, but in the short to medium term the dynamics are different.
At the time of writing, there are US$6.4 billion Dai in existence and over 42% of the collateral is backed by the custodial stablecoin USDC. Only around 10% of the collateral is Ethereum. The collateral is also diversifying further, into traditional investment funds, even purchasing real estate. Part of the justification for this is that the Maker community are concerned that they are too dependent on USDC, which could at any point freeze all the balances at the request of US law enforcement. It is believed diversification will reduce this risk. A catalyst for this was the US Treasury’s August 2022 decision to sanction the Ethereum mixer contract Tornado Cash.
The final stablecoin we will discuss is UST, part of the Terra blockchain, which launched in April 2019. Like BitUSD years earlier, it can be considered an algorithmic stablecoin. There was a sister coin to UST, a proof of stake cryptocurrency called Luna. Luna was also the collateral backing up the stablecoin. If UST traded at a discount to the US Dollar, Luna tokens could be issued and UST holders should be able to, at least in theory, redeem their UST for US$1 worth of Luna. However, this mechanism has a clear and extremely obvious potential flaw. The new issuance of Luna could increase the supply of Luna in the market, which could cause the Luna price to decline. Then more and more Luna would need to be issued to maintain the peg, causing further price declines and a potential collapse of the system. However, for some reason, many seemed to overlook this and regarded UST as relatively safe.
The Terraform Labs company behind the chain was co-founded by Do Kwon and was incredibly successful in the 2021 period. By 2021, TerraForm Labs (TFL) was one of the VC darlings of the cryptocurrency space, regarded as having top technology and top marketing capabilities. In the company’s most recent funding round in July 2021, Terraform Labs secured funding of over US$150 million, from investors such as Arrington Capital, BlockTower Capital, Delphi Digital, Galaxy Digital, Hashed, Lightspeed Ventures and Pantera Capital.
The UST stablecoin was a huge success in 2021 and early 2022. At the start of 2021 around US$180 million of UST were in issue and by the peak, in May 2022, the issuance reached over US$18 billion, with steady and consistent growth in the intervening period. The stability mechanism worked reasonably well. However, there were some periods where the algorithm did not achieve the desired stability and TFL had to intervene to maintain the peg. For instance, in December 2020, when the price of UST fell to US$0.85.
It is certainly inaccurate to describe Mr Kwon as a modest individual. He is known for aggressively defending and promoting Luna and UST on Twitter. For example, when cryptocurrency analyst and fund manager Eric Wall criticised some of the mechanisms inside Luna, Mr Kwon Tweeted:
Reminder that unlike me, your rando ass blogs for a living. Do your job, or stay in irrelevance. That’s all I ask.
In November 2021, another analyst provided a brief outline of how a wealthy attacker could break the UST system and profit, by conducting a “Soros style Black Wednesday attack”. Mr Kwon responded by saying:
Probably the most retarded thread I’ve read this decade. Silence is a perfectly acceptable option if stupid. Billionaires in my following, go ahead, see what happens.
Some of the arguments put forward by Do in defence of UST’s stability systems, had echoes of the arguments put forward by Dan Larimer seven years earlier. Do compared UST to the way US Dollars are produced in the traditional banking system.
I do not think algorithmic stable coins fail because of algorithms, I think it fails because of the stability of the economies that are backing them. What I mean is an algorithmic stable coin and a national currency are actually very similar constructs. It is all based on the quantity theory of money. The idea is that you should be able to absorb contractions and expansions in currency and the best way to do that is to build a robust economy surrounding it.
UST’s sister coin, Luna, was one of the best performing coins in the bull market of 2021. In January 2021 the coin was trading at around US$0.60. By the end of 2021 Luna traded at around US$90, amazing 15,000% returns. The coin continued to perform into 2022, reaching a peak price of around US$120 and a peak market capitalisation of over US$40 billion.
In September 2021, Do Kwon and TerraForm Labs (TFL) announced something called Project Dawn. This was described as “a new funding initiative for critical infrastructure improvements and core technologies to supplement the accelerating growth of the Terra ecosystem”. This essentially involved unlocking Luna tokens for use for TFL’s operating expenses. The statement from Do Kwon read:
Project Dawn has commenced as of today with a 5 million LUNA unlocked and distributed by the TFL Genesis wallet (market value $150M). Further to Project dawn, TFL is committing to unlock at most 3 million Luna per month for all operating costs with details around each unlock transparently relayed to the community.
The spot price of Luna at the time was around US$30. This therefore implied TFL intended to spend US$90 million per month on operating expenditure, which is an exceptionally high budget. As for the details and transparency surrounding this, it’s not clear such details were ever made available.
In February 2022, Terraform Labs announced a sponsorship deal with Major League Baseball’s Washington Nationals, using US$40m of funds inside the Terra DAO. Luna holders had to vote inside the DAO to approve the sponsorship deal. The deal would mean the baseball team would sell tickets using UST.
In January 2022, a foundation called the Luna Foundation Guard (LFG) was set up in Singapore. Do Kwon was a founding member and director. The initial capital was a 50 million Luna gift from TFL. The LFG then began acquiring Bitcoin. On Twitter on 5 May 2022 the company stated:
The LFG has acquired an additional 37,863 Bitcoins totalling ~$1.5 billion in OTC swaps with @GenesisTrading and 3AC. The OTC swaps included 1 billion $UST for $1 billion worth of $BTC with Genesis, and the LFG purchased an additional $500 million worth of BTC via 3AC. The acquired BTC brings the LFG’s total holdings to ~80,394 Bitcoins.
As a result of this the LFG had around US$3.5 billion of Bitcoin and the company announced that it planned to increase the holding to US$10 billion. These funds were essentially obtained from the proceeds of the Luna coin offering and Luna coins allocated to TFL. The purpose of the Bitcoin holdings was to diversify the stability mechanism for UST. This could improve the robustness of UST, similar to how Dai diversified its collateral. Rather than only Luna backing up UST there could be a separate stability algorithm operating in parallel involving Bitcoin. Bitcoin is more liquid and stable than Luna, therefore this would reduce risk. In addition to this, with the Bitcoin backing UST, more Bitcoin would not be created if UST traded at a discount, instead reserves would be used and therefore there is no risk of a cascading collapse. Reserves were either sufficient or they were not. The idea was to construct a bridge between this custodial Bitcoin held by the LFG and the Luna blockchain, such that when inside Luna, the Bitcoin stability system could operate autonomously. These algorithmic stability systems involving Bitcoin were never constructed, nor was the bridge. Instead, the LFG just conducted manual interventions in the market using their reserves. The UST system never achieved an effective collateralisation level, unlike Dai for example, which was over collateralised. In addition to Bitcoin, the LFG also held an alternative coin called Avalanche (AVAX), for a similar reason but in a smaller quantity.