AI-Based Stable Coin
By Yulin Liu on ALTCOIN MAGAZINE
Disclaimer: The views expressed in this article are those of the author and do not necessarily reflect the official position of the DFINITY Foundation.
In 2008, the global financial crisis, caused by over-leveraged mortgage-backed securities and unregulated market speculation, led to widespread public distrust in the centralized, interdependent monetary and financial systems. The distrust was later on fueled by the central banks’ controversial policies. After lowering the nominal interest rate to the effective zero lower bound, there was not much room to further cut the policy rate. The major central banks have been relying on the unconventional monetary policy, e.g. quantitative easing. In layman’s terms, quantitative easing means the central bank channels a large amount of newly printed money to the real economy in the hope that firms with new bank loans could expand the business and hire more workers.
However, in the aftermath of the financial turmoil, global economic development has stagnated. Banks and investors lost interest and confidence in the real economy. The newly printed money flowed mostly to the housing market (e.g. in China) and the stock market (e.g. in the U.S.), instead of to the small and medium enterprises that hired the most workers and were short of bank loans. The effect of the unconventional monetary policy was limited in boosting the economy. Yet the cost of quantitative easing was high. The wealth of fiat currency holders was diluted and the middle class suffered the most. The wealthy with easy access to cheap funds and high-return investment opportunities became richer and the poor poorer. The unconventional monetary policy accelerated the wealth redistribution from the middle class to the super-rich.
As wealth inequality gets exacerbated, it is not surprising that we have witnessed the election of Donald Trump, Brexit, the Yellow Vests Movement in France and the recent riots in Hong Kong. As the saying goes, the capitalist does not know how to divide the pie while the socialist does not know how to grow the pie. Society keeps a delicate equilibrium between the middle class and the wealthy when the pie grows. However, once the pie stops growing, social problems pop out one after one. There is no doubt that there will be more populist movements as the capitalists keep swallowing a bigger part of the pie while the government is not helping, and may actually be exacerbating the situation (e.g. in 2013 the Cypriot government decided to freeze bank accounts and impose bank deposit levy).
Cryptocurrency, a synthetic asset, and programmable money was born against this backdrop. Unlike government-issued fiat currencies whose total supply is at the discretion of the central bank, the supply of cryptocurrency can be predefined and capped. The path of money growth can be programmed and committed, which solves the time-inconsistency of the monetary policy. The time-inconsistency here refers to the scenario where the central bank commits to certain policies and later changes it abruptly, which typically causes an abrupt market reaction and sometimes even financial turmoil. With coded monetary policy in the blockchain consensus, the decision on money growth is no longer dictated by a group of centralized agencies like the government and the central bank. The inviolability of private property is engraved in the DNA of certain cryptocurrencies.
The first cryptocurrency, Bitcoin, was invented as a “peer-to-peer electronic cash system”. However, the value of Bitcoin has been very unstable since its birth in 2009. The wide usage of most cryptocurrencies is precluded by its volatile price swings. Stable coin, a bespoke cryptocurrency, is designed to tackle the problem of value instability. The different designs of the stable coin are presented at length in the next section.
Stable coin is nothing new. Currently, more than 60 countries either peg their currencies to the US dollar or use the US dollar as their legal tender. The reason for this is those small economies are more susceptible to economic shocks and thus have more volatile exchange rates, which impedes international trade. Pegging to currencies of large economies, like the US dollar and Euro, helps lower the trading uncertainties.
Stable Coin Taxonomy
Fiat-Backed Stable Coin
A fiat-backed stable coin is, in essence, IOUs. Users wire fiat currency to the bank account of a centralized institution (i.e. the debtor) and receive stable coins issued by the debtor as a receipt. The stable coin is a claim of the fiat deposit in the bank. In other words, the stable coin is the tokenized fiat currency. Users could redeem the fiat deposit by sending back the stable coin to the issuer. The issuer then burns these returned stable coins. In doing so, every existing stable coin in circulation is backed by one fiat currency unit in the bank account.
So far most fiat-backed stable coins pegged to the US dollar. Examples are USDT issued by Tether, TUSD by TrustToken, USDC by Circle, GUSD by Gemini and PAX by Paxos Standard. Fiat-backed stable coins, especially Tether, are by far the market leader. It has been popular among users partly because its stability mechanism is very intuitive: every USDT in circulation has one US dollar stored in the bank.
However, fiat-backed stable coin raises severe concerns. It is prone to counter-party risk, as the issuance of stable coin is controlled by a centralized institution. The issuer could easily print more stable coin than the bank deposit or even abscond with the bank deposit. In particular, Tether is notorious for appropriation of the funds due to lack of transparency and supervision.
In addition, the funds saved in the bank are prone to third-party risk. Banks could default or the fund could be frozen or confiscated by the government. Therefore, for the fiat-backed stable coin to function well, users must trust the issuer and the custodian banking institution.
Moreover, it is cost inefficient. Fees paid to the middleman, e.g. auditors, custodian banks, financial firms, regulators, lawyers and so on, are ultimately undertaken by users. To make things worse, fiat currency itself has been losing purchasing power as the Federal Reserve targets a 2% annual inflation rate. Anchoring a stable coin to unstable government-issued IOUs is a ludicrous temporary solution.
Crypto-Backed Stable Coin
In real life, borrowers could pledge their properties such as real estate as collateral to banks. Banks create money out of thin air and lend it to borrowers. When the loan matures, the borrower needs to pay back the principal and interest of the loan. Failing to do so would result in the sale of the collateral to recoup the bank loss. As the market price of the collateral is usually fluctuating, the bank loan is over-collateralized. That is, the value of the loan granted by banks is less than the market value of the collateral.
In a similar vein, in the crypto world, users could pledge their crypto assets as collateral in a smart contract (i.e. autonomous bank on the blockchain). Then, the smart contract automatically generates stable coins (i.e. loan in crypto fiat) against the collateral to the user. Because the value of the crypto collateral is volatile, the stable coin is over-collateralized to buffer the price swing of the crypto collateral.
As banks are replaced by the autonomous smart contracts and multifarious administrative work is largely simplified, users are exempted from the middleman fees and can easily create stable coins 24/7 by a few clicks in their crypto wallet.
Unlike fiat-backed stable coin, the crypto-backed stable coin is decentralized, transparent and traceable. Users’ collaterals are locked in the smart contract. The issuance of a stable coin can be publicly tracked and verified. Moreover, compared to the fiat-backed stable coin, the liquidation of crypto-backed stable coin is instant. Users could withdraw their crypto assets within a few seconds by sending back the stable coin to the smart contract at any time.
The deficiencies are also obvious. The stability mechanism relies on the underlying crypto collateral. The stable coin could become under-collateralized if the value of the underlying crypto assets drops too fast.
A related challenge is the cascade effect. In the presence of a collateral price drop, some stable coin loans become under-collateralized. The smart contract automatically liquidates the collateral to buy back the stable coin. The firesale further drives down the price of the collateral if the loan is sizable, which could trigger the under-collateralization of some other loans, causing an avalanche of non-performing loans and crash of the underlying collateral. It is, therefore, more resilient to adopt multiple collaterals to diversify the idiosyncratic risk of each crypto collateral.
Moreover, the system is vulnerable to hacks of the crypto assets locked in the smart contracts if the codes are not well written. In such cases, the recourse of collateral is almost impossible.
Seigniorage-Shares Stable Coin
The exchange rate of sovereign currencies mirrors the interplay between the demand for and supply of the currency. When the supply exceeds the demand, the currency depreciates and vice versa. To maintain a stable exchange rate, nowadays central banks tend to adjust the money supply by buying and selling government bonds to counteract the varying money demand. Specifically, when the currency depreciates, the central bank sells government bonds to the market to withdraw money from the market. By doing so, the money supply contracts and the currency appreciates. Vice versa, when the currency appreciates, the central bank buys government bonds and injects money into the market.
Seigniorage-shares stable coin adopts a similar approach. The system generates money to bondholders in exchange for their bonds in order to increase the money supply. Vice versa, the system could contract the money supply by selling newly-issued bonds to the market. The challenge is that government bonds are backed by the government’s ability to levy taxes. Bond buyers are assured of the value of government bonds. What backs the bonds issued by the seigniorage-shares stable coin system?
Take Basecoin as an example. In order to contract the money supply, Basecoin sells bonds to the market at prices lower than $1. Stable coin holders could exchange their stable coins for bonds and they are entitled to receive $1 stable coin for each bond in the future monetary expansion. That is to say, bond buyers sell their stable coins to the system in exchange for the seigniorage-share in the future, i.e. receive more stable coins in the future.
During money contraction, the bond is sold to the market via auction and the price of the bond is lowered until enough stable coins are collected from the market. In the case of a black swan event or a long period of money contraction, users lose confidence in the stable coin. The bond price can quickly fall into the death spiral: users demand a lower price of the bond to account for the systemic risk. Then the system has to issue more bonds to purchase a certain amount of stable coin. With more bonds issued, bonds become even cheaper. A death spiral is triggered. Ultimately, the bond price could fall to zero while the money supply is still excessive. Consequently, the bond becomes worthless and the stable coin becomes unstable.
The stability mechanism of the seigniorage-shares stable coin is vulnerable. It cannot withstand a long time shrinking demand for the stable coin. It is also fragile in the presence of a rumor-triggered drop of the bond price. The stability is contingent on the unstable market sentiment of its fundamentals.
Nevertheless, it is an ambitious design and paved the way for the future design of a stable coin. It differs from the fiat-backed and crypto-backed stable coins, as no collateral is needed. Therefore, it is not relying on or tied to other crypto or fiat currencies. Instead, it creates a new crypto asset, its bond, to maintain the stability of its stable coin. By shifting the volatility of its stable coin to the bond tokens, the system tries to maintain the stability of its stable coin. The problem lies in the fact that the bond lacks endorsement. Building a stable coin on top of an unbacked asset is like building on sand.
The Ultimate Form: AI-based Stable Coin
The stable coin needs to be endorsed by some fundamentals that support its value. A fiat-backed stable coin is backed by the fiat currency and the fiat currency is endorsed by the government’s fiscal revenue. A crypto-backed stable coin is backed by its crypto collateral. The value of the underlying crypto asset is supported by the utility its native blockchain provides to users, whether as a store of value (e.g. Bitcoin) or as a platform of smart contracts (e.g. Ethereum).
The fundamental problem of the collateral-backed stable coins is that their success relies on the underlying collateral. Fiat currencies have been losing purchasing power over time as most central banks target a 2% inflation rate. And no fiat currencies can be guaranteed to maintain a stable value. Major sovereign currencies like Euro, US Dollar, and Swiss Francs have all experienced drastic value plunges, either intentionally by the government in a trade war or unintentionally due to the sunspot market cycle. Not to mention crypto assets. A substantial amount of crypto assets have emerged, developed, soared and vanished within a couple of years. Building stable coin on top of these collaterals involves a huge systemic risk.
Seigniorage-shares stable coin does not rely on exogenous assets. Instead, it endogenously creates its own underlying asset — its bond. However, the bond itself lacks value endorsement. An unendorsed stable coin without fiscal revenue is doomed to fail. To fix it, one needs to find a sustainable revenue model that supports the value of the stable coin. As stable coins are vehicles for value exchange, naturally the system could charge transaction fees from users and utilize the generated revenue to back the value of its stable coin.
To illustrate, consider the AI-based Stable Coin by AISC. The system charges transaction fees between 0–1% paid in AISC. In the case of appreciation, the transaction fee is lowered. Consequently, the amount of transactions increases and the flow of AISC in the market increases. Vice versa. When the exchange rate depreciates, the system increases the transaction fee to suppress the supply of AISC tokens in the market. That is to say, the fluctuating demand for the AISC token is counteracted by the elastic supply of AISC, which is adjusted by the transaction fee. The transaction fee is adjusted stepwise according to a predefined function. The transaction fee will be an effective tool in managing the supply of AISC token when the system matures. Some other measures in safeguarding its stability are presented below.
The augmented transaction fee in the presence of a downward breach of the parity could also resolve the death spiral, a common challenge faced by stable coin systems. When the exchange rate is below 1, token holders fear to be the last one to sell and tend to sell in panic which further drives down the exchange rate. Increasing transaction fees in such circumstances could largely alleviate the panic firesale. To further incentivize holders to hoard stable coins in this abnormal time, the system could even distribute the transaction fees proportionally to the holders.
It is interesting to compare the decision making between the committee-based Federal Reserve and the AI-based Stable Coin System. The dual mandates of the Federal Reserve are price stability and maximum employment. Generally speaking, when the inflation rate is lower than the 2% target or the employment rate falls below its natural rate, the Federal Reserve lowers the interest rate to stimulate the economy. For AI-based Stable Coin System, the human decision is replaced by the algorithm and the algorithm evolves over time when more on-chain data are fed into its machine learning system. In other words, the AISC system codifies central bankers. The principle of “Code is a law” contrasts with the “dictator”-style central banking system which has severe time-inconsistency problems (there are numerous examples of how central banks failed to deliver what they had promised) and claims to be independent of the government and immune from the pressure of politics. Apparently, Donald Trump does not think so.
An algorithm-based automated system processes more information in a more efficient and objective manner than the information-constrained, politically-biased and labor-based decision making of monetary and fiscal policies. With the emergence of distributed ledger technology, it is time to reflect on questions such as what is the justification for the government being the monopoly of currency issuance and what makes central banks better at issuing money than others.
The financial system could be made more open and decentralized so that the ruling class does not possess unduly amounts of financial power. This would hopefully prevent such scenarios where the ruling class may be tempted and enabled to abuse their position in further increasing the financial inequality between the classes.
That said, the AISC system still needs a committee of monetary experts to vote for the decision in the case of black swan events. Human intervention is necessary when the system derails, especially at its infant stage. A group of financial scholars and practitioners are randomly selected to correct the AI-based system. Small adjustments of transaction fees in the range of [-0.1%, 0.1%] within 24h are automatically executed by the system. A large deviation requires voting of the committee. Such reliance can be relaxed in the long run when the system becomes smarter and more resilient.
Anchor
Most stable coins peg their value to the US dollar. Placing faith in the Federal Reserve and tying its hands to a centralized entity does not seem to be wise in the long run. In particular, users from outside of the US are exposed to the daily exchange rate volatility of the US dollar relative to their domestic currencies. However, pegging to the US dollar is still a dominant strategy to bootstrap the system, as the US dollar is the most commonly used international currency.
To diversify the idiosyncratic shocks of one single sovereign currency, the anchor will transition from the US dollar to a basket of major sovereign currencies in the midterm. In the long run, it should decouple from the government-issued fiat currency and peg to the price index of a mix of a basket of consumer goods and a pack of major dApp services.
Reserve
The transaction fee charged by the system accrues to the Reserve Pool managed by a smart contract. The smart contract interacts with the Decentralized Exchange (DEX) to purchase some other crypto assets, e.g. Bitcoin. The diversified portfolio consists of multiple established crypto assets (denoted as reserve tokens) in order to reduce its risk exposure.
The Reserve Pool also plays the role of insurance in the presence of a sharp drop in AISC price. The Reserve Pool smart contract automatically swaps the reserve tokens with the AISC tokens using the DEX to reduce the supply of AISC in the market.
The Reserve Pool expands as the transaction fee accumulates. The system gains resilience with the number of reserve tokens. There is a self-fulfilling stability mechanism: as the size of the Reserve Pool grows, the stability of AISC tokens reinforces, which attracts more users and generates more transaction fee revenue, making the system even more robust to shocks.
Oracle
To maintain the stability of the exchange rate of the stable coin vis-à-vis the sovereign currency, the system requires the retrieval of exchange rate data from the off-chain centralized exchanges. It is prone to manipulation and collusion if the system relies on a trusted agent or a group of elected agents to feed the exchange rate data. To make it more decentralized and trustworthy, the system should be open and allow anyone who stakes a certain amount of AISC tokens to upload the exchange rate to the system. The feeds are aggregated and weighted by the number of AISC stake behind each exchange rate. The system takes the weighted median value as the final exchange rate. To incentivize stakeholders to upload the true value, the system rewards those whose feed is within one standard deviation of the final value. Those who fall outside of three standard deviations are slashed a certain fraction of their staked AISC tokens. The confiscated tokens accrue to the Reserve Pool.
Loan
The AI-based Stable Coin System also operates an autonomous commercial banking system on the blockchain. Users could generate AISC loans by depositing reserve tokens as collateral in the smart contract. The loan rate is adjusted by the system. In the presence of AISC appreciation, the system lowers the loan rate stepwise to bring the exchange rate back to the parity. Vice versa. When the AISC depreciates, the loan rate is lifted to reduce the supply of AISC tokens in the market. The loan interests accrue to the Reserve Pool as well.
Evolution
The loan rate and transaction fee charged by the system first follow a simple rule. With enough data collected over time, machine learning kicks in and finds out the best response function. The system will wean off the fiat-pegged mechanism when the market cap hits $100 billion and transition into a CPI-pegged mechanism. The system weans off the price oracle and adjusts the transaction fee and loan rate based on the on-chain data, e.g. transaction volume, number of active accounts, median and average values of the transactions, prices of commercial goods mapped on chain by dApps etc. The stability of the AI-based stable coin will be safeguarded by the growing Reserve Pool and the increasingly smart algorithm.
The author of this article, Yulin Liu (yulinzurich@gmail.com), currently leads research on token economics and governance system at DFINITY. He also serves as Affiliated Economics Professor at Huazhong University of Science and Technology. Yulin specialises in monetary theory, bank supervision, cryptocurrency, token economics and blockchain governance system.
He holds a Master of Science in Quantum Computation and a Ph.D. in Economics from ETH Zurich. Yulin was a visiting scholar at the European System of Central Banks and has been invited for talks at major central banks and conferences worldwide.
