WTF is Self Collateralized Stablecoin And How Does It Work?

A scalable and highly stable USD alternative

Cryptocurrencies have many advantages when it comes to speed and privacy. They are, however, also very attractive for speculation and this tends to keep their price very volatile.

People always have a need for low volatility assets for transaction and store of value purposes, stablecoins answer that need by targeting a stable value, a “peg”, against the dollar or other financial assets.

Stablecoins are going through a phase of very rapid expansion; growing from a market cap of USD 4.9 billion at the start of 2020 to about USD 60 billion today.

At the moment the most widely used stablecoins are backed by bank deposits at traditional banks. Since traditional banks are heavily regulated, this makes these stablecoins quite sensitive to government interference and regulations.

Crypto-backed stablecoins like DAI are resistant to government control but need large amounts of crypto collateral and might face problems with future growth.

Algorithmic stablecoins do not use collateral but almost all projects use incentives to stimulate the stablecoin buying and selling needed to bring the price back to the peg. If the price is below the peg, the buying incentives need to be large because they are offered when the system is under pressure.

Over time, handing out these expensive incentives can undermine the economics of the project up to the point that market participants lose confidence that any further incentives will be paid out at all.

At that point, the peg needs to be abandoned. This is exactly what happened to all incentive-based algorithmic stablecoins that have been launched so far.

This article presents a design for pure crypto stablecoin that does not undermine its own economics and is exponentially more stable as a result. This design features two new elements:

1. The protocol will not rely on voluntary action. Investors can make large amounts of money when the stablecoin is expanding but these are required to help out when buying is needed to support the price. If investors refuse to buy stablecoin when it is needed, they lose their claims to future profits.

This avoids having to convince people to put money in when the system is under pressure and makes the stablecoin a lot more stable. In effect this design allows the stablecoin to use its own future profits as collateral, a “Self-collateralized Stablecoin”.

2. The amount of buying that is needed to bring the price back to the peg is determined by looking at the actual oversupply in the market.

This article first discusses current stablecoin designs and their challenges, with special attention to USDT, DAI, ESD, FRAX, and FEI. It then explains how the Self-collateralized Stablecoin keeps its price at the peg and how it functions under stress scenarios.

For more details on existing projects and the ideas in this article can be found in the Optimal Stablecoin Design white paper.

Stablecoin market cap showing enormous growth. Source: CoinMarketCap

Challenges for Established Stablecoins

The first part of the article discusses the challenges faced by existing stablecoins. Currently, the largest stablecoins by far are fiat-collateralized stablecoins backed by traditional bank deposits like Tether (USDT), USDC and Binance USD (BUSD). Five of the six biggest stablecoins that are included in the graph above are fiat-backed. 

These fiat backed coins are not “decentralized”, this means that there is one central entity, in this case a corporate entity, that has the ultimate decision-making power over the project. No decentralization in this case also means that fiat-collateralized stablecoins need bank accounts at traditional banks. This exposes them to government interference and potential regulation. These features make fiat-collateralized stablecoins likely a poor long-term foundation for a decentralized economy. 

Crypto-collateralized stablecoins are decentralized since they don’t need bank accounts or corporate entities. This makes them a good intermediate solution, but because crypto prices are so volatile a lot of collateral is needed. With a 5% stablecoin market share and a fourth place in terms of stablecoin size, DAI is the most well-known example of a crypto-collateralized stablecoin. Creating new DAI depends on crypto holders depositing crypto collateral in the protocol in order to borrow freshly minted DAI. These borrowed DAI are then used for to facilitate levered investment or consumption. The protocol obtaining these loans on favorable terms might become a problem in the future if demand for DAI outgrows the demand for these crypto backed loans.

Algorithmic Stablecoins

Algorithmic stablecoins try to resolve the reliance on collateral described in the previous section by using the rules of the protocol itself to hold the price at the peg. When keeping the price stable an algorithmic stablecoin has to deal with two possible situations:

Too much demand: Happy times! This is when the system is in expansion, demand for new stablecoins will push the price above the peg and new stablecoins can be minted at a profit. The profit from minting new coins is called “seigniorage”, a word from the 15th century when minting physical coins in the name of a sovereign was good business. The seigniorage profit can get pretty large for algorithmic stablecoins; if no collateral needs to be put aside, the sale value of newly minted stablecoins is 100% profit.

Too much supply: Party is over. Now the system is contracting, oversupply will start pushing down the price and the excess stablecoin needs to be taken off the market somehow. Destroying stablecoin during a contraction to increase their scarcity is called “burning”. 

The first and most successful algorithmic stablecoin so far is Ampleforth (AMPL) which launched mid-2019 and currently has a market cap around USD 200 million. AMPL was based on an excellent paper by Ferdinando Ametrano “Hayek Money: The Cryptocurrency Price Stability Solution”. It features a simple and elegant design: if the price drops by a certain percentage, the same percentage of all coins are burned in each wallet. The idea behind the burning is that increased scarcity would subsequently drive the price back up. Although this mechanism is successful in keeping the AMPL price around the peg, the chart below demonstrates the price shows quite large gyrations.

Ampleforth price gyrating around the peg. Source: CoinMarketCap

Besides the variation in price, the instability in wallet balance makes AMPL less than ideal as a means of payment and store of value. If you reserve some stablecoin in your wallet to buy something in one week, you want to be sure you still have the same number of AMPL one week from now.

Seeking Wallet Stability: ESD

Other designs like Empty Set Dollar (ESD) try to solve these problems by enlisting a subgroup of active traders to buy excess supply in the market and subsequently burn these stablecoins. In return for doing this these active traders receive rewards. This reward-based system means that active traders can choose not to help in stress situations.

Enticing active traders to put fresh money into the project when the system is under pressure means large rewards need to be offered before they take action. It is very difficult to get the size of these rewards right; if they are too high, they will attract speculators, if they are too low, there might be insufficient buying to defend the peg. This can cause a dynamic where high perceived rewards attract many speculators right after a project launches. After a while, profit taking by these speculators combined with a lack of real users start to prevent structural expansion and dilute the economics. When confidence in future expansion is lost no level of rewards is enough to support the peg and the stablecoin collapses. Unfortunately, this is what happened to ESD about two months after its launch as shown in the graph below.

Empty Set Dollar price dropping after confidence in future expansion was lost. Source: CoinMarketCap

A number of similar tokens were launched last year like Basis Cash (BAC) and Mithril Cash (MITH) and Dynamic Set Dollar (DSD), unfortunately they all faced similar challenges as ESD.

Mixed Stablecoin: Frax and FEI

Mixed stablecoins combine the use of collateral with algorithmic techniques. They try to reduce the dependence on collateral as far as possible while still being able to defend the peg. As the Self-collateralized Stablecoin will also use a base level of crypto collateral we will discuss two interesting members of this stablecoin family.

Frax

The price of the Frax (FRAX) stablecoin is supported by a flexible collateral mix consisting of other stablecoins and a separate “seigniorage token” called FXS that is issued by the Frax protocol. Using a separate seigniorage token is a good idea as we will explain in more detail below. The Frax structure does seem to have a drawback as FRAX is swapped for FXS when FXS prices will tend to be low and FXS will be swapped for FRAX when FXS prices will tend to be high. That can exacerbate FXS’s price moves and could lead to FXS dilution over time. This effect might explain the Frax’s interface showing an increase in FRAX supply that does not seem to translate into a significantly lower FXS supply outstanding (*see notes below).

As Frax still relies on voluntary arbitrage possibly resulting in dilution, its required collateral ratio might be relatively high compared to the Self-collateralized Stablecoin design. If the project turns out to need a significant amount of other stablecoins as collateral this will somewhat undermine its use case as an independent decentralized stablecoin. 

FEI

Fei Protocol (FEI) is new project that has just launched. Its collateral is in ETH and all of it will be deployed in the ETH/FEI liquidity pool on the Uniswap decentralized exchange. When the Uniswap price goes below the peg the protocol uses direct trading incentives to stimulate buying as well as a fine on selling, called the “dynamic burn”. This dynamic burn rises exponentially when additional selling causes the price to drop further.

One potential issue with the project is that the project might be sensitive to price moves as it retains all its collateral in ETH and features no overcollateralization. A more structural issue might be that the dynamic burn fine will function as an exit fine: it will protect the collateral pool from selling pressure, but it can destroy stablecoin liquidity just when people need it most.

On the 4th of April, FEI launched after raising more than USD 1 billion in ETH. This was followed by large selling pressure on the stablecoin and abandonment of the peg. The most likely explanation being investors being surprised by and subsequently panicking because of the exit fine.

Summary Stablecoin Features

Below we summarize the main features of the projects discussed with the proposed Self-collateralized Stablecoin (SCS) in the bottom row. 

Self-collateralized Stablecoin: From Voluntary to Required Action

The Self-collateralized Stablecoin shields the stablecoin from the activities of active traders by issuing a separate seigniorage token. These seigniorage tokens contain all the mechanics to keep the price at the peg while the stablecoins avoid all this volatility and will be used for transaction and store of value purposes only. Seigniorage tokens can be bought in the market, like the stablecoin. This principle was first proposed by Robert Sams in an insightful 2015 write-up “A Note on Cryptocurrency Stabilisation: Seigniorage Shares” and is now being implemented by the new version of Dynamic Set Dollar (DSD). 

There is one challenge with this approach that needs to be dealt with however. As suggested in the case of FRAX discussed above, a market mechanism that involves buying seigniorage tokens when their price is likely low and selling them when their price is likely high, can result in seigniorage token dilution. This is a general problem, all existing algorithmic designs rely on the issuance of new claims on future profits to stimulate buying below the peg. These new claims can be called new seigniorage tokens, rewards, coupons or debt, but they are all claims on future seigniorage profits. Issuing these claims during a contraction, when their risk will be high and their value low, continuously dilutes the value of future seigniorage profits. As described in the white paper, Appendix I, 3.3 “The Dilution Problem”, this is very unproductive as a lower valuation of future seigniorage profits will make the peg exponentially more sensitive to being abandoned.

The Self-collateralized Stablecoin avoids the issuance of new claims and the resulting dilution. When its price goes below the peg for a certain amount of time, existing seigniorage token holders have a time window to buy and burn an amount of stablecoins. If seigniorage token holders don’t meet this “burn requirement”, their tokens will automatically be impounded by the protocol and given to token holders that do not fail to act.

Self-collateralized stablecoin response to excess supply and demand.

Threatening to impound seigniorage tokens only works as long as the tokens are always worth more than the expected cost of the stablecoins that are required to be burned. Otherwise the token holders will simply walk away and the peg would be lost. So at any time the following must hold: 

Value of future cash flows from seigniorage tokens > Expected burn cost

Later in the article we will find out that the above condition will hold in some very extreme scenarios. But first let's see how the seigniorage token profit and value is determined.

Seigniorage Token Profit

The seigniorage tokens receive all seigniorage profits resulting from new stablecoin being minted. These tokens should derive significant value from this, a value which will not be diluted as the number of tokens will never increase. We will now try to determine the seigniorage profits in different scenarios. Let’s assume we are dealing with a USD stablecoin and all numbers are in USD. 

The three variables we need to estimate seigniorage profits are the:

 1. Amount of stablecoin outstanding,
 2. Demand growth for the stablecoin,
 3. Collateral ratio applied to minting. 

In the table below we define four scenarios from a 100 million stablecoin upstart to a mature project with 100 billion outstanding. For each scenario different assumptions are made for the stablecoin growth rate and the collateral ratio applied to minting new stablecoin.

Four different scenarios for the Self-collateralized Stablecoin. 

The stablecoin growth rate is important because it directly determines the minting of new stablecoin and hence seigniorage profit. The growth rates assumed here are quite conservative. Tether grew at a more than 710% annual growth rate since passing the USD 10 million mark at the start of 2017. It grew more than 750% over the last twelve months, reaching a size of about USD 40 billion today.

As the token matures its growth rate will tend to drop. For the mature scenario we assume no further growth in market share as a cash equivalent. This means long-term demand growth for cash balances in an economy should become dependent on 1. the increase in price level (inflation) and 2. the increase in transaction volume (real economic growth). These two numbers make up the so-called nominal growth. US nominal economic growth amounted to 4% in 2019. As 2019 was a low nominal growth year, using 4% as the long-term expected growth rate for a mature stablecoin seems quite conservative.

The collateral ratio is the percentage of stablecoin outstanding that is covered by collateral value. As the stablecoin matures, the collateral ratio is assumed to drop from 80% to 20% as it builds more trust and becomes more stable. So in the mature scenario there is USD 20 billion of collateral backing 100 billion of stablecoin. 

The collateral ratio is the cost of creating a new stablecoin. During an expansion phase people want to buy stablecoin, and in a regular cryptocurrency an increase in demand would increase the price. In the case of a stablecoin the quantity is increased instead because arbitrageurs can buy newly minted stablecoin from the protocol at the peg price. The profit generated from minting these stablecoin is equal to the difference between the peg price and the value of any collateral we have to put aside. As shown in the diagram below, this means the seigniorage profit margin is equal to the peg price (100%) minus the collateral ratio.

USD 0.80 of seigniorage profit resulting from 1 Self-collateralized Stablecoin (SCS) being minted at a 20% collateral ratio. Profit distribution can either be directly in crypto or as a seigniorage token buyback.

The amount of new stablecoins that is being minted is determined by the stablecoin growth rate. Multiplying this stablecoin growth rate by the seigniorage profit margin gives the seigniorage profit rate, the seigniorage profits expressed on the stablecoin outstanding. The calculation in the table below shows that for the mature scenario this seigniorage profit rate will be 4.0% * 80.0% = 3.2%. 

As this seigniorage profit rate is expressed on the stablecoin outstanding it can be multiplied by that number to get to the seigniorage profit in USD. During our four stages of stablecoin growth the annual seigniorage profit increases from USD 50 million to 3.2 billion.

Calculation of seigniorage profit rate on stablecoin outstanding and the seigniorage profit in millions. The seigniorage profit increases when the stablecoin matures.

Seigniorage Token Valuation

To get from an annual profit to a valuation involving all future profits, we should apply the correct Price-to-Earnings (or “PE”) ratio. The S&P 500 PE ratio is not a bad road sign for the PE ratio of our mature scenario as both involve an earnings stream dependent on long-term nominal economic growth. We know the PE ratio of the S&P 500 is about 33 today, we will use a PE ratio of 15 to be conservative.

The three scenarios that occur before the mature scenario feature much higher growth so if anything they should probably be valued at a higher PE ratio than the mature scenario, comparable to how high-growth stocks trade at a relatively high PE also. Even though higher PE ratios are likely to apply, to be conservative, we will assume a PE of 15 for these scenarios as well.

The value of seigniorage tokens increases as token matures, but decreases in relation to stablecoin outstanding.

The table above applies our estimated PE ratio to the seigniorage profits of each scenario. This results in the value of the seigniorage tokens increasing from USD 714 million to 45.7 billion as the stablecoin grows to maturity. Note that although the absolute valuation grows as the stablecoin matures, the seigniorage token value drops in relation to the stablecoin outstanding (from 750% to 48%).

Now we can answer the key question: if there is a real challenging contraction, will seigniorage token holders choose to support the peg instead of giving up their tokens?

Stress Analysis

If the market slowly realizes a shift in stablecoin demand is occurring or even if the realization happens in a few distinct phases, the seigniorage token holders are likely to support the peg as each additional commitment will be quite small compared to their token value. The worst-case scenario for the Self-collateralized Stablecoin is an immediate, permanent drop in demand. Think about a number of big holders stating they are dumping their holdings and stating they never want to see our stablecoin again.

In this section we will look at the effect of an immediate and permanent 25% drop in stablecoin demand. The results will only be presented briefly, more detail on assumptions and mechanics can be found in the Optimal Stablecoin Design white paper, Section 3.3. “Stress Analysis”.

A 25% drop in stablecoin demand will result in the seigniorage token value dropping by the same percentage. At the same time the token holders will incur a burn requirement equal to 25% of the initial stablecoin outstanding. The cost of this burn requirement is partially mitigated by the purchase price of the stablecoin which will be lower than the peg. Another factor reducing the burn cost is the collateral, as the collateral backing any burned stablecoin can be distributed to the seigniorage token holders immediately. The table below presents the results.

Scenario results. All percentages in the table are expressed on the initial stablecoin outstanding (except the trading level and the collateral ratio).

The calculation shows that, across all scenarios, the value of the decision to burn remains positive by a wide margin. Depending on the scenario seigniorage token holders can save up to 47.9% of their token value by meeting the burn requirement.

Note that the mature scenario is the most risky with the lowest relative seigniorage value and the highest relative burn cost. As described by Haseeb Qureshi in his 2018 article “Stablecoins: designing a price-stable cryptocurrency”, an algorithmic stablecoin will always need some positive long term expected growth rate to function, and the lower the expected growth rate, the higher the risks.

Using the Collateral Asset

In a crisis situation with a lot of sudden stablecoin supply, all seigniorage holders need to be sure the other holders will also support the stablecoin. If a lot of seigniorage token holders refuse help and the stablecoin fails, the ones that don’t would lose both the burn cost they incurred as well as their seigniorage token value. This means there always needs to be a healthy buffer between the remaining value of the seigniorage tokens and the burn investment, especially in a panic situation.

Up till now we have left the remaining collateral out of our analysis. The seigniorage tokens will have voting based governance rights over the protocol and hence control the collateral. This means any distributions to the token holders from the lowering of the collateral ratio over time and from returns generated by the collateral should be added to the seigniorage token valuation. This will add significant value to the seigniorage tokens, especially when the collateral ratio is still high.

In the mature scenario the collateral ratio can not be expected to drop much further, but even then a 20% collateral ratio is available as a last line of defense. In case of an emergency, a governance vote can deploy this collateral to help seigniorage token holders buy and burn stablecoins. If the seigniorage token itself trades at an irrationally low level, some of the collateral could be profitably deployed by providing a floor under its valuation.

Analysis presented in the Optimal Stablecoin Design white paper (Appendix II “Scenario Analysis”) shows that by deploying half of its remaining collateral, our slow growth mature stablecoin should be able to survive a 35% immediate and permanent demand drop with a very healthy buffer between the value of the seigniorage tokens and the required burn investment.

Determining the Intervention Amount

After the price is below the peg for a certain amount of time the protocol will have to determine the burn requirement. The Self-collateralized Stablecoin includes a new method to determine this number. Algorithmic stablecoins have so far relied on the assumption that after a given percentage drop, the exact same percentage of stablecoin should be burned to get the price back up. So, a 5% drop in price results in a 5% burn target to get the price back up to the previous level.

This assumes that just before the burning begins, all participants are satisfied with the value of stablecoin they hold and want to keep that value constant. If this assumption holds and the quantity is reduced by 5%, people will start buying till the price goes up by 5%, so all holders can get back to holding the previous stablecoin value that they were happy with. Unfortunately, stablecoin prices are determined in a sometimes very fast-moving market with the involvement of a minority of holders. We know many non-active holders are not initially involved in price moves and seem to be slowly updating their desired stablecoin holdings according to how the situation evolves. This makes the assumption that all holders are satisfied with their stablecoin value just before the burning starts highly dubious. If that assumption cannot be counted on, responding to a percentage price move by burning the same percentage of stablecoins becomes a pretty random thing to do.

What all algorithmic projects are really looking for is the amount of buying that is needed to push the market back to the peg. This amount is equal to the sum of all amounts that market participants are willing to sell below that price. We will approximate this excess supply by adding all the sell orders on all (verifiable) on-chain limit order books that are offered at prices below the peg. To this, we will add the amounts that would be needed to push on-chain automated market maker liquidity pools like Uniswap back to the peg. This total amount will be the required burn amount that the seigniorage token holders will have to burn within a certain time period.

The calculation below shows an example with only one limit order book and one automated market maker liquidity pool. Both books feature the Self-collateralized Stablecoin (SCS) versus USDC, for simplicity’s sake we are assuming USDC trades at exactly one Dollar.

Limit orders below the peg and disbalance in automated market maker books are added up to calculate the required burn amount. Note that as long as the USD price of the other crypto is known, any price can be converted into an implied stablecoin price vs the dollar.

Because the protocol will not see sell orders on centralized exchange order books, adding up the excess supply from on-chain trading books is very likely to undershoot the actual quantity needed to bring the price back to peg. Even though this means there is a high likelihood that the protocol will not reach the peg after our initial response, we do know that the response will be proportional to the pressure in the market. What is even more important, the response will almost never overshoot. If the response overshoots the peg, stablecoin volatility will be higher, and it might even lead to progressively larger swings if the level of overshooting exceeds the initial deviation. In our case, after undershooting, the intervention exercise can simply be repeated until the price reaches the peg: snapshot on-chain trading books, calculate burn requirement, communicate burn requirement to token holders, wait a bit, impound tokens from holders that don’t burn.

Other Features

It is important to ensure that seigniorage token holders actually buy the tokens they are supposed to burn. Otherwise, they might deliver at least their first allotment stablecoins from a standby wallet instead of buying them in the market. Buying is stimulated by paying a fee to token holders that buy their burn requirement on verifiable on-chain exchanges. These “verified buyers” can also receive the impounded tokens from those that “fail to burn”.

The Optimal Stablecoin Design white paper also contains some proposals to make the stablecoin popular for transaction and store of value purposes. Activities like yield farming that attract speculators should be avoided, but the protocol could use some of the large initial seigniorage profits to implement a blockchain transaction cost rebate for example. The stablecoin could also distribute a multiple of the consumer inflation rate to stablecoin holders during its initial growth phase.

Potentially Attractive Investment

An ICO of the seigniorage token could be a very attractive proposition for investors. They get all the rights to the future seigniorage profits, which might represent an excellent investment if the stablecoin becomes a valid monetary medium. At the same time the initial collateral the seigniorage token holders invested remains under their governance control. The investors could also receive a significant amount of initially extremely well collateralized stablecoins that they can spend.

Conclusion

Similar to how the legacy USD is collateralized with expected government revenues, it is possible to collateralize a stablecoin on its own expected use value in a growing economy. This might inspire some technically skilled people to use these ideas to build a highly stable, decentralized stablecoin that can truly function at scale.

The strengths and weaknesses of existing projects and the ideas in this article are discussed in more detail in the Optimal Stablecoin Design white paper. 

Don’t hesitate to get in touch for further discussion.

----------------- Notes -----------------

* The Frax’s interface shows that despite net FRAX supply increasing (left hand panel below), the total FXS supply seems to remain stubbornly close to the original issuance amount of 100 million (right hand panel below). This is the case even with the collateral ratio having dropped from the initial 100% to 86% now.

FRAX supply (left), Frax Seigniorage token FXS supply (right). Source: FRAX Finance

Let's have a look at some numbers from the period from the 1st of February onwards when the amounts outstanding seem to be more stable and the collateral ratio stays at 86%.

Despite a large increase in the FRAX outstanding, the FXS supply hardly moves while it should be reducing significantly due to the FRAX creation process. This could be an indication of FXS dilution.