2018 is the year of the stablecoin. (If you’re not familiar with that term, start by reading this excellent summary by my friend Haseeb Qureshi.) A handful of stablecoins have raised significant funding recently, giving the impression that the stablecoin problem must be on track to being solved. Surely at least one of the projects that have collectively raised hundreds of millions of dollars, or one of the dozens that are still in hiding, will produce an effective solution… right?
We think this impression is an illusion. Many have forgotten past prominent stablecoin attempts, which have failed. (Some consider Tether a failure given the apparent lack of trust it has garnered in the ecosystem, and we do speculate that the market cap of Tether would be substantially higher if it were a more trusted solution, but it’s obviously the best and most successful stablecoin that has been made and distributed so far!) The reason we’ve gone through all the effort to design another stablecoin protocol is that we are convinced that all of the existing attempts are likely to fail as well. Leaders of the crypto ecosystem have been aware of the need for a stablecoin for years. The financial incentive to create the winning stablecoin is immense. Economic protocol design of this type is hard — if it were easy, it would have been solved years ago.
Some say that these new projects will update their plans, copy one another’s work, and all land on functional protocols before launching. We think this is overly optimistic. Since economic protocol design is hard, a team without the knowledge to design a protocol free of flaws over the course of several months may simply be incapable of grasping all that needs to be understood in order to make all the relevant final design decisions and arrive at something functional. If a team had designed a flawed consensus mechanism for a new smart contract platform but had believed it was solid enough to raise millions of dollars on, would you trust them to fix the flaws later? And even if these teams have the knowledge needed to succeed, in many cases they will find themselves in a mess when it turns out the token they’ve pre-sold to dozens or hundreds of investors is not all that similar to the token they would ultimately need to build, requiring them to violate or renegotiate legal and social contracts in order to pivot.
In order to correctly assess which stablecoin designs have the potential to work — which the industry really needs to be able to do right now — you’ll need to understand some important principles, and be willing to spend considerable time applying them.
Exchange Rate Pegs
Every stablecoin design proposed to date is implementing an exchange rate peg, which is when a currency’s exchange rate with another currency is stabilized. It’s important to understand the fundamentals of exchange rate pegs in order to assess stablecoin designs.
Middle-income countries often peg their currencies to more prominent currencies in order to maintain a stable purchasing power. So in a sense, stablecoins are just an old technology being ported to a new domain.
To understand currency pegs, you need three background concepts: buy walls, sell walls, and arbitrage.
A buy wall is when a market participant is willing to buy a very large amount of an asset for a single price. For example, suppose Alice were willing to buy up to 1,000,000 BTC on GDAX for the current market price, say $10,000 each — that is, she had posted a limit order for 1,000,000 BTC at $10,000, where the most recent trade had also been at $10,000. Now suppose Bob wanted to sell 1,000 BTC on GDAX at the best rate he could get. He posts a market sell order for 1,000 BTC. Alice and Bob would be matched, and Alice would buy all of Bob’s 1,000 BTC for $10,000 each. Alice would still have 999,000 BTC left up to purchase at $10,000 each. If Carol then wanted to sell 1,000 BTC at the best available price, she would also be matched with Alice, and Alice would now have 998,000 left for sale. The market price — the price of the most recent trade — would remain at $10,000. Until market participants had sold 1,000,000 total BTC to Alice, the market price would not be able to go down. A buy wall gets its name from the shape it produces in a market depth chart:
A sell wall is the same as a buy wall, but in reverse. Instead of stopping the market price from going downward, it stops it from going upward:
Arbitrage is a simple but sometimes misunderstood concept. It doesn’t just mean buying low and selling high; it means buying low and selling high at the same time. Suppose David can buy 1 BTC for $10,000 on GDAX, and simultaneously sell 1 BTC for $10,100 on Gemini. He finishes with the same position in BTC, and has earned $100. His purchase on GDAX will raise the market price (so long as there isn’t a sell wall) and lower the price on Gemini (so long as there isn’t a buy wall). This simple action, which comes with a strong economic incentive, keeps the prices on all BTC exchanges relatively uniform.
If what I just said didn’t make sense, you’ll want to google around a bit, spend some time thinking, re-read it a few times, ask your friends, or whatever else it takes until you can read through it and have the “yep, makes sense” experience, before the rest of this will add up. Please take the time to do this! We need more people who are equipped to think clearly about stablecoins :)
In order to peg currency A to currency B, a party only needs to do two things: (1) maintain a buy wall in the market between currency A and currency B on a single exchange, and (2) maintain a sell wall in the market between currency A and currency B on that exchange. These buy and sell walls keep the market price on the exchange fixed, so long as the party doesn’t run out of currency A or currency B. If other exchanges have different market prices at any given point in time, arbitrageurs will simultaneously buy on one exchange and sell on the other in order to earn the risk-free profit. This brings the market price of the other exchanges closer to the exchange the party is maintaining the buy and sell walls on, but it doesn’t move the price on the exchange with the buy and sell walls, since they stop it from moving. As a result, the markets between currency A and currency B are brought to the same price as the exchange with the buy and sell walls.
Now we can ask: When do pegs break? The answer is simple: When the party runs out of currency to maintain a buy or sell wall. Once there is no wall on either side, the market price can move in that direction, just like it does with BTC.
When a nation pegs its currency to another currency, it does so by (a) holding foreign currency in reserve and using it to maintain a buy wall for its domestic currency at a particular price, and (b) minting as much of its domestic currency as needed to maintain a sell wall on it for that same price, against whichever foreign currency it’s pegging to. (b) is the easy part, since a central bank can mint as much of its domestic currency as it wants. (a) is the hard part — when it runs out of foreign currency, or decides not to spend any more of its foreign currencies in reserve to defend the peg, it can’t maintain the buy wall anymore, and this is what leads to pegs breaking.
This means that there are only two features of a traditional exchange rate peg one needs to understand in order to predict whether it will hold:
- The total value of foreign currencies the party holds in reserve
- Whether the party will choose to spend these currencies maintaining a buy wall
One more important nuance is that in order to peg currency A to currency B, a party doesn’t necessarily need to hold currency B in reserve. It can hold any valuable asset that has liquid and accessible markets. For example, a nation could hold gold in order to peg its currency to USD. So long as it’s always willing to pay one dollar worth of gold for one unit of its domestic currency (at the present USD market price of gold), it will be able to stabilize the price of its currency at $1.
Again, I encourage you to do what it takes to be able to read the previous section and feel totally comfortable with every claim. The simple two-piece framework above is crucial for thinking about any new stablecoin proposal and has helped the Reserve team analyze dozens of potential designs we’ve invented and encountered.
The Three Common Stablecoin Approaches
Again, every existing stablecoin proposal is implementing an exchange rate peg, typically to USD. The work that goes into stabilizing the purchasing power of a floating currency is significantly more complicated, and likely not fit for automation in smart contracts any time soon, if at all.
In monetary policy, there is a long-running debate on the value of “rules versus discretion” — whether it’s better to allow humans to make discretionary choices in how to manage the money supply, or to codify fixed rules in advance and follow them closely. Nations like the US that modify purchasing power via manipulating interest rates (and in other ways) need to make a complicated set of decisions that often require discretion. Nations that peg their currencies to other major currencies often have what’s called a “currency board,” which just maintains reserves and uses them to buy and sell their local currency — a process which is much more easily codified and so can be automated.
When considering whether any given stablecoin will work, it’s very helpful to translate its design into these concepts. Let’s look at the three main approaches to stablecoins that have been proposed so far, and translate them into the language of monetary economics. For simplicity let’s assume all stablecoins are intending to peg one-to-one to the US dollar.
Traditional Asset-Backed Stablecoins
Most prominent example: Tether
A traditional asset-backed stablecoin is simple to describe, and only gets complicated in practice. The issuer sells tokens for $1 each, and holds all of the dollars from those sales in reserve in a bank account. Any time someone wants to, they can redeem tokens for dollars with the issuer. This design is best with a trustworthy audit mechanism, like Trust Token is using for TrueUSD, in order to reassure holders that the money is there.
This means that the issuer can maintain a buy wall equal in size to 100% of the circulating supply of the stablecoin at all times. If it’s a trustworthy actor and does in fact maintain that buy wall, it can implement a perfect currency peg. This system is a lot like Hong Kong’s currency board, which has classically maintained very high USD reserves, and had no trouble weathering the storm in the Asian Financial Crisis we referenced earlier, in which many exchange rate pegs broke.
But in practice there are two central problems with traditional asset-backed stablecoins:
- Counterparty risk — What if the issuer chooses not to maintain the buy wall at some point? We think the best traditional asset-backed stablecoins will be issued by trusted banks and perhaps trusted governments, both of which can solve this problem fairly well. But even banks and governments aren’t trusted everywhere in the world.
- Risk of government intervention — What if the local government decides to freeze the assets, or a powerful foreign government decides to put pressure on the issuer? Governments have well-developed capacity to prevent money laundering, which has been used to shut down multiple virtual currencies in the past. In practice this may not reach the stage of asset seizure, and may just lead to the issuer losing banking relationships again and again, since banks want to stay in the good graces of their local governments.
These two core problems create a fundamental tension:
- The better the audit mechanism, the more credible the promise to use reserves to maintain the buy wall and thus defend the exchange rate peg, but the easier the job of the government to locate and freeze the assets.
- The worse the audit mechanism (all the way to just not providing any audits and keeping banking relationships secret, or not even telling the banks they are holding assets used by a crypto company), the harder the job of the government to locate and freeze the assets, but the harder it is for the market to assess the issuer’s solvency and ability to maintain the buy wall.
If users all have to go through a KYC process, this significantly limits the viability of the token to permit free movement of money across borders.
This gives us the following breakdown of how traditional asset-backed stablecoins can play out:
We don’t see either of these outcomes becoming the new global reserve currency. The non-trustworthy option is clearly out. The trustworthy option doesn’t serve what has been the main use case for BTC other than speculation: free transfer of money across all borders. In order to supplant the US dollar and become the new reserve currency, we think free transfer across borders is essential.
Collateralized Debt Stablecoins
Most prominent example: Maker DAO
A collateralized debt stablecoin offers asset holders a way to take out loans backed by their cryptoassets, which then become the collateral in the system. Users can deposit a cryptoasset into a smart contract, which then mints a new stablecoin for them. The stablecoin must be of lesser value than the collateral, so let’s say the user deposits $2 worth of ETH and receives a stablecoin that has a target price of $1. The user can now go spend that stablecoin on goods and services, without having to give up their position in ETH. When they want their ETH back in their control to spend or trade, they have to repay their stablecoin loan to withdraw it. A common use case for this is going margin-long on cryptoassets; the user can sell the $1 stablecoin for ETH, and then is holding $3 worth of ETH instead of just $2 worth. If ETH appreciates, they’ll earn more, and if it depreciates, they’ll lose more.
Even if you understood what you just read, which will take you a few minutes if you aren’t familiar with this particular space, a good next question to ask is: Where’s the exchange rate peg in all of this? In the Maker system, there are three answers.
First, and easier to understand, is that if the price of all of the locked up collateral starts to go down too fast, or something else wonky happens, a group of people with special rights can choose to wind down the system. If they do this, all of the stablecoin holders can trade their coins for $1 worth of the collateral cryptoassets. (The system can then start up again, it isn’t gone forever.) Since this promise exists, in theory, speculators will want to buy the stablecoin if it is ever below $1 in order to be able to either (a) sell it for $1 when the secondary market price goes back up, or (b) redeem it for $1 of collateral assets in this wind-down scenario. In theory, these speculators will provide a buy wall at some discounted price slightly below $1, and will maintain the peg so long as they have capital to do so.
The problem with this mechanism is that it will be difficult to tell how much capital is available in the hands of speculators who want to take this bet before they have actually taken it. So the implied reserve to implement the buy wall may or may not be large, and may change on a dime based on market sentiment and events that require or incentivize speculators to move their capital elsewhere. After all, they are not a dutiful currency board charged with maintaining the exchange rate peg, they are speculators looking to earn a profit, and this bet requires allocating capital for what may be a long time.
Second, if the collateral that’s locked up falls in value enough, the system will begin automatically closing out collateralized loans. It does this by re-purchasing stablecoin tokens in an auction using some of the collateral, destroying the stablecoin tokens, and releasing the remaining collateral back to the original debtor.
However, the protocol must do this at whatever the market price for the stablecoin is, so this will provide indiscriminate upward pressure on the stablecoin price, not a buy wall at the target price. Assuming the collateral doesn’t depreciate very quickly (which could be a problem), the main issue with this mechanism is that it reduces the supply and increases the market price of the stablecoin at exactly the time when demand for stablecoins is likely to be rising — when other cryptoasset prices are falling. This means the market and the system will both be driving the price upwards at the same time, not bringing it to a target price.
Third, and somewhat more subtle than the first two, is that whenever a debtor sells their stablecoins after taking out their loan, they have an incentive to repurchase the stablecoins and repay the loan any time they can do so for less than they sold them for. Suppose you deposit $2 of ETH, receive 1 stablecoin, and sell it for $1. Then the market price of the stablecoin drops to $0.95. Now you can repurchase the stablecoin at a discount, close your loan (which destroys the stablecoin), and receive your $2 in ETH back, earning $0.05 in free profit. Every debtor has this incentive. So as long as all debtors have sold their stablecoins for $1, this provides a nice buy wall at just under $1!
The problem with this third mechanism is that not all debtors will have opened their collateralized loan and sold their stablecoin for $1. If they have sold them for a different price, say $0.95 or $1.05, they will have the same incentive, but relative to that other price. So rather than producing a perfect buy wall at $1, this mechanism will produce a cloud of demand at various prices around $1. This is pretty cool, but not exactly what you want in order to maintain a stable peg. If things start fluctuating and more people are taking out loans and selling their stablecoins at prices that are further from $1, we worry that this could lead to even more fluctuation, and self-enforcing negative feedback loops are something to be avoided in a stablecoin protocol.
Maker’s design has the virtue of decentralization, and thus censorship resistance (apart from the small group of people who decide when to trigger the system winding down, which could perhaps be distributed amongst more token holders). But it doesn’t implement a strong and predictable currency peg. So we predict that users seeking stability will be unimpressed by the fluctuating price history that will likely develop and look elsewhere, so an approach like this will also be unlikely to yield the next reserve currency for the world.
Future Growth-Backed Stablecoins
Most prominent example: Basis
A future growth-backed stablecoin offers speculators portions of the future growth in stablecoin market cap in exchange for providing the capital to peg a currency as needed. This design was originally proposed in an informal paper by Robert Sams a few years ago. In the original design, there are two tokens, stablecoins and shares. When the price of the stablecoin is above the target price, the system mints more stablecoins and offers them in an auction. The currency used to buy stablecoins in the auction is the share token — so only share token holders can participate, and the highest bidders are the recipients of the newly minted stablecoins. The increase in stablecoin supply presumably reduces the market price back down to the target. When the price of the stablecoin is below the target price, the reverse happens — the system mints new shares, and auctions them off for stablecoins. By doing this, the system can reduce the supply of stablecoins and bring the price back up.
Why would people buy shares? The idea is that you can trade your stablecoins for N shares now, and then trade those shares back for >N stablecoins later when the system needs to increase supply to keep up with a new big wave of demand.
This is an interesting design in that there is no collateral from outside the system — it’s all self-contained. So then we must ask: If it’s implementing an exchange rate peg, where is the reserve? How do we use our framework to analyze the reliability of the exchange rate peg?
We can translate it into the terms of our framework in order to do that. The reason we have to look at the size of the reserve is to see how much the supply of the pegged currency can be reduced in order to respond to a decrease in demand. In this system, the level of supply reduction that can be achieved is simply the number of dollars worth of shares that can be minted and sold for stablecoins at any given point in time. Let’s call this the “implied reserve.”
The implied reserve is made up of assets that are held by speculators. You can visualize the big crowd of people and trading bots out there, ready to buy shares in an auction under the right conditions; however much money they would be willing to fork over at any given moment is equivalent to the amount of money that a nation implementing a peg is willing to fork over from its reserves. This is what makes up the buy wall — any time the price of the stablecoin is less than $1, the system starts offering shares, and in order to buy shares you need stablecoins, so the speculators all buy coins on the secondary market at a price just below $1.
The central problem with this design is that if at any point those speculators lose interest in purchasing shares, the peg breaks, since no stablecoins can be taken out of circulation. This is analogous to a nation holding reserves and deciding not to spend them to defend the peg for some reason. If growth in the market cap of the stablecoin is perceived to be highly probable by the market, this would be unlikely to happen, since it would be clear to the market that there is always money to be made by purchasing shares at some price. But in the early stages of adoption, growth in the market cap of the stablecoin may sometimes not be perceived as highly probable.
Given this problem, a future growth-backed stablecoin has to solve the bootstrapping problem of getting adopted enough that further growth appears nearly inevitable. If it does, we believe it can reach a state of stability, and if it doesn’t, it can fail, perhaps catastrophically. I mentioned that Basis is the most prominent design of this type, and we have concerns about this type of failure with the Basis protocol.
If a future growth-backed stablecoin does reach this state of stable equilibrium, it has the benefit of being totally decentralized and thus censorship resistant, and able to scale up easily in response to increasing stablecoin demand. But because of the bootstrapping difficulty and risk of catastrophic failure, most designs of this type we’ve looked at don’t appear to be safe approaches to building a world-wide cryptocurrency.
Most prominent example: Havven
This design works by charging a transaction fee to stablecoin spenders, and using the money from that revenue stream to reward those that put up capital to back the stablecoin when supply needs to be decreased. I’d like to walk through this architecture in more detail but unfortunately don’t have time before publishing this post. We’ve looked at several designs like this and thought through a few more ourselves, and find it to be a potentially interesting avenue, if done right. Like all of the other approaches, there are many ways to do it wrong.
The Incredible Responsibility of Stablecoin Producers
Creating a stablecoin is very different from creating a normal crypto token. With a normal token, the standard expectation at this point is high volatility in market price. That’s not the case for a stablecoin, at least once it has established an initial track record of functionality. Users will expect the price to stay the same, and will use the token to store value that they can’t afford to lose. Indeed, the people in the world who I believe will benefit most from a stable full-stack open currency are those with no access to any stable store of value right now, who often don’t have a lot of money to start with. If a stablecoin reaches prominence among this demographic and then crashes, it will have done great harm to some of the most vulnerable people on the planet. While creating speculative assets that induce some retail investors to lose money is already problematic, this would be a lot worse.
I believe stable cryptocurrencies will be the first introduction to crypto for millions or billions of people. If the first experience is a massive violation of trust and the creation of unnecessary economic hardship, this could be a major setback instead of a leap forward for the open currency movement. It could make it much harder for legitimate cryptocurrencies to take hold, and could give governments the political will to massively hamper innovation in the crypto space.
This is why I think it’s crucial for us as an industry to enter this new era with caution. Stablecoins are a hot topic right now, and are on track to raise a lot of money in 2018. To further compound this issue, in many cases investors and issuers could earn a profit from a stablecoin project reaching prominence even if it breaks later, so long as they liquidate at the right time. This misalignment of incentives means that any stablecoin project founder will have to make choices that are not in their own economic interest in order to act responsibly. I’ve lost sleep over this myself.
All of this is what has motivated the Reserve team to begin putting resources towards educating the industry more broadly. We’re just at the beginning of this educational process. If you would like to get involved in this effort, we are looking to bring on conference organizers, online forum moderators, writers, and careful thinkers to help us build the open currency movement the right way. If you found this post engaging, would like to learn more about how to think about stablecoins, and are interested in helping us communicate these ideas to the broader ecosystem, please get in touch.