First published on May 14, 2018
Thanks to Myles Snyder from Multicoin Capital for providing feedback on this post.
Basis is an algorithmic stablecoin system under development. It was previously known as “Basecoin.”
Right now, it’s one of the most notable ‘Seigniorage Shares’-type stablecoins, and together with Maker DAO (read our analysis of the Maker protocol here), one of the upcoming stablecoins that is receiving the most attention. This attention is well-deserved as the Basis team has made important contributions to stablecoin developments. They’ve brought cryptoassets as a whole closer to becoming real currencies by successfully popularizing the idea that you can’t have real currencies without stability.
Before Basis, the stablecoin landscape was stagnant. There was little funding and there were no promising projects other than Maker DAO. Today, new projects pop up every week — some even inspired by the Basis protocol.
Basis took Robert Sams’ Seigniorage Shares concept and developed it into a simple and elegant decentralized stablecoin protocol, which they very skillfully communicated to the public.
Unfortunately, despite an honest and pioneering effort, I expect Basis to fail. They have taken many steps in the right direction, but ultimately they will not succeed in creating a reliable stablecoin.
The reason why is relatively simple to state, but takes in-depth analysis to understand. I’ll state it here first in bullet-points and spend the rest of the post describing precisely how Basis works and how its design appears to be irredeemably flawed. At the bottom you can read about my motivations for writing this piece.
The Basis protocol is described very clearly in their white paper. If you haven’t read it yet, you’ll want to do that before reading the rest of this post. I’ll re-cap the protocol behavior here to get us started:
The Basis protocol functions via the use of three different tokens — ‘Basis’ (the stablecoin), ‘bond tokens’ and ‘share tokens.’
Basis tokens are stabilized tokens pegged at $1. Bonds are tokens that get converted to Basis tokens when new Basis tokens are minted. Shares act like equity in the system: shareholders get newly minted Basis tokens, if and only ifall outstanding bonds have already been converted to Basis tokens and more Basis tokens need to be minted.
The Basis token itself is stabilized through on-chain supply management. What this means is that the market price of Basis is monitored by smart contracts through an oracle and when the market price differs from $1, this contract tries to bring it back to $1 by increasing or decreasing supply.
If the price of Basis is below $1, then the contract needs to decrease Basis supply to restore market price. Supply is decreased by the contract acquiring Basis that it then burns. To acquire Basis tokens, the contract mints and sells bonds in an auction (with a 10 cent minimum bond price). Basis tokens are acquired in exchange for these bonds and acquired Basis tokens are then burned. The supply reduction restores the $1 price.
If the price is above $1, then the system needs to increase supply to decrease market price. Supply is increased by minting new Basis tokens. These newly minted tokens are used to pay off outstanding bonds on a first-in-first-out basis. Repaid bonds are burned. In return for a bond, the holder gets 1 Basis token. Any remaining newly minted Basis tokens are distributed among shareholders.
Contrary to bonds, shares are not burned when paid. Also, bonds expire after five years, but shares never do. Expired bonds do not pay out Basis tokens.
Why would people buy the bonds? People buy bonds to profit from future Basis minting. Each bond pays out 1 Basis token, which is worth at least $1 when minted, since tokens are only minted when the market price is above the peg. If you buy bonds for, for example, 75 cents, you make a 25%+ profit. Buying bonds is worthwhile as long as the time-discounted profit at payout outweighs the risk that the bond will expire before paying out.
I’ve learned from the stablecoin community that Basis plans to use off-chain funds — money that is not trustlessly controlled by smart contracts on the blockchain — to keep their stablecoin pegged. These funds will be held and controlled by the company behind Basis.
The Basis team plans to get off-chain funds to stabilize their token from two sources. The first source is their fundraise. They will take [source] a portion of the money raised and use it to set up large initial off-chain reserves, controlled by the Basis company, called Intangible Labs. This will work initially. Funds raised will be sufficient to protect the peg early on, when the market cap is small. However, this protection is not scalable. Should Basis ever grow to even a moderate market cap, the crowdsale funds will no longer be enough to protect the peg. In the long run they will need more funds than they have in order to protect the peg.
This brings us to the second planned source of funding: proceeds from Basis shares owned by the company. It is known in the stablecoin community that the Basis team plans to keep a large amount of Basis shares, possibly even as high as 80%, and use them to get more funds as the Basis token’s supply initially grows.
As we saw above, the system grows by minting Basis tokens, which it gives to shareholders whenever all outstanding bonds have been paid. The company can sell off the new Basis tokens they receive for fiat money, and store that money in bank accounts, the way Tether does. Owning a large portion of Basis shares give Basis a scalable source of off-chain reserves. If they have 80% of the shares, they’ll get up to 80% of newly minted Basis tokens and can use them to maintain up to 80% off-chain reserve backing. The team will also have the option of selling off more shares once the platform is live to raise additional capital, though this would reduce their income stream as the network begins to scale up.
Off-chain reserves can be used to stabilize Basis in several ways. For example, they can put large buy orders right below the peg (buy-walls) and large sell orders right above it (sell-walls) on exchanges. Alternatively, Basis can fund automatic bots that defend the peg in a more dynamic way.
The Basis team will likely use their fiat reserves to defend the peg if their protocol’s stabilization mechanisms fail, regardless of whether they do it manually or automatically through bots. If the peg breaks, or is about to break, they will buy up Basis tokens and stabilize the Basis price at the peg.
In what follows, I’ll go over the 3 ways in which the Basis system is irredeemably flawed: (1) how the off-chain stabilization system is not a sustainable solution, (2) the fragility of Basis’ algorithmic stabilization system, and (3) its illegible and volatile backing.
There is no reason to have an off-chain stabilization mechanism unless you expect that your on-chain stabilization mechanism won’t work. This may very well be the Basis team’s motivation. Perhaps they share the concerns I express in the next section about their on-chain stabilization mechanism’s inability to keep its peg.
Regrettably, the workaround of off-chain stabilization undermines Basis’ ability to become a widely adopted decentralized currency. To the extent that the protocol relies on off-chain funds, it is no different from Tether, and runs into the same problems as Tether and other centralized digital currencies relying on off-chain funds.
There are three reasons why off-chain stabilization is not sustainable in the long run.
First, like Tether, Basis will have a hard time finding a bank that’s willing to hold their money. There is ample evidence that Tether has had a hard time maintaining banking relationships.
Second, given that centralized coins require users to trust the issuer, it incurs counterparty risk. The issuer can steal, misuse, misallocate, or just plain lose off-chain assets.
Third, centralized coins risk getting shut down by governments. The US government alone has shut down more than half a dozen centralized digital currencies.
The Basis team knows this. They too criticise Tether for counterparty risk and risk of getting shut down by governments. They wrote as much in their white paper: “a centralized solution like Tether will likely never work in the long term, and sentiment toward Tether has been extremely negative as a result.” (p.5)
Why would the Basis team want to rely on a stablecoin model they know is unsustainable? I suspect their plan is to use off-chain reserves in the short run and then hope that their algorithmic stabilization will be able to hold the peg in the mid and long run. Unfortunately this too will not work, for reasons described below.
To the extent that Basis’ on-chain stabilization does not work, Basis is another Tether, except with 80% or less reserve backing rather than 100%.
Basis’ algorithmic stabilization requires unbroken market confidence to reliably handle drops in demand for the Basis token. This is not realistic. Even in a best-case scenario where a cryptocurrency succeeds at reaching large-scale adoption, it has to survive confidence drops on the way there.
Temporary losses of confidence in Basis’ future growth undermine Basis’ ability to defend the peg. Basis defends against these price drops by issuing bonds. The expected future value of these bonds is determined by the expectation of future growth in supply of Basis tokens. Reduced confidence in this future growth thus reduces the bond value.
To illustrate, consider the effects of a demand drop when confidence is low. A demand drop brings down the Basis price. The system starts selling bonds to take Basis tokens out of circulation and bring the price back up. Initial bond sell-offs build up the bond queue. As the bond queue grows longer, the risk of bonds expiring before being paid out increases. This increased risk decreases each new bond’s value. The expected payout time for new bonds also gets longer. Time discounting makes bonds that pay out later less valuable, independently of the risk that the bonds will not pay out. So the more bonds are sold to keep the peg, the faster their value sinks. When bonds’ expected value drops below the 10-cent floor, the system stops selling bonds.
No bond sales means the system cannot further reduce Basis supply to stabilize the peg. The peg then breaks. The peg breaking is, of course, strong evidence against Basis’ viability and likely reduces confidence even further.
In this sense Basis’ fragility is worse than Maker DAO’s sensitivity to black swans. A black swan is an event that is not just unlikely but so unlikely it lies outside the bounds of our normal expectations. A black swan may be considered absurdly unlikely prior to happening. The Basis system, in contrast, can be taken down by ordinary dips in confidence.
The Basis team knows that their protocol is vulnerable to loss of market confidence in this way. They agree that a confidence crisis will eventually break their peg, but argue that it will not be not fatal. They believe their 5-year bond expiration period will help the peg recover eventually:
The protocol is designed such that when a crisis of confidence occurs, once bond prices drop below a threshold that we call the bond price floor, the protocol would simply stop creating new bonds. When [bond prices drop below the floor], we would expect Basis price to dip below $1 for a period of time because demand has dropped but bonds are no longer being created. However, having a bond expiration means that old bonds gradually expire. When old bonds expire, the bond queue gets shorter. At some point, and usually quite rapidly, we would expect the shorter bond queue to cause bond prices to rise above the floor again, helping to restore Basis price back to $1.
The suggestion is that it is not problematic for Basis’ peg to stay broken for some time, until enough bonds expire that the peg is restored. Then business can continue as usual.
However, bond expiration takes a lot of time, and Basis cannot afford to stay de-pegged for extended time periods. With a 5-year Bond expiration period, Bond expiration happens on a scale of years. That’s not going to restore the peg soon after it breaks.
This means that the real matter is how bad it is to stay de-pegged for periods of years at a time. Breaking the peg does lasting damage to a stablecoin in two ways: loss of confidence and loss of users. The initial peg breaking hurts user and market confidence in Basis. It is a sign that the system has failed to fulfill what it was designed to do: stay stable. When Basis stays de-pegged for an extended period of time, users are likely to rapidly lose any remaining confidence in the system.
The stablecoin space is becoming more and more competitive. Some, like Maker DAO, are ahead of Basis. Others will come. As soon as Basis stays de-pegged for an extended period of time, its users will stop using Basis and switch to competitors, further depressing the price. Stablecoin users have no need for another volatile coin. Due to lost confidence in Basis, peg recovery after an extended break is unlikely.
Even if Basis were to somehow recover its peg, it would not recover its customers. Why would someone buy into a stablecoin that has been broken for five years when there are other successful, stable, widely adopted alternatives? In particular, the people who held Basis when the peg broke will not return to it. Many of them will have lost a lot of money — for example, if Basis de-pegged and dropped from $1 to $0.50 and they sold off in fear of further losses, they will have lost half of their funds placed in what was supposed to be a trusted stablecoin.
The combination of lost confidence and lost users makes extended peg breaks unacceptable for a stablecoin. If Basis stays de-pegged for months it will die. So Basis remains powerless to endure confidence crises, despite its bond expiration period.
One might ask whether this fragility would not have come up in a stability analysis and wonder whether the Basis team did one at all. They did. In this section I’ll detail exactly how their stability analysis did in fact reveal this fragility, despite their claims otherwise.
Their price stability analysis ‘reveals’ that the Basis token’s $1 value survives a wide range of market situations. However, this is only so because the model in their simulations depends on a crucial flawed assumption: it assumes that the market has stable high confidence in the protocol’s ability to repay its bonds, even in periods of drastic demand reduction or periods with a broken peg. It is this groundless assumption that allows them to forecast good outcomes for Basis.
This stability analysis is unfortunately unpublished, despite a 6-month old promise by the Basis team to have it “posted to our website in the coming days.” This makes it impossible for the public to assess Basis’s claims to stability. I ended up with a copy of the document, so have had a chance to analyze it. I want to respect their decision to not publish it yet, but in order to justify some of the claims I’m making here, I will include a few screenshots.
To analyze how the Basis stablecoin handles market scenarios, the Basis team performed simulations based on full future demand histories for Basis. These histories are datasets that represent possible ways the demand for Basis could develop over the coming decades.
However, predicting the future is hard. How do you generate a good projection of how the future demand history for Basis will look? One way is to look at past demand histories of similar assets. Another is to make up random data. Basis uses both methods and thus produces two series of simulations.
The stability analysis uses a model for how the Basis system and the market will react to a given demand history. Given a demand history for Basis as input, the model outputs estimates for what the market price history of Basis will be. Among other things, the outputted price history shows when Basis loses its peg, according to the model.
In the first set of simulations based on past asset history, they use the market cap history for Bitcoin 2011–2017 to approximate what the demand history of Basis might look like early on.
They also use the S&P500 stock index 1928–2017, as an approximation of what the Basis demand history might look in the mid run. Finally, they look at the historical M1 money supply of the Mexican peso, the Canadian dollar and the US dollar as examples of how the Basis demand might look in the long run.
In the second set of simulations, they use randomly generated market cap data, with varied settings for demand growth rate and demand volatility.
Both sets of simulations share a fundamental flaw. The flawed assumption is that there is a constant expected risk of “systemic default” — which is when Basis fails to grow enough to pay off its outstanding bonds in a reasonable amount of time. Basis assumes that the market has a constant expectation of this risk of default, regardless of market conditions. In reality, the expected risk is wildly variable; it will depend on people’s confidence in Basis. When times are bad, confidence naturally drops and the expected risk increases notably.
Here’s Basis own description of their default risk assumption:
They use a constant 20% expected one-year default risk assumption in their Bitcoin demand history simulation. This is equivalent to an assumption that the market thinks the chance of Basis repaying their bonds in a year is 80%.
Basis lowers their risk assumption to 2% for their S&P 500 simulation. This equals a constant expected 98% chance of bond payout in a year.
For their M1 money supply simulations they use 1% or 2% risk assumption (we’re only quoting their Canadian Dollar example here):
In normal times 20%, 2% and 1% are indeed fairly conservative numbers for the market’s estimation of the risk of Basis defaulting in the next year. However, Basis’ challenge is not in staying stable in normal times, but in staying stable when times are bad. That’s when the constant expected risk assumption is problematic.
The Bitcoin dataset includes the December 2013 Mt.Gox crash, with a market cap drop of 83% in 13 months and no recovery for 3 years. Throughout this gloomy market situation, Basis’ simulation assumes that the market expects a stable 80% chance of newly bought bonds to get repaid within a year.
Worse, Basis’ S&P500 dataset includes the Great Depression, where demand dropped by 86% between 1929–1932 and took 17 years to recover. Here Basis assumes a constant market belief in a 98%(!) chance of bonds getting repaid within a year. That assumption is patently absurd. For all bonds to be repaid it is not enough for the market to fully recover, it must expand from the previous demand high point enough to pay for whatever premiums the people who bought the bonds required. Imagine being an ordinary market participant in the Great Depression that followed the 1929–32 drop. How confident would you be that the market will fully recover and then grow some more before your bond expired? How confident would you be that others would continue buying bonds to keep the system afloat in these conditions? If the world is in depression, and the Basis peg has been broken for an extended period, would you still expect a 98% chance that the system would continue functioning for the next year?
With assumptions like these it is no wonder that the Basis token survives a wide range of market situations in their simulations. Under realistic conditions, crashes like the Mt. Gox crash or the Great Depression will erode market confidence in Basis, starting the self-reinforcing low-confidence crisis I described above.
The Basis team assumes that a peg break would not hurt market confidence in Basis. In their simulation, Basis loses its peg, drops below $0.70 for weeks, and stays below $1 for another month before re-pegging. In reality a peg break like this would irreversibly hurt market confidence in Basis. The reason people would presumably trust and use Basis is because of the strength of its peg. Once that is broken, the game is over. In Basis’ simulation the market retains its 98% confidence despite the peg break. No wonder the peg recovers.
A superficial interpretation of the Basis stability analysis and simulation would be that Basis can handle a wide variety of market situations. But once you look closer, these unrealistic assumptions point directly at the system’s fatal flaw.
A currency’s ability to hold its peg is, in fact, determined by its effective reserve backing. To keep the peg, the issuer needs to guarantee that you can always trade the pegged currency for the currency it is pegged against at the target price.
Let’s look at the Danish Krone for an example. It is pegged to the Euro, at a rate of 7.46038 Krone for one Euro. Fiat currencies’ pegs are backed by central banks. It is easy for the Danish central bank to ensure that you can always exchange a Euro for (at least) 7.46038 Krone. If the market price of Krone increases relative to the Euro, so you get less than 7.46038 Euro per Krone, the Danish central bank can just print more Krone and bring down the Krone market price to the peg rate by selling them for 1/7.46038 Euro.
The tricky part of guaranteeing a peg is if the price of Krone decreases relative to the Euro. To ensure that the Krone isn’t worth less than 1/7.46038 Euro, the central bank needs to be able to sell 1 Euro for 7.46038 Krone. Normally, this is done by holding foreign currency reserves. The total value of reserves that the bank holds that can fund Krone buybacks for Euros determines the bank’s ability to stabilize Krone in the upwards direction.
Generally then, the peg is maintained insofar as people believe in the central bank’s ability to ensure exchange between the currencies at the promised rate. This trust in turn depends on the size of the effective reserves relative to the currency in circulation.
Basis aims to be an algorithmic central bank and so we can analyze it like a central bank. The Basis protocol differs from a central bank in that it does not hold any foreign reserves. Their ability to buy back Basis tokens at the $1 peg rate at any time is determined by their ability to sell bonds at that time. So Basis’ effective reserve in a time period is the amount of fiat currency people are prepared to pay in total for all bonds Basis can issue in that time period.
Basis’ effective reserve has two problems. First, it is practically impossible to know the size of the reserve. At best, it can be estimated. Central banks keep known amounts of foreign currency in their reserves. Cryptocurrencies with collateral backing, like TrueUSD, also keep known amounts of reserves. For Basis we can know the market price of bonds at any given time when the protocol is issuing bonds, but we never know in advance how much value Basis can raise in total by selling off more and more bonds.
When people don’t know the size of the effective reserve, they can’t tell how stable the currency is, because they don’t know how well it can survive demand decreases and reduced confidence. This uncertainty makes it hard for Basis to get the market to trust their stability. If people start doubting Basis’ ability to defend their peg, there is no way to respond to doubters by showing a solid reserve backing. Lack of trust can become a self-fulfilling prophecy. What people are willing to pay for bonds depends on their expectations of Basis’ future, which is determined by people’s confidence in Basis’ ability to defend their peg.
Uncertainty about the size of Basis backing also reduces people’s willingness to adopt the currency. We see this with Tether. It doesn’t get as much use as it would if people were certain that it is as backed as it claims.
Second, Basis’ effective reserve is volatile. As people’s expectations of Basis’ future expansion changes, the value of bonds changes, and so does the effective reserve size. If the future looks bright, the reserve increases, but if it looks bleak, the reserve decreases.
The effective reserve’s volatility will not just depend on the market’s beliefs about Basis’, but also on people’s beliefs about cryptocurrencies’ prospects in general. Current cryptoassets’ valuations depend heavily on the market’s expectations of how well cryptoassets will do generally. We can see this by noting how their market caps all rise and fall together, in response to general crypto news.
The same will be true for the market’s expectation of Basis’ future prospects. For example, if Japan bans cryptocurrencies, then Basis’ effective reserve will decrease.
This volatility of Basis’ effective reserve is troubling because what matters is not average reserve size, but how often the reserve backing is insufficient. Compare two currency reserves with the same average backing where one is stable and the other is volatile. The stable reserve gives 100% backing, and the volatile reserve oscillates between 50% and 150% backing with a 100% average. A pegged currency backed by the stable reserve is never at risk of its peg breaking. The volatile currency, on the other hand, is at risk whenever its reserve is below 100%. The good periods with more than 100% reserve backing do nothing to protect the peg in bad periods, despite bringing the average backing up to 100%.
For Basis, the volatility of their effective reserve creates periods of increased vulnerability. When confidence in Basis is lower, the effective reserve goes down, Basis becomes more fragile, and the risk of a fatal peg break increases.
The lack of fungibility of bonds further depresses the effective reserve sizes.
The value of a particular bond depends on its probability of paying out and when it will pay out if it does. Both are determined by a bond’s place in the bond queue. Given that all bonds have different places in the queue, it follows that all bonds have slightly different values. Given that they all have different values, they aren’t fungible. This lack of fungibility makes it hard to form markets that trade bonds. This probable lack of a secondary market for bonds causes two problems.
First, the fact that these markets won’t exist makes it so that people won’t be willing to pay as much for bonds, as they’ll charge an illiquidity premium. Thus, the value of bonds is immediately penalized by their lack of fungibility. Because the bond value is what determines how much decrease in demand Basis is able to handle, this reduced value penalizes Basis’ stabilization ability by decreasing the effective reserve size.
For example, if bond buyers pay half as much for illiquid bonds as what they pay for liquid bonds, then Basis’ effective reserve is cut in half — the amount of money the buyers will pay to get all future minting is cut in half, and that in turn cuts the total value Basis can raise to defend their peg in half.
Second, a lower bond valuation also makes it more expensive for Basis to stabilize the peg against decreasing demand. For example, if people pay half as much for a bond, then twice as many bonds have to be issued in order to take Basis tokens out of circulation when the demand for Basis tokens decreases. This means that twice as much of the value of minted Basis tokens goes to bond purchasers. That value is taken from shareholders, who would have gotten it otherwise. So shareholders get less profit because bonds are illiquid.
Like I noted in the beginning of this post, I think that the Basis team has helped the cryptocurrency community by sharing correct insights and correctly directing investor attention. I also think they are well-intentioned — which might raise the question of why I’m writing a detailed analysis of how the combination of their poor algorithmic stabilization, illegible and volatile backing, and off-chain stabilization plan will cause them to fail.
The harm Basis could cause won’t happen if Basis survives in the long run and becomes a global currency. Neither will it happen if it fails in the short run. However, Basis is most likely to fail in the mid run, and a mid-run failure will harm a lot of people.
Basis is unlikely to fail early on because the team can stabilize it with centralized off-chain money. (This is assuming they won’t misallocate the funds or be shut down by the US government.) Off-chain stabilization with lots of funds will both prevent and mitigate confidence crises and lengthy periods of de-pegging.
In the success scenario where Basis does really well and manages to become a global currency like the dollar, Basis will stabilize naturally. A global currency has a large base of diversified stable demand that limits its volatility.
However, before that success can be reached, the off-chain reserve stabilization hack will not be able to protect the peg. If Basis does manage to get people to adopt their currency, then their off-chain reserves will eventually not be enough to back it, and it’s likely to fail.
Who are Basis’ mid-run users that will be affected by this failure? Ordinary people. Unlike investors or speculators, they will be neither aware nor prepared to handle the risks involved in holding unstable assets. A stablecoin at this point in the game will be held in large part by people in the developing world who put their life savings in stablecoins to avoid the problems of inflationary currencies.
If Basis gains widespread but not total adoption, their off-chain reserve stabilization system won’t be enough to stabilize inevitable fluctuations. When the peg breaks, ordinary folks will lose significant amounts of money. Poor people may lose their entire life savings.
Not only would this be crippling for those ordinary people, but it would be crippling for the reputation of stablecoins and of the cryptocurrency community in general.
If you have any feedback on the accuracy of this post or want to discuss how to turn cryptocurrency into real-world money, email me at [email protected]