Updated: Sep 17, 2018
This is an instalment in a series of blog posts on crypto-token design and economics, aka “Tokenonmics”. This article was initially be published on the Meritt blog.
In the world outside of crypto currencies, people are getting paid in fiat currencies, and products and services are priced in in fiat currencies. Companies and individuals could of course choose to price their offering in crypto currencies, but in a competitive market this is suboptimal: either providers adjust their prices whenever the value of their chosen crypto currencies changes against fiat — in which case they are really running a fiat-linked system with a certain amount of settlement volatility — or they end up mis-pricing their offering against their competition, with the obvious impact on business and earnings.
The alternative is a platform or protocol whose native token is fiat-linked: there is a mechanism in place that ensures that the token’s volatility against a chosen fiat currency is zero, or at least very low. Note that the token does not necessarily need to remain constant against fiat as it can move in a predictable, planned manner: if the token goes up over time then this can be considered an interest reward for holding the token, and if it goes down over time then this can be considered a fee to be paid by token holders. The former would encourage people to hold the token over time (but the interest payment would need to come from somewhere), and the latter would discourage people to hold the token over and beyond what they’d need in the foreseeable future (also depending on the transaction costs of trading into and out of the token), but it would provide some income for those running the stable coin platform.
It is not necessarily the native token that has to be fiat-linked. For example on an Ethereum-like platform the stablecoin token could be an ERC20 token, and there could be one stablecoin token for every currency, or even multiple competing ones. Smart contracts in turn could recognise certain stablecoin tokens, and instead of expecting payment in the native currency they’d expect payment in the respective stablecoin token, with the native coin being reduced to paying the gaz needed to pay for compute.
Token ownership does not give rise to any particular rights, so governance is not relevant in this respect. However, keeping a token stable against fiat is a strong economic promise that depends on an actor — or a group of actors — to deliver on that promise. First and foremost this is an economic problem and therefore discussed in the section on token economics. However, the platform governance must foresee the situation when this actor or those actors do not deliver on that promise and the fiat-link gets broken, and determine what should be done in this case to remedy the situation.
Token economics and valuation
To start with the easy part: as the token is linked to a fiat currency its value is whatever it is designed to be, unless things go wrong, which we will discuss below. First we will discuss however the more interesting question, which is how we can ensure that the price of the token is indeed what it is meant to be.
Basic microeconomics suggests that the exchange rates of two assets A and B is equal to unity if and only if there are sufficient buyers and seller who are willing and able to do the exchange at that price, in bo th directions,and in a reasonable volume. If you want the exchange rate to be unity at all times then you the presence of those buyers and seller must be asserted at all times.
Let’s assume F is an established fiat currency (say the USD), and T is some token that we want to hold at par against this currency. One way to do it is to run the currency board model or ETF model. To start with a definition, HKMA has defined currency board as follows:
A Currency Board System is a monetary system that complies with the Monetary Rule requiring that any change in the Monetary Base should be matched by a corresponding change in foreign currency reserves in a specified foreign currency at a fixed exchange rate. In operational terms, the Monetary Rule often takes the form of an undertaking by the Currency Board to convert domestic currency into foreign currency reserves at the fixed exchange rate.
So in order to create a stablecoin token T we can establish a currency board / ETF structure with a chartered actor who can issue and redeem the T tokens, and who (like for an ETF) undertakes to (1) issue the tokens at par to whomever deposits the fiat currency F, and to (2) redeem them at par as well to whomever deposits the tokens with them (it is not strictly necessary that everyone can demand and redeem tokens; indeed it is sufficient if a competitive group of actors have the right to do so, but this is beyond the scope of this discussion).
If we allow for some residual volatility then we can allow for a bid-offer spread between issuance and redemption, and/or a fixed fee at every exchange. This has a number of interesting consequences
a bid/offer spread encourages people who regularly spend and receive tokens to hold on to them rather than to exchange them to fiat; on top of this, a fixed exchange fee encourages people to change larger amounts / less often, which also encourages holding the token
any transaction costs at the official exchanges encourages secondary providers to move in and to make a tighter market; it can also additional encourage service providers to accept the token as payment from those that earn it through work on the platform (eg gas stations accepting ride hailing tokens)
excess volatility and excessive transaction costs however can lessen the attractiveness of the platform: if eg service providers can only liquidate their fiat-linked tokens at a 10% cut, and customers pay a 10% premium for acquiring the tokens this will probably not be good for the volume on the platform
The only way that this chartered actor can ensure that they can always fulfill their obligations to exchange the token into fiat and vice versa is if (a) they control the token supply, and can issue and redeem tokens at will, and (b) they keep all the fiat currency they receive in escrow, so that they are at any point in time to meet redemption request. In a positive interest rate environment, and assuming the exchange rate is kept constant, there is a seignorage gain due to the ability of earning interest on the fiat currency received. In a negative interest environment this turns into a loss. Also, in the more general case where tokens can also earn interest — meaning the forward curve is not flat, and the promised exchange rate changes over time — seignorage profits are positive if and only if the fiat interest is higher than the increase in the price of the token during the same period.
There are two related structure that give almost the same result, and that can be economically more attractive to run: the money market fund model and the bank model. In the money market fund model, like in the eponymous fund, the token issuer does not hold the funds received in central bank cash, but in slightly higher yielding assets, typically because they are longer dated (say, up to one year maturity). Those assets should not usually have a credit risk, so they could be government securities, or possibly AAA-ratedcommercial paper or other short dated, ultra-high quality privately issued securities.
If we assume that the no credit risk part has worked out well (which admittedly is a big assumption ), then the only risk here is liquidity risk: too money token holders show up that the same time reclaiming their funds, and whilst the assets are there in principle they are in longer-dated securities and not in cash. If those assets can be sold at or above par this is not an issue, but in many realistic scenarios (eg, a liquidity crisis, or some doubt on the credit quality of the assets) this is not the case. Ultimately what is needed here is a liquidity line, ie a commitment by a reliable player (or multiple reliable players) to lend at par against those assets if ever there are excess redemptions.
In the bank model the issued tokens are the equivalent for certificates of depositissued by a bank, ie they are a liability (in the legal / accounting sense of the word) by the issuer, and more precisely promise by the issuer to repay them upon presentation, like a sight deposit. Typically deposits pay interest, and so could those tokens, eg through a redemption value that increases by x% per year. In the bank model, the restrictions of the assets in which the proceeds can be invested are very loose, or even non-existent. Instead, that value of the tokens is protected by an equity cushion: there must be someone else in thecapital structure of the issuer who takes the first-loss risk, and who earns the excess profit, ie the profit on the asset investment over and beyond what is being distributed to the token holders.
The trick here is to ensure that the equity cushion is large enough, and commensurate with the riskyness of the assets that the proceeds of the token issuance are invested in. For example if the investment goes into AAA-rated, floating rate government bonds, this equity cushion can be very small. For retail mortgage or corporate loans it must be bigger, and for highly volatile investments like stocks it must be even bigger. Like for the money market fund there is the risk that redemption requests exceed the liquid assets of the issuer. This can in particular happen in case people lose trust in the issuer and are afraid the equity cushion is not big enough compared to possible losses, in which case there is the risk of what in the case of a bank would be called a bank run.
Like in the world of banks, there are two instruments that can address this risk: a backstop liquidity provider, which in the banking world is usually a central bankwho will lend against the long dated assets held in case of excess redemption requests, and a deposit protection scheme that undertakes to indemnify the token holders against any losses due to issuer default, and which — if sufficiently trusted — should limit the redemption requests even in case there is a break down of trust in the issuers. The quid-pro-quo for promising emergency liquidity assistance and deposit protection is that the issuer submits to a prudential regulation regime, including regular audited reporing of the assets, and a maintaining a minimum size of the equity cushion that is commensurate with the risk of the asset portfolio.
An apparent alternative to the three models discussed above is a central bank (intervention) model, ie a central-bank like structure that keeps the token within its allocated band by open market purchases: whenever the fiat-value of the token goes towards the lower end of the bank the central bank purchases it, and whenever it gets towards the higher end the central bank sells it on the open market.
Two observations to start with: firstly, this system only ever will be able to provide a trading band, not a hard peg, so there will be some residual volatility. Moreoever, it only works in an environment where liquidity is usually provided by other market makers that operate on the platform — if the central bank ends up being the counterparty of choice for end customers because there is noone else trading we are de facto in the currency board model. On that basis, any central bank system will probably start up with a currency board system, and only over time the central bank will pull back from being a guaranteed backstop provider to someone only intervening in the markets if need be.
Experience in the fiat world suggest though that this system is stable until it is not. In the world of international finance it (sort of) works because there is a belief that the central bank in question will be supported by its government and in particular its associated tax raising power. By definition we do not have this here, so any central-bank model will implicitly rely on one of the three models provided above:
the currency board model where the central bank holds fiat currency backing the issued tokens 1:1,
the money market fund model where the issuance proceeds are invested into (supposedly) risk free securities with a slightly longer duration, and where it therefore runs a liquidity risk, and
the bank model where the issuance proceeds are invested in risk assets and there is an equity cushion ensuring there are aways sufficient funds available to repay the issued tokens at par.
As discussed above, all models require other parties to work well: the currency board model requires access to a deposit facility, ideally with a central bank to ensure that there is zero credit risk. The fund and bank models require a committed liquidity provider and possibly a protection scheme, otherwise they are prone for a run.
There are possibly some programmatic, smart contract-based solutions that could recreate some of those institutions, eg a committed lending smart contract that promises to lend against assets at par. Ultimately though whenever you analyse those kind of solutions in detail you will find that you either have full collateralisation somewhere — ie you are running the currency board model — or you have simply pushed the risk scenario further into the tail: it is less likely, but more severe if ever it does crystallise.
A good analysis of existing stable coins is for example in this Cointelegraph article. They use the term fiat-collateralised stable coins for what we have called the currency board or possibly the money market fund model above, and they put Tether, Digix and Petro (with reservations) into this class. Their second model is the crypto-collateralised stable coin which in our classification would be a banking model where the asset side is only crypto, and where the protection is provided by an equity cushion, albeit without lender of last resort and/or prudential supervision. They cite Bitshares, MakerDAO and Havven as examples. The last category is what they call non-collateralised stable coinswhich is their name for stable coins organised by the principle of seignorage shares. Note the non-collateralised moniker is misleading: in our classification this would be a bank model where the collateral pool are sharesin the terminology of Sams’ paper. The put Basis (ex Basecoin) and Saga in this category.
A second good article on this topic is the one in Multicoin. They mention a few other seignorage coins, notably Fragments, Carbon, and Kowala. What we call the currency board model they call the Centralised IOU Issuance model and they include here additional currencies like TrueUSD, Arccy and Stably. What we call the bank model they call the Collateral-Backed-On-Chain model, and they include here also Sweetbridge and Augmint.
Security or utility token
I am not a lawyer, and this is definitely not legal advice. Having said this, the token is certainly not a speculative investment as there is absolutely no room for a future gain, and it can only be used on the platform, so on that basis it is arguably a utility token. On the other hand it is exchangeable at par against fiat currency at any point in time, so arguably it is a fiat currency for all practical purposes: people can use it to settle fiat claims that are in no way related to the platform. On that basis, AML and KYC regulations will almost certainly have to be taken into account.
Stefan Loesch is a partner at LexByte. He recently published “A Guide to Financial Regulation for Fintech Entrepreneurs”. He is a serial tech entrepreneur, and he previously worked at J.P. Morgan, McKinsey & Co, and Paribas.
(c) Copyright Stefan Loesch 2018.