That thing about ICOs and Monetary Systems

With the Tezos thing going on and ICOs being all the rage — with over 2 billion in funds raised so far, and the largest ICOs raising a couple of hundreds of millions on their own, this starts being serious financing, and it is worth thinking about what is going on.

So what is an Initial Coin Offering, or ICO? Let’s start what it is not: it is not a Kickstarter-like prepayment of services offered by a company. But let’s not get ahead of ourselves, and let’s first introduce introduce the two companies that I’ll use throughout this blog post: Alison Dee are setting up a minicab service with a slick app, and in order to fund the development they issue Alison Dollars, short A$, that they’ll later accept in exchange for rides. Those Alison Dollars are freely transferrable between owners, so they can be — and we assume they are — listed on an exchange where they can be bought and sold. A$ are what I want to call pre-paid service tokens, which are importantly not ICOs.

The other company, Hyss, on the other hand is setting up a ride sharing platform: they also build a slick app, but they will not be offering the rides, but they’ll be a market place for potential drivers and potential customers to interact. They also issue a token — either called Hyssi Fits, or H Bombs — but theirs operates differently. Instead of like A$ being a prepayment of services the issuing company is meant to provide, HFs/HBs are token that everyone using the app must use to transact: customers pay their fare in HF/HB, and the drivers receive HF/HB — up to the customers to get the tokens when they need them, and up to the drivers to exchange them into something else if they want to eat. Like A$, HF/HB are freely tradeable, and listed on an exchange.

You might be wondering why Hyss’ token have two names. It is because I want to distinguish between two cases: in the first one, where the company issues Hyssi Fits, they fix the price of rides in advance, say one HF corresponds to 1km ride. In the second one, where the company issues H Bombs, the price is either fixed in fiat terms, or will at least be adjustable if the cost of HBs is too far out of line with either competing minicab operations, or the price of inputs that go into a ride. We want to call HFs and HBs network tokens, the former pegged, and the latter floating. I’d argue all distributions of network tokens are ICOs, but as far as I know all ICOs thus far have been of the floating type.

Alison Dollars (pre-paid service tokens)

Alison Dollars represent a real service to be provided — say 1km of ride— and are somewhat akin to a bank deposit, with two key differences:

  • the liability is denominated not in a fiat currency, but in a different numeraire, notably 1km of ride
  • if Alison Dee is a start-up, there is probably quite a significant risk of default, and there is no deposit protection in place

I do not want to get into a discussion here when a private sector liability can be considered money. If Alison Dee is a start-up with a significant non-performance risk it is probably too risky to be used in everyday transactions. However, if Alison Dee is a major established taxi-cab company with thousands of cabs who simply wants to fund the development of an app this might be a different story as the default risk might be reasonably small, and depending on the interest rate they offer (because they will have to sell the tokens at a discount) it might be an interesting financial investment even for people who do not want to use their services.

So to conclude this section: Alison Dollars are structurally similar to certificates of deposits (tradeable term deposit notes issued by banks), and depending on the liquidity of the token and the associated risks (default risk, and price risk of the numeraire against fiat) this might be a financial investment that can compete with savings deposits, or it is something a bit more racy. In any case, it is credit that is extended by a retail customer to a company, and is therefore substituting regular bank lending.

Hyssi Fits (pegged exchange tokens)

Similar to A$s, HFs are tokens that are pegged to a certain level of services: one HF corresponds to 1 km of ride exchanged on the platform. The key difference here is that HFs are not liabilities of the platform — ultimately it is the decision of drivers to provide their services on the platform and to accept 1 HF in exchange for a ride of 1km, or to rather drive for another platform where they can earn more. Drivers’ switching cost are not dramatic, so they can take this decision on a daily basis, or even an hourly basis if it makes sense for them.

Let’s first think about the steady state that we eventually want to achieve: we know that a token corresponds to a 1km ride, and we also know what 1km rides in comparable services like Alison Dee cost, so this gives us a range where those token should trade. There are complications however. Let’s look at the launch phase for example: a lot of people will hold Hyssi Fits, but there are very few drivers. HFs can not be converted into their nominal cash value, so chances are that (some) holders are willing to sell them at a discount to people who actually want to take rides.

This poses a problem for the drivers: to the extent that there is a big selling overhang in the market they will only be able to convert the Hyssi Fits they earned into cash below par as potential customers will not buy the drivers’ tokens at par if they can get them at a discount from the investors. This in turn means that drivers will not drive for the platform because on other platforms they’ll earn more, and the market collapses.

The only way to deal with this is if someone — naturally the platform provider — inserts himself into the process and allows the drivers to sell their earned HFs at par. Effectively this means that the platform provider is running a central bank for the currency they issued. The analogy does not hold entirely as they do not maintain the peg for everyone, but only for drivers to the extent that they have earned their HFs, but the principle is the same.

There are two options where this could end up: either the market eventually achieves a steady state, or the platform provider runs out of cash, and the market collapses. But in this scenario even steady state is not quite what we might expect it to be: there is very little incentive for anyone to hold HFs before they need them. So almost the entire HF money supply will end up at the HF central bank, and customers will buy HFs from the central bank just before they need them, and the drivers will sell them back to the central bank pretty much immediately as well.

To conclude this section: the end game in this scenario is that most of the Hyssi Fits will need to be bought back by the platform-sponsored central bank at par. If they kept enough of the money they received at the ICO in reserve that’ll work. If not, they will run out of money, therefore the platform will not be able to pay service providers at par, and therefore the service providers will abandon the platform and trading collapses.

H Bombs (floating network tokens)

Like HFs, H Bombs are network tokens, except that they are not pegged to anything: they are worth whatever people think they are worth, and it is up to the service providers and service buyers on the platform to negotiate how many HBs are to be paid for a 1km ride, or possibly that price is fixed by the platform but updated in real time to ensure that customers pay the same they’d pay on competing platforms in fiat, and that service providers received the same they’d receive on competing platforms in fiat.

The issue is that now we have a multiple-equilibrium problem. And when I say multiple, I really mean Aleph-null (or Aleph-one if we assume prices to be real numbers rather than fractions): pretty much every possible exchange rate of H Bombs versus fiat currencies is a possible equilibrium. The only constraint is that depending on the trading facilities in place there is a certain monetary amount needed for settling active transactions. But if we assume that riders can buy their HBs just before they pay the drivers, and drivers immediately convert HSs back into fiat once they have received them this amount tied up in the system is de minimis.

Now most systems have multiple equilibria, and actually both Alison Dollars and Hyssi Fits have them. Those equilibria however are of the discrete type, and they are the system is running and the system has collapsed. The regime change in both cases happens when token holders lose trust, eg because they believe Alison Dee will go bust, or because they believe Hyss’ central bank will run out of cash: in this cases they’ll fall over themselves dumping the tokens, and they’ll end up trading at zero, or some kind of vulture value driven by people who think they’ll ultimately go back.

The Aleph-null equilibrium is different in that it is continuous: a little push and it moves. In the presence of momentum investors (and arguably all investors in those tokens are momentum investors anyway as tokens do not have any intrinsic value that fundamental investors could latch on to), small moves tend to be amplified as brilliantly illustrated in the famous cartoon below:

So there is little chance the floating tokens will ever settle to any fixed value, and they’ll be moving around depending on Keynes’ animal spirits. Whilst speculators might like this, it is a nuisance for platform users at either side of the transaction: If I take (or drive) a car from Victoria to Liverpool Street I want to know how much it will cost me (or how much I will get paid) in fiat money terms. If I want to speculate on a token I open an account with City Index or someone similar, but there is absolutely no value of bundling speculation with the provision of a service. Given that both customers and drivers are able to take their business to other platforms at any point in time, only those platforms where the fiat value of the trip is in the correct range will be able to intermediate business; all others will be abandoned either by drivers (if they are too cheap) or by customers (if they are to expensive).

There are two possibilities to address this conundrum on the platform level: either the token value is kept reasonably stable, or the price in token terms is permanently adjusted — during the ride if need be — to ensure the fiat price of the ride is what both parties expect it to be. The latter is easy to implement: the ride price is quoted in fiat, and it is only converted into tokens at the end of the ride, in a transaction where both parties get exactly the same exchange rate and therefore no one is taking on token price risk. This of course is a fiat transaction in all but name, and there is no need to use a token at all.

The former is harder: like for Hyssi Fits, it requires a central bank. Here the central bank must implements a fiat peg, ie it intervenes in the market, buying and selling tokens to maintain the price of tokens relatively stable to the fiat currency to which it is pegged. Experience shows that this works until it does not — an example to look at is what happened to the pound in the European Exchange Rate System. Once there is a doubt whether the central bank has the necessary resources to defend the H Bomb they’ll get their own Soros who will attack it, and the peg will probably break.

I also want to mention a second solution that could be implemented, which is a currency board where every token is backed by fiat currency held in escrow, and where the central bank is willing to issue and redeem tokens at par. This solution keeps the price stable, but it has two issues: firstly it requires a lot of cash, and if the sponsor has spent some of the funds raised in the ICO for platform development a full currency board will not be possible. Secondly it is probably illegal under money laundering regulations because it might not be allowed to issue cash-like anonymous bearer tokens.

To conclude this section: without stabilisation, H Bombs will just move erratically, and users of the platform will require a method to put in place that de facto removes the price risk from the system, in which case for all intents and purposes the platform runs on fiat. Token stabilisation is in principle possible, but in practice hard to do: any fiat peg is susceptible to attack and to breaking, and running a currency board requires too many resources and is probably illegal anyway.


I have looked above at three different mechanisms how tokens for service offerings could be pre-sold, notably

  • Pre-paid service tokens (Alison Dollars)
  • Pegged network tokens (Hyssi Fits), and
  • Floating network tokens (H Bombs)

The conclusion was the pre-paid service tokens can work nicely, and that they are disintermediating the credit supply that is usually provided by banks. Whether or not the resulting tokens are money-like depends on the exact characteristics of the transaction*. Pegged service tokens, where a token is tied to a specific quantity of a specific service, can work, but they most likely require a central bank that redeems tokens for services that have been provided at par, otherwise the underlying market will collapse. Floating service tokens (which most of not all ICO tokens are) that are not tied to anything, but that they are the required means of payment on the platform lead to a continuum of equilibria and probably to significant volatility. This most likely can only be addressed by either de-facto removing the token from the transactions and operating the platform in fiat, or by creating a token central bank, in which case the system is likely to be either unstable, or insufficiently resourced and illegal.

Maybe I have missed something — and in particular on monopoly service providers some of this analysis does not apply, because customers and service providers do not have a choice — but from this analysis, the future of tokens might not be that bright.

As an aside, arguably fiat currencies are a big confidence trick where something has a certain value because some magician says it has, and everyone is mesmerised and believes it. However, at least this confidence trick is run by (mostly) democratically elected governments, and it (mostly) works. ICOs of the H Bomb type are pretty much the same thing, except they are run by private corporations with no accountability to the wider public, which is quite ironic given where the crypto currency movement is coming from.

*Note that in principle any token could fall short of AML regulations, this is not a problem confined to a currency board only; however, generally the closer a token is to being money the more likely that regulators will want AML regulations to be applied