Wednesday, March 15, 2017

The dematerialization of cash

"One dollar bill," watercolour by Adam Lister (source)

R3, a company specializing in distributed ledger technology, has just posted a paper I wrote for them entitled Fedcoin: A Central Bank-issued Cryptocurrency. And here is a nice write-up on the Fedcoin idea in American Banker, which unfortunately is behind their paywall.

The paper is pretty wide-ranging, but one thing that's worth focusing on is the ability of Fedcoin to provide some of the same features as banknotes, in particular anonymity and censorship resistance. That a Fedcoin system can be designed to provide the same degree of privacy as cash runs counter to some of its early critics, who see in Fedcoin a coming financial panopticon.

One really neat things about good ol' cash is that, like bitcoin, it is a decentralized network. The opposite of this is a centralized network, say something like the deposit banking system. In the banking system, storage of value is handled by the issuing bank through accounts hosted on the bank's database. Conversely, in a banknote system, the issuing central bank offloads the task of storing value onto us. We the public—think of us as nodes in a decentralized network—are responsible for choosing how and where to keep our cash, say in a wallet, or under our bed, or in a safety deposit box, as well as bearing those storage costs. The central bank doesn't care how we manage this task, though they'd prefer that we don't mangle the notes too much.

Or take the process of securely transferring value. Centralized actors like banks handle all the stages of moving deposits from a buyer to a seller, including verifying identities, ensuring adequate account balances, updating ledger entries etc. But in a transfer of banknotes, the transaction process is entirely devolved to the buyer and seller, who must physically move the cash to the right location, count out by hand the necessary quantity of banknotes, and then come to a consensus that the transaction has been settled. As for the central bank's ledger of notes, there is nothing that needs updating. Unlike private bankers, central bankers don't care who owns their circulating liabilities.

The task of screening for counterfeit notes is also outsourced to the public. Each time a shopper accepts a banknote, say as change, they'll give it a once-over to verify that it hasn't been run off by a teenager using an inkjet printer. Retailers deal in cash all day and are familiar with banknote anti-counterfeiting devices, and thus can exercise more judgement in checking for fakes. And banks, the recipient of notes from retailers at the end of the day, will catch many of the counterfeits that have slipped through the system.

Banknote systems aren't entirely decentralized, of course. The central bank has the final say on whether a note is counterfeit or not. It also regulates the purchasing power of those banknotes, either by toggling interest rates higher or lower, repurchasing money using its portfolio of assets, or issuing more money in return for assets.

Because they are at least partly decentralized, banknote systems inherit two nice features: anonymity and censorship resistance. The first feature is self-explanatory: the central issuer makes no effort to determine the identity of a banknote owner. Proceeding from this, the issuer lacks enough information to censor, or prevent any particular party, from using the banknote network. These are completely open systems. By contrast, centralized systems like banks can easily censor members of the public from making payments. Take the Huntingdon Live Sciences episode in 2001, for instance, in which a UK-based company involved in drug-testing was cut off by British banks in response to pressure from animal rights activists. Other examples of censorship by banks include the blockade of Wikileaks and the monetary embargo of Iran.

Now in theory a banknote system could be modified by introducing more centralization, thus removing anonymity and introducing censorship. Each banknote has a unique serial number. The central bank could set a rule that for every cash transaction, the buyer and seller are obligated to log in to a government-provided account where they register the note's serial number into a tracking database. To get these accounts, users would be required to submit documents and ID. This would destroy the anonymity of cash users and open the door for censorship. In practice, though, the guardians of banknote systems have chosen to preserve anonymity by ignoring serial numbers.

One of the major trends over the last decades has been dematerialization, the replacement of paper by bits and bytes as a medium for holding data. This saves on costs like printing, storage, and distribution; improves speed; and reduces waste. We saw it with stock & bond certificates a few decades ago, books, newspapers, music, record keeping, bills. And one day we may see it with cash. The question is: how to allow for the dematerialization of cash without losing its useful features like anonymity and censorship resistance?.

Bitcoin is one answer. But Bitcoin only goes half-way to solving the problem since it does not recreate one of cash's other key features, its stability. A nation's prices are conveniently denominated in terms of its paper money (i.e. U.S. retailers set prices in dollars, Japanese retailers in yen), and since prices tend to stay sticky for 4.3 months or so on average, the public has a huge degree of certainty over the medium-term purchasing power of the money in their wallets. This is a very nice feature. Bitcoin prices? Not so stable.

Fedcoin may be able to recreate this stability while still providing anonymity and censorship resistance. A government copies the source code of a proven cryptocoin (maybe bitcoin, maybe zcash, maybe ethereum), boots the system up, and promises to peg the price of each coin to its existing banknotes, say 1 coin = $1 banknote.

As with bitcoins, anyone would be able to hold Fedcoins without the necessity of providing identification. And like Bitcoin, Fedcoin could be designed in such a way that a distributed collection of Fedcoin nodes (or miners) validate transactions by referring to the system's shared history. Remember how the Fed allows banknote users to anonymously come to a consensus about the validity of a banknote transactions i.e. they do not have to log in to an account to register note serial numbers? Likewise, Fedcoin could be designed in a way that nodes have the ability to remain anonymous. This would preserve a degree of censorship resistance and openness. After all, if validators must unveil themselves, governments and other powerful actors might compel those nodes to censor transactions.

This is just one way of setting up anonymous central bank money. I'm sure there are many others. There are also ways to set up non-anonymous central bank money, but these are less interesting to me, a point I made here. As I point out in the R3 paper, I think my preferred set-up would be to allow individuals a rationed amount of anonymity, targeting some sort of “sweet spot” such that there is enough anonymity-providing exchange media for regular consumers but not enough for criminals. But I can't wrap my head around how to design something like this. Anyways, go read the paper, there's plenty more.

PS: I also recently had an article on bitcoin published in the Common Reader, a publication of Washington University in St. Louis.

And since we're on the theme, I should link to some other public appearances by yours truly, including recent-ish podcasts with David Beckworth and Alex Millar.


  1. "Fedcoin could be designed in a way that nodes have the ability to remain anonymous. This would preserve a degree of censorship resistance and openness. After all, if validators must unveil themselves, governments and other powerful actors might compel those nodes to censor transactions." Would be great to hear more about this. Ultimately, a transaction must be recorded in the ledger i.e. it's recorded that key A transferred money to key B. I know zerocoin talked about doing this with zero knowledge proofs that would allow you to prove you owned coins without revealing their source. I would be interested if you could write more about this and explain more of the details because this feature is not available in bitcoin.

    1. You're right, bitcoin isn't sufficient because it only provides pseudonimity. As I pointed out in the paper, to offer actual anonymity Fedcoin would have to be a clone of something like Zcash, which uses zero knowledge proofs. Since Fedcoin would be a long-term project, ie. unlikely to debut for many years, central bankers have a lot of time to monitor the various private cryptocoins to see which one offers the best features.

  2. Great stuff, JP! Now I don't need to think what I should read the next hour or so. Thanks.

    Btw, I got to read the full American Banker article with no trouble. Perhaps just lucky.

    1. Thanks for the heads up. They must have removed the paywall after 24 hours or so.

  3. JP-
    A question came to mind today and Ifigured you would be the guy who might have the answer.

    I know countries like the US indirectly earn substantial seigniorage through banknotes being collected or stored indefinitely - BUT is there any modern example of a country directly charging for banknote distribution? The argument being that this would level the playing field for competition between cash and cash alternatives.

    1. Good question. Historically central banks provided commercial banks with cheap/free cash handling/distribution, thus subsidizing the use of cash. Over the last two decades central banks have begun to recover the true cost of cash handling/distribution by privatizing these processes and/or requiring banks to takeover more cash handling activities from the central bank.

      Here is the Bank of Canada:

      Here is the Riksbank: