Andy Haldane and BOEcoin

The 1995 British two pound "Dove" coin

The Bank of England's chief economist Andrew Haldane recently called for central banks to think more imaginatively about how to deal with the technological constraint imposed by the zero lower bound on interest rates. Haldane says that the lower bound isn't a passing problem. Rather, there is a growing probability that when policy makers need three percentage points of headroom to cushion the effects of a typical recession, that headroom just won't be there.

Haldane pans higher inflation targets and further quantitative easing as ways to slacken the bound, preferring to focus on negative interest rates on paper currency, a topic which gets discussed often on this blog. He mentions the classic Silvio Gesell stamp tax (which I discussed here), an all out ban on cash as advocated by Ken Rogoff, and Miles Kimball's crawling peg (see here).

According to Haldane, the problem with Gesell's tax, Rogoff's ban (pdf), and Kimball's peg is that each of these faces a significant 'behavioural constraint.'  The use of paper money is a social convention, both as a unit of account and medium of exchange, and conventions can only be shifted at large cost. Tony Yates joins in, pointing out the difficulties of the Gesell option. Instead, Haldane floats the possibility of replacing paper money with a government-backed cryptocurrency, or what we on the blogosphere have been calling Fedcoin (in this case BOEcoin). Unlike cash, it would be easy to impose a negative interest rate on users of Fedcoin or BOEcoin, thus relaxing the lower bound constraint. Conventions stay intact; people still get to use government-backed currency as a medium of exchange and unit of account.*

While I like the way Haldane delineates the problem and his general approach to solving it, I'm not a fan of his chosen solution. As Robert Sams once pointed out, Fedcoin/BoEcoin could be so good that it ends up outcompeting private bank deposits, thus bringing our traditional banking model to an abrupt end. Frequent commenter JKH calls it Chicago Plan #37, a reference to a depression-era reform (since resuscitated) that would have outlawed fractional reserve banking. If Haldane is uncomfortable with the Gesell/Rogoff/Kimball options for slackening the lower bound because they interfere with convention, he should be plenty worried about BOEcoin.

---

I do agree, however, with Haldane's point that the apparatus adopted to loosen the constraint should interfere with convention as little as possible. We want the cheapest policies; those that only slightly impede the daily lives of the typical Brit on the street while securing the Bank of England a sufficient amount of slack.

With that in mind, here's what I think is the cheapest way for the Bank of England to slacken the lower bound: just freeze the quantity of £50 bills in circulation. Yep, it's that easy. There are currently 236 million £50 notes in circulation. Don't print any more of them, Victoria Cleland.**

I call this a policy of embargoing the largest value note. How does it work?***

Say that in the next crisis, the Bank of England decides to chop rates from 0.5% to -2.0%. Faced with deeply negative interest rates, the UK runs smack dab into the lower bound as Brits collectively try to flee into banknotes. After all, banknotes offer a safe 0% return, the £50 note being the chosen escape route since those are the cheapest to store and convey.

Flooded with withdrawal requests, banks will quickly run out of £50s. At that point the banks would normally turn to the Bank of England to replenish their stash in order to fill customers' demands. But with the Bank of England having frozen the number of £50s at 236 million and not printing any new ones, bankers will only be able to offer their customers low denomination notes. But this will immediately slow the run for cash since £20s, £10s, and £5s are much more expensive to store, ship and transfer than £50s. Whereas people will surely prefer a sleek high denomination note to a deposit that pays -2%, they will be relatively indifferent when the choice is between a bulky low denomination cash and a deposit that pays -2%. Thus the lower bound has been successfully softened by an embargo on the largest value note.

Once negative interest rates have served their purpose and the crisis has abated, they can be boosted back above 0% and the central bank can unfreeze the quantity of £50s. Everything returns to normal.

A few conventions will change when the largest value note is embargoed.

1. People will no longer be able to convert £50 worth of deposits into a £50 note. Instead they'll have to be satisfied with getting two £20s and a £10. That doesn't seem like an expensive convention to discard. And if folks really want to get their hands on £50s, they'll still be able to buy them in the secondary market, albeit at a small premium.

2. In normal times, £50 notes always trade at par. Because their quantity will be fixed under this scheme, £50s will rise to a varying premium above face value whenever interest rates fall significantly below zero. For instance, at a -2.0% interest rate a £50 note might trade in the market at £51 or £52.

The par value of £50 notes is a cheap convention to overturn. The majority of the British population probably don't deal in £50s anyways. Those who do use £50 notes in their daily life will have to get used to monitoring their market price so that they can transact at correct prices. But the inconveniences faced by  this tiny minority is a small cost for society to pay in order to slacken the lower bound.

3. Importantly, there will be no need to proclaim a unit of account switch upon the enacting of an embargo on £50s; the switch will be seamless.

Because the £50 was never an important part of day-to-day commercial and retail existence, come negative interest rates no retailers will set their prices in terms of a £50 standard. If they do choose to set sticker prices in terms of the £50 note, they will find that if they want to preserve their margins they will have to levy a small surcharge each time someone pays with £20s, £10s, and £5s and bank deposits. Given the prevalence of these payment options, that means surcharging on almost every single transaction. That's terribly inconvenient. Far better for a retailer to set sticker prices in terms of the dominant payments media—£20s, £10s, and £5s and bank deposits—and provide a small discount to the rare customer that wants to pay with £50s.

It's entirely possible that the majority of retailers will not bother offering any discount whatsoever on £50s. This would effectively undervalue the £50 note. Gresham's Law tells us that given this undervaluation, the £50 will disappear from circulation as it gets hoarded under people's mattresses. For the regular British citizen, never seeing £50s in circulation probably won't change much. And anyone who does want a £50 can simply advertise on Craig's list for one, offering a high enough premium to draw it out of someone's hoard.

In closing, a few caveats. The figures I am using in this post are ballpark. It could be that a policy of freezing the supply of £50 notes allows the Bank of England to get to -2%. But maybe it only allows for a level of -1.75%, or maybe it slackens the bound so much as to allow a -2.5% rate.

Haldane mentions that the Bank of England could need 3% of headroom to combat subsequent recessions. But as Tony Yates has pointed out, in 2008 bank officials calculated that a -8% rate was needed. The Bank could get part way there by not only embargoing the £50 but also the next highest value note; the £20. But that probably wouldn't be enough. As ever smaller notes have their quantities frozen, this starts to intrude on the lives of the people on the street, making the policy more costly. If it needs to slacken the lower bound in order to allow for rates of -8%, I think the Bank of England should be planning for a heftier policy like Miles Kimball's crawling peg. After all, when the sort of crisis that requires such deeply negative rates hits, the last thing we should be worried about is disturbing a few conventions. Until another 2008-style crisis hits, embargoing large value notes might be the least intrusive, lowest cost option. 

---

*Of these policies, I think Miles Kimball's plan is by far the best one.
**Specifically, the Bank would only print new bills to replace ripped/worn out bills. Otherwise the outstanding issue will wear out and become easier to counterfeit. As for Scotland, which issues 100 pound notes, their quantity would have to be fixed as well.
*** I first mentioned the idea of embargoing large notes in relation to the Swiss 1000 CHF note, and later elaborated on it in the Lazy Central Banker's Guide to Escaping Liquidity Traps.

Fedcoin

Recent posts by Adrian Hope Baille and Sina Motamedi have got me thinking again about the idea of the Federal Reserve (or any other central bank for that matter) adopting bitcoin technology. Here's an older post of mine on the idea, although this post will take a different tack.

The bitcoin ethos enshrines the idea of a world free from the totalitarian control of central banks. So in exploring the idea of Fed-run bitcoin-style ledger, I realize that I run the risk of being cast as Darth Vader (or even *yikes* the Emperor) by bitcoin true believers. So be it. While I do empathize with the bitcoin ideal—I support freedom in banking—I rank the importance of bitcoin-as-product above bitcoin-as-philosophy. And at the moment, bitcoin is not a great product. While bitcoin has many useful features, these are all overshadowed by the fact that its price is too damn volatile for it to be be taken seriously as an exchange medium. This volatility arises because bitcoin lacks a fundamental value, or anchor, a point that I've written about many times in the past. However, there is one way to fix the crypto volatility problem...

Enter Fedcoin

Setting up the apparatus would be very simple. The Fed would create a new blockchain called Fedcoin. Or it might create a Ripple style ledger by the same name. It doesn't matter which. There would be an important difference between Fedcoin and more traditional cryptoledgers. One user—the Fed—would get special authority to create and destroy ledger entries, or Fedcoin. (Sina Motamedi gives a more technical explanation for how this would work in the case of a blockchain-style ledger)

The Fed would use its special powers of creation and destruction to provide two-way physical convertibility between both of its existing liability types—paper money and electronic reserves—and Fedcoin at a rate of 1:1. The outcome of this rule would be that Fedcoin could only be created at the same time that an equivalent reserve or paper note was destroyed and, vice versa, Fedcoin could only be destroyed upon the creation of a new paper note or reserve entry.

So unlike bitcoin, the price of Fedcoin would be anchored. Should Fedcoin trade at a discount to dollar notes and reserves, people would convert Fedcoin into these alternatives until the arbitrage opportunity disappears, and vice versa if Fedcoin should trade at a premium.

As for the supply of Fedcoin, it would effectively be left free to vary endogenously, much like how the Fed currently let's the market determine the supply of Fed paper money. This flexibility stands in contrast to the fixed supply of bitcoin and other cryptocoins. The mechanism would work something like this. Should the public demand Fedcoin, they would have to bring paper dollars to the Fed to be converted into an equivalent number of new Fedcoin ledger entries, the notes officially removed from circulation and shredded. As for banks, if they wanted to accumulate an inventory of Fedcoin, they would exchange reserves for Fedcoin at a rate of 1:1, those reserves being deleted from Fed computers and the coins added to the Fedcoin ledger.

Symmetrically, unwanted Fedcoin would reflux to the central bank in return for either newly-created cash (in the case of the public) or reserves (in the case of banks), upon which the Fed would erase those coins from the ledger. The upshot is that the Fed would have no control over the quantity of Fedcoin—it would only passively create new coin according to the demands of the public.

Apart from that, Fedcoin would be similar in nature to most other cryptoledgers. All Fedcoin transactions would be announced to a distributed network of listening nodes for processing and verification. In other words, these nodes, and not the Fed, would be responsible for maintaining the integrity of the Fedcoin ledger.

Why implement Fedcoin?

The main reasons that the Fed would implement Fedcoin would be to provide the public with an innovative and cheap payments option, and to provide the taxpayer with tax savings.

The public would enjoy all the benefits of bitcoin including fast transaction speeds, cheap transaction costs, and the ability to transact almost anywhere and with almost anyone as long as all parties to a transaction had a smartphone and the right software. At the same time Fedcoin's stability would immediately differentiate it from bitcoin. No longer would users have to fear losing 50% of their purchasing power prior to making a transaction.

Fedcoin's distributed architecture would be both complementary and in many ways superior to Fedwire, a centralized system which currently provides for the transferal of Fed electronic reserves among banks. I won't bother getting into the specifics: see this old post.

By introducing Fedcoin, the Fed would also lower its costs. While I haven't done the calculations, I have little doubt that running a distributed cryptoledger is far cheaper than maintaining billions of paper notes in circulation. Paper currency involves all sorts of outlays including designing and printing notes, collecting, processing and storing them, as well as constantly defending the note issue against counterfeiters. A distributed ledger does all this at a fraction of the cost. As Fedcoin begins to displace cash, and I think that this would steadily happen over time due to its superiority over paper, the Fed's costs would fall and its profits rise to the benefit of the taxpayer.

Fedcoin would have no impact on monetary policy

Fed officials might balk at giving the idea a shot if they feared that adopting a Fed cryptoledger would impede the smooth functioning of Fed monetary policy. They needn't worry.

The Fed currently exercises control over the price level by varying the quantity of reserves and/or the interest paid on reserves. The existence of cash doesn't get in the way of this process, nor has it ever gotten in the way. Bringing in a third liability type, Fedcoin, the quantity of which is designed to fluctuate in the same way as cash, would likewise have no impact on monetary policy. The Fed would continue to lever the return on reserves in order to get a bite on prices while allowing the market to independently choose the quantity of Fedcoin and cash it wished to hold.

Well, almost none: Interest on Fedcoin and the zero lower bound

Ok, I sort of lied in the last paragraph. While it happens only rarely, there are times when cash does get in the way of monetary policy, and so would Fedcoin if it were implemented. If the Fed needs to reduce rates on reserves to negative levels in order to hit its price and employment targets, the existence of cash impedes the smooth slide below zero. With reserves yielding -2% and paper notes yielding 0%, reserves would quickly be converted en masse into cash until only the latter remains. At that point the Fed would have lost its ability to alter rates—cash doesn't pay interest nor can it be penalized—and would no longer be capable of exercising monetary policy. This is called the zero-lower bound, and it terrifies central bankers.

Fedcoin has the potential to alleviate the zero lower bound problem. Here's how.

As Fedcoin adoption grows among the public, cash would steadily be withdrawn. And while it might not shrink to nothing—the public might still choose to use some cash—at least the Fed would have a good case for entirely canceling larger denominations like the $100 and $50.

Consider also that it would be possible for interest to be paid on each Fedcoin  (unlike bitcoin and cash), the rate to be determined by the Fed. And just as Fedcoin could earn positive interest, the Fed could also impose a negative rate penalty on Fedcoin. This would effectively solve the Fed's zero lower bound problem. After all, if the Fed wished to reduce the rate on reserves to -2 or -3% in order to deal with a crisis, and reserve owners began to bolt into Fedcoin so as to avoid the penalty, the Fed would be able to forestall this run by simultaneously reducing the interest rate on Fedcoin to -2 or -3%. Nor could reserve owners race into cash, with only low denomination and expensive-to-store $5s and $10s available.

So by implementing something like Fedcoin, the Fed could safely implement a negative interest rate monetary policy.

(Lastly, monetary policy nerds will notice that the displacement of non-interest yielding cash with interest-yielding Fedcoin is a tidy way to arrive at Milton Friedman's optimum quantity of money, or the Friedman rule.)

The big losers: banks

Fedcoin has the potential to tear down the private banking system. Interest yielding Fedcoin would be able to do everything a bank deposit could do and more, and all this at a fraction of the cost. As the public shifted out of private bank deposits and into Fedcoin, banks would have to sell off their loan portfolios, the entire banking industry shrinking into irrelevance.

One way to prevent this from happening would be for the Fed to make an explicit announcement that any bank could be free to create its own competing copy of Fedcoin, say WellsFargoCoin. Like the Fed, Wells Fargo would promise to offer two-way convertibility between its deposits/cash/Fedcoin and WellsFargoCoin at a rate of 1:1 to ensure that the price of its new ledger entries were well-anchored. The bank could then implement features to compete with Fedcoin such as higher interest rates or complimentary financial services. Even as Wells Fargo's deposit base steadily shrunk due to technological obsolescence, its base of WellsFargoCoin liabilities would rise in a compensatory manner.

The resulting lattice network of competing private bank crypto ledgers built on top of the Fedcoin ledger would work in a similar fashion to the current banking system. Wells Fargo would make loans in WellsFargoCoin and take deposits of FedCoin as well as competing bankcoins, say CitiCoin or BankofAmericaCoin. Intra-bank cryptocoin payments would be cleared on the books of the Federal Reserve with reserves transfers over the Fedwire funds system, although Fedcoin might eventually take the place of Fedwire. A change in the value of Fedcoin or reserves due to a shift in monetary policy would be transmitted immediately into a change in the value of all private bankcoins by virtue of  the convertibility of the latter into the former.

Nor would it be necessary to start with Fedcoin and then introduce bankcoins. Why not begin with the latter and skip Fedcoin altogether? Why aren't private banks at this very moment switching out deposits and replacing them with cryptoledgers?

KYC: Know your customer

'Know your customer' regulations would make implementation difficult, but not impossible.

With bitcoin, the location of a coin (its address) is public but the identity of the owner is not. However, laws require banks to gather information on their customers to protect against money laundering. As these laws are unlikely to change with the advent of new technology, banks would probably require anyone wanting to use bank cryptoledgers to have an account with a regulated bank. This would not be too onerous given that most Americans already have bank accounts.  However, it compromises anonymity, one of the key ideals of bitcoin, since each coin would be traceable by the authorities to a real person.

Perhaps there is still a way to preserve some degree of anonymity. Historically the Fed has always been spared from KYC rules since it has never had to document who uses cash. By grandfathering KYC exemption to Fedcoin, any user who wanted to preserve their anonymity could use Fedcoin rather than any of the multiple bankcoin ledgers, just like today they prefer to use anonymous Fed cash rather than bank accounts to transact.

In summary

So that's a rough sketch of Fedcoin—a decentralized, flexible, and well-backed payments system that grants one user, the Fed, a set of special privileges and responsibilities. Feel free to modify the idea in the comments section.

And just so we are keeping tabs, these are the institutions that Fedcoin could eventually make obsolete: bank deposits, banks (unless the latter are allowed to innovate their own bankcoins), the credit card networks Visa and Mastercard, bank notes, Fedwire, and even bitcoin itself, which would be unable to compete with a stable-value copy of itself.

Bitcoin true believers may not like this post, but perhaps they can take something constructive from it. Fedcoin is one of the potential competitors in the distant horizon. Now is the time for the rebels to figure out how to create a stable-price version of bitcoin, before Darth Vader does it himself. Otherwise they may someday find themselves closing down their bitcoin startups in order to write code for the Empire.

---

Note: My apologies to readers for my having succumbed to the constant temptation to adorn all blog posts with Star Wars references.

Why the Fed is more likely to adopt bitcoin technology than kill it off

Paul Krugman has a recent post in which he casts bitcoin as a retrogression from our current fiat system. He's wrong about this. In the next few years, I put decent odds on the guardian of our fiat system, the Federal Reserve, adopting the very bitcoin technology that Krugman finds so dubious. Here is Krugman:

One thing I haven’t seen emphasized, however, is the extent to which the whole concept of having to “mine” Bitcoins by expending real resources amounts to a drastic retrogression — a retrogression that Adam Smith would have scorned.

Krugman has taken bitcoin's colourful jargon a bit too literally. It's best to think of "bitcoin" as a distributed ledger, or a record, and not as physical coin. And while Bitcoin miners do "mine", they're not performing a function that is analogous to gold mining. Rather, they're contributing to the tending and maintenance of the information that makes up the bitcoin ledger. The mining community listens for ledger transfer announcements, processes and verifies them, and then updates the distributed record. The reward for being the first to successfully add to the ledger is new bitcoin. Distracted by this reward, Krugman misses the underlying verification process it represents, thus drawing an incorrect analogy between bitcoin miners and gold miners.

It's better to think of a bitcoin miner as a gold assayer who verifies that a circulating gold coin isn't a fake, or, in our fiat world, as part of the verification process in a credit card payment. A bitcoin miner listens, processes, double checks, and polices the distributed ledger. Protecting a ledger is a valuable use of resources.

Rather than being retrograde, let's see how bitcoin technology could be adopted by the Fed.

The Federal Reserve owns one of the world's most important ledgers. The Fed's 3,000 or so member banks maintain accounts at the Fed. Put differently, they own allocations in the Fed's ledger. Every business day banks trade Fed ledger space amongst each other. The name for the system that facilitates these transfers is Fedwire and the name for the ledger space being transferred is "reserves".

The quantity and size of Fedwire transactions is breathtaking. In the fourth quarter of 2012, 33.8 million transfers were made with an average transfer value of $4.45 million. A total of $150 trillion in ledger space, or reserves, was exchanged.* Many types of transaction are carried out over Fedwire. When company A buys out company B for a billion dollars, the payment will probably be made over Fedwire. If the Fed conducts an open market purchase of $5 billion, it'll pay for that purchase by transferring ledger space over Fedwire. (The size of a single Fedwire transactions is currently limited to $9.99 billion). When John Doe wires a college $80,000 to pay for his daughters tuition bill, ledger space is being moved from John Doe's bank to the college's bank via Fedwire. The lifeblood of the US commerce pumps through Fedwire.

The Fedwire infrastructure is currently hosted at the Fed's East Rutherford Operations Center (EROC) at 100 Orchard Street, East Rutherford, New Jersey (see map below). While most of the world's attention is usually focused on the Fed's Washington headquarters at 20th Street & Constitution Avenue, the importance of Bernanke's office pales in comparison to 100 Orchard Street. All vital information pertaining to the Fed's ledger is maintained on computers at EROC. If a bank wants to transfer ledger space to another bank, the payment is routed to EROC where it is processed and the Fedwire database updated. This system is a hub and spoke system, with EROC serving as hub for its member bank spokes. In the picture at top, it is the system on the left, a centralized network.


View Larger Map

Because Fedwire is so important, it needs to be resilient. Should disaster strike at the East Rutherford hub, a secondary back up data center at the Federal Reserve Bank of Richmond is designed to resume Fedwire operationality 60 - 90 minutes later. A third backup center exists at the Federal Reserve Bank of Dallas.** The weakness of the system is that if the hub is destroyed (and the second and third backups) then the entire payments infrastructure disintegrates.

An alternative (and perhaps cheaper) way to build a resilient payments system would be for the Federal Reserve to adopt a bitcoin-style distributed ledger. The Fed's ledger would no longer be stored at EROC. Rather, all member banks would hold a copy. Much like a bitcoin miner "mines", a member bank would be an independent node responsible for helping to maintain the ledger's integrity. Should a bank want to exchange ledger space, the transaction would be announced to the network of member bank nodes who would in turn poll each other to verify the legitimacy of the transaction. Once a consensus has been arrived at, the payment would be processed and the Fed's ledger updated. As a condition of membership in the Federal Reserve System, banks would be required to act as verification nodes.

What would be accomplished is a decentralization of the information contained in the Fed's ledger. The ledger would be everywhere rather than at one spot. Transfers of ledger space would no longer be patched through the central processor at EROC but would be handled by a distributed network of cooperating nodes. Whereas the current hub and spoke system has two levels of redundancy, Richmond and Dallas, a distributed system has no central hub and therefore much more layers of redundancy. It would be very difficult to destroy it. Such a system is represented in our top chart by the network on the right, in which no entity is more important than the other.

All of this is an exercise in speculative economics, of course. But I like to think there's a grain of truth in it. Whether you agree with me or not on the possibility or likelihood of the Fed adopting a bitcoin-style distributed ledger as the basis for Fedwire, at least you'll see why Krugman has been too hasty in writing off bitcoin as a retrogression. Distributed ledgers are cutting-edge and will have many applications in the future.

---

PS. In the burst of attention over the last few weeks, the press and pundits have all become a bitcoin experts, just like they were Cyprus experts in the previous news cycle. There are three blogs worth reading that offer more stable and permanent bitcoin fare. Peter Surda wrote his thesis on bitcoin, so do go by and check his blog The Economics of Bitcoin. Jon Matonis has payments industry experience and has been following bitcoin for far longer than myself or any journalist. He blogs at The Monetary Future. Finally, Mircea Popescu owns and operates MPEX, the largest bitcoin stock and options exchange, and knows all the excruciating detail of the system's inner functioning. He blogs at Trilema.


* Fedwire statistics
**Payments, Clearance, and Settlement: A Guide to the Systems, Risks, and Issues by the General Accounting Office (1997) PDF