Thursday, February 26, 2015

Sweden and peak cash

The Swedes really don't like cash. First, consider that Sweden is the only country in the world that I'm aware of where reliance on paper money is in decline. Second, no country's central bank has produced a nominal deposit rate as negative as Sweden's, for as long. Yet even at -0.85% per year, Swedish banks who own those deposits haven't fled into 0% cash, providing some indication of the degree to which they hold banknotes in disdain.

ABBA won't accept paper

As the chart below shows, cash outstanding continues to grow in almost every country except Sweden. Japan and Denmark are the only countries that come close to pacing the Swedes, although both nations continue to show incremental growth in demand for banknotes. Even Kenya, where m-pesa has taken hold, shows strong cash demand.

Sweden reached "peak-cash" somewhere between 2007 and 2008. The reason for this change of heart is public preferences, not government diktat. The monetary authorities can only indirectly influence the demand for cash, say by introducing/removing various banknote denominations, or altering the quality of its note issue (say by making notes harder to counterfeit). By virtue of deposits being convertible into cash whenever the depositor desires (and vice versa), the allocation between cash and deposits is primarily up to the public, not the monetary authorities.

One theory is that Sweden become more law-abiding in 2008, thus reducing their demand for paper kronor. Cash is typically demanded by criminals and tax evaders to avoid creating a paper trail. That Sweden's underground element suddenly decided to go legit doesn't seem very plausible to me. Demographics is a more likely contributor to peak cash. As a nation's population growth slows, the demand for cash peters off with it. This can't be the entire explanation, however, since other countries that also suffer from poor population growth profiles (like Canada) show rising cash demand. This leaves technology as the most likely culprit. As electronic payment options improve, it makes little sense to endure the hassle of withdrawing and holding a small horde of dirty paper in one's wallet.

According to a MasterCard study, 89% of transactions in Sweden are cashless, compared to 80% in the U.S. Situation Stockholm, the street paper sold by homeless vendors in Sweden's capital, can be purchased with a card rather than cash, and while London's buses went cash-free earlier this year, bus fares disappeared several years ago in Stockholm. Unlike the U.S. and other laggards, Sweden has a near real-time person-to-person payments system called Bankgirot which has been active since 2011. Bank customers can download an app called Swish, which allows them to make immediate mobile payments over the Bankgirot network. In southern Sweden, Vicar Johan Tyrberg has installed a card reader to make it easier for worshipers to make offerings. And finally, despite having a hit song entitled Money, Money, Money, ABBA refuses refuses to accept cash at the ABBA Museum. Apparently ABBA member Bjorn Ulvaeus is leading a crusade against banknotes after his son's apartment was burgled twice.

No zero lower bound, at least not yet

As a second illustration of Swedish cash abhorrence, consider that no other central bank has maintained a negative deposit rate as low as Sweden's central bank, the Riksbank, for as long. The Riksbank reduced its deposit rate to -0.85% this month after having maintained it at -0.75% since October 2014. A few nations come close. The Swiss, for instance, reduced rates to -0.75% in January, as have the Danes—but both are behind the pace set by the Swedes.

The key point here is that with Riksbank deposits being penalized 0.85% per year, one would assume that that they'd be quickly converted into Swedish banknotes. Cash, after all, pays 0% a year, superior to -0.85%. But that hasn't happened. As the chart below shows, cash left on deposit at the Riksbank stands at around 150 million kr, roughly the same level it has been at for the last twelve months (and far above 2008-2012 levels).

Who keeps funds on deposit at the Riksbank? As the business day progresses and Swedes make payments among each other, banks who maintain settlement accounts at the Riksbank will find themselves in a surplus or deficit position. While those in surplus can elect to park their excess at the Riksbank's overnight deposit facility, they'll usually try to lend these positions to deficit banks in the interbank lending market or participate in Riksbank fine-tuning operations at the end of the day, both of which provide superior returns to the deposit rate. For whatever reason, Swedish banks typically leave a small portion of their surplus in the deposit facility, bearing the awful return on deposits for the sake of enjoying whatever conveniences the deposit facility offers.

That this 150 million kr in -0.85% yielding deposits hasn't been converted into cash is an indication of just how low Swedish opinion on cash has sunk. Consider the myriad number of costs a bank that wants to cash out its balance will have to incur. A Brinks truck must be hired in order to transport the cash to the bank's vaults. The cash must be counted, requiring a diversion of tellers' resources from other important activities. With the majority of Swedish bank branches having gone cashless, they may need to reinvest in handling infrastructure before they can take delivery of a truck full of banknotes. Next, that cash needs to be vaulted, which means displacing other valuables from being safeguarded (like a client's jewels), forcing the bank to forfeit a vaulting fee. Finally, the cash needs to be insured from theft. No wonder Swedish banks continue to use the Riksbank's deposit facility, even at a -0.85% rate; like the public, banks don't consider cash to be a convenient option.

Could Swedes one day re-embrace cash?

Swedes are proud of their move towards digital payments, but this trend could very rapidly come to an end. In an effort to hit its inflation target, the Riksbank may have to push interest rates even deeper into negative territory. At much lower levels, even the most cash-hating Swedes will have to re-consider their aversion to paper. The consequences could be significant. While only a small quantity of deposits are kept in the Riksbank's deposit facility, much larger amountsaround 30 billion kroneare invested at the Riksbank via overnight fine-tuning repos, which currently pay -0.2%. If the Riksbank reduced the fine-tuning rate much lower (say to around -1.5%), these repos would be rapidly converted into cash by banks. The public holds many hundreds of billions more worth of deposits at Swedish commercial banks. Should Riksbank rate reductions force banks to respond by ratcheting down their own deposit rates to -1 or -2%, how long before Swedes empty out their bank accounts, turning Sweden into a cash-only economy?

To avoid reverting to a cash-only economy at extreme negative rates, the Riksbank would do well to hitch itself to the cashless trend. One way to go about this without calling in all banknotes would be to reign in the number of paper products the central bank currently offers consumers, in particular high denominations of notes. Just stop allowing conversions into the 1000 krona note, and maybe the 500 krona note too, or consider canceling these large denominations outright. Not only would this reduce the central bank's printing costs, but it would provide more room for further rate cuts into negative territorywithout the threat of a mad dash into Swedish cash.

As for the rest of us...

As in Sweden, I'm pretty sure that cash demand in places like Canada and the US will eventually peak as continued advances in payments technology and a more rapid adoption of those technologies lead consumers to demand less of the stuff. Central banks might consider adapting to this trend ahead of time by reducing the number of paper products they offer to the public. Do the Swiss really need a 1000 SFr note? Do Europeans need a €500 note, and Canadians $100 notes? Alternatively, why not do what Bill Woolsey advocates? Let's gradually privatize the issuance of paper currency. If anyone can make cash relevant again, it's innovators in the private sector. And if they can't, then maybe the banknote deserves to die a slow death.

Secondly, Sweden shows that the so-called zero-lower bound isn't actually at zero, but some distance below that. Cash is awfully burdensome, as evidenced by Swedish banks who are willing to hold deposits at -0.85% despite the option to earn 0%. Central bankers at the Federal Reserve, ECB, and elsewhere would do well to heed this Swedish data point. If they need to loosen monetary policy in order to hit their targets, they can go well below -0.5% before having to fear mass conversion into cash. The world's central bankers have much more interest rate ammunition than they let on.

Tuesday, February 17, 2015

A lazy central banker's guide to escaping liquidity traps

For lazy central bankers, this post describes three lite strategies for getting interest rates below the zero lower bound. Rather than requiring drastic action, these methods can be quickly deployed without having to spend too much energyleaving plenty of time for the afternoon squash game.

1) Let's start at the beginning. What is the zero lower bound? If a central bank reduces interest rates below 0% then banks will rapidly convert all their central bank deposits into cash. No point accepting a -2% return if you can get 0%, right?

2) The zero lower bound is a problem. From time to time, a central bank may need to venture into negative territory to hit its monetary policy targets. Cash impedes the smooth descent into negative territory.

3) We already have a few go-to plays for dealing with our inability to get below zero: quantitative easing, forward guidance, and fiscal policy, each with its own set of warts. While quantitative easing has become a popular tool over the last few years, theory tells us that purchases are irrelevant at the zero lower bound. Promising to keep rates at 0% for longer than is prudent has also been used by a few central banks, notably the Bank of Canada. Unfortunately the market finds forward guidance confusing and may see little credibility in it, given the fact that the central banker who initiates the promise may not be in office to carry it out. This problem is called time inconsistency. And lastly, while fiscal policy may be a good way to evade the zero lower bound problem, it hinges on flaky political processes and arduous negotiations.

4) A more direct way to get around the zero lower bound problem may be necessary. Our lazy central banker might have to make alterations to the very nature of cash itself.

5) A sure-fire way to remove the lower bound is an outright abolition of cash. It gets around the time inconsistency problem and the flakiness of politics. But abolishing cash is a drastic step. Banknotes serves a role in protecting privacy and are popular with the unbanked. Abolish cash and you hurt both. Given the degree of preparation and effort needed to remove cash, and the political wrangling this option would require, a lazy central banker may want to take a pass.

6) Rather than removing cash, just harm it. Silvio Gesell's stamp tax, for instance, attacks cash's pecuniary return. In this spirit, Miles Kimball's crawling peg between electronic currency and paper currency burdens those who own cash with a capital loss. The crawling peg banishes the zero lower boundwithout requiring the drastic step of immediately removing all banknotes. It's an elegant solution, you can read the details here (pdf).

7) There are a few drawbacks to a crawling peg. Driving a wedge between paper and electronic currency creates two different sets of prices at the till, one for deposits and the other for cash. A chocolate bar, for instance, might have a sticker price of $1.00 in electronic money, but require a cash payment of $1.05. This will be confusing and inconvenient for shoppers, necessitating an expensive and costly education campaign by our central banker. According to Kimball, instituting a crawling peg requires that a nation enact a unit of account switch. Prices must be set in terms of electronic currency, not paper currency, otherwise the central bank will lose control over the price level. While a switch in standards is by no means impossible, it does require time and effort.

8) Which finally gets us to our lite strategies for lazy central bankers. These options don't suffer from time inconsistency or flaky politics. They get us below zero without requiring the abolition of cash, nor do we get two different sets of prices at the till, nor do we need a nation to switch to a new unit of account. In short, if enacted, they'd keep our system pretty close to the current system.

9) Laziness isn't without a cost. Unlike the abolition of cash and Miles Kimball's crawling peg, the lite methods don't free us entirely of the lower bound. They only soften it up a bit, re-situating the bound a few percentage points lower. This buys room for central banks to cut rates,  but not infinite amounts. If extremely negative rates are necessary, say -6%, then there is no lazy option: best get off the couch and go with a full-out crawling peg.

10) There are a number carrying costs on cash holdings, including storage fees, insurance, handling, and transportation costs. This means that a central bank can safely reduce interest rates a few dozen basis points below zero before flight into cash begins. The lower bound isn't a zero bound, but a -0.5% bound (or thereabouts).

11) The various lite strategies all exploit the fact that differences in carrying costs among the various note denominations mean that banknotes are not naturally fungible. Put differently, bills aren't perfect substitutes for each other. Large denomination notes, say $100 bills, incur lower storage and handling fees than small denomination notes like $10s. After all, a hundred-thousand $10 bills (worth $1,000,000) take up ten times more storage space than a ten-thousand $100s (also worth $1,000,000). However, a central bank renders the two types of notes equivalent by offering to convert pesky low denomination $10s into sleek large denomination $100s at no cost to the owner. This means that the public can avoid the nuisances of small denomination note storage, for free. So at any point in time the note-owning public is bearing the carrying cost of the highest denomination note, not the lowest ones.

12) To get a bite, the following three lazy techniques all boost the carrying cost of cash.

13) They do so by interfering with the traditional smooth switch out of small denomination notes and deposits into large denomination notes afforded by a central bank. The effect is to put an end to banknote fungibility. The public, previously sheltered from the hassles of holding pesky low denomination notes, must now bear those costs, while being barred from racing into sleek high denomination notes.

14) By implementing any one of these lite techniques, the additional carrying costs now imposed on cash remove any incentive to convert increasingly negative yielding deposits into banknotes. A central bank that had previously reduced its deposit rate to, say, -0.5% before finding itself snug against the lower bound, will now be able to reduce its deposit rate to a much lower level, say -2.5%, without fear of mass flight into cash.

15) The first method a lazy central banker should consider is the abolition of large denomination notes. A central bank issues a proclamation giving people one month to bring in all $100s for conversion into ten $10 bills. Any large denomination notes remaining in circulation after one month will be disavowed. Once all high-value denomination are demonetized, the market clearing carrying cost on cash holdings will no longer be the superior rate on $100s, but the much heftier one on $10s. The expected return on cash holdings having been diminished, a central banker who had previously found him or herself stuck against the lower bound now has room to go lower without fear of mass flight into cash.

16) Even with the $100 having been abolished, the remaining low denomination notes in circulation can continue to serve a role in protecting privacy and serving the unbanked.

17) The second method involves closing the high denomination "conversion window." Specifically, a central bank ceases converting both low denomination notes and deposits into high denomination notes. The only window the central bank will keep open is between deposits and low denomination notes. This means that anyone who converts deposits into cash can now only get pesky small notes, forcing them to bear the higher carrying costs of $10s rather than the minimal inconveniences of sleek $100s. A central bank can now cut its deposit rate much deeper into negative territory than before since depositors are far less likely to flee into bulky cash.

18) If we close the conversion window, won't those who hold negative-yielding deposits and low denomination notes simply trade them for zero-yielding high denomination notes on the secondary market? Sure, but the opportunity will be a fleeting one. The closing of the conversion window effectively freezes the quantity of high denomination notes in circulation. The price of $100s will immediately rise to a premium over bulky low denomination $10s and negative-yielding deposits. After all, $100s impose much lower carrying costs than the other two instruments. This premium removes any incentive to flee deposits and low denomination notes.

19) Won't the public suffer the inconveniences of having two different sets of prices? Not really. Rather than having an electronic currency price and a cash price (as in point 7), the closing of the high denomination conversion window will create a combined electronic/low denomination price and a high denomination price. The public, which almost never transacts in high denomination $100s anyways, can conveniently ignore the high denomination price level.

20) Nor does our lazy central banker need to worry about switching standards. Given that consumers only rarely pay with high denomination notes, it's highly unlikely that retailers currently set prices in terms of $100s. In fact, even now retailers often refuse to accept large value notes. It's likely that we probably already live in a world with a low denomination/electronic currency standard.

21) Which brings us to our third method: vary the conversion rate between low denomination notes/electronic currency and large denomination notes. Central banks currently allow free conversion between deposits, low value notes, and high value denominations. The idea here is to keep the conversion window open, but levy a fee, say three cents on the dollar, on anyone who wants to convert either deposits or low denomination notes into high denomination notes. Conversion between low value notes and deposits remains free of charge.

22) A central banker can now safely guide rates to a much more negative rate than before, say to -2.5% rather than just -0.5%. Prior to instituting a conversion charge, the public would have fled from deposits to cash at such low rates. Now, while people can still convert deposits at no cost into low denomination notes, this offers them no real advantages given the high carrying costs on such notes. And flight into high value notes is forestalled by the conversion fee.

23) As with the second lite method, the third creates two different sets of prices: one for low denomination notes/deposits and one for high denomination notes. But this doesn't matter, see point 19. Nor do we have to switch standards, see point 20.

24) The main difference between the second method and the third one is that the exchange rate between high denomination notes and low denomination notes/deposits is allowed to float in the former versus being fixed under the latter.

25) The third lite method is akin to Miles Kimball's crawling peg, except that the conversion penalty is set on high denomination notes only, not cash in general. But if we steadily widen the peg so that it includes mid-value denominations, and then add small denominations, then the third lite technique isn't so lite anymore. It has basically become Kimball's peg. At some point along that transition, we start to inherit the inconveniences of the crawling peg (see point 7). For instance, the dual price level becomes much more inconvenient, especially once $10s (and lower) are included. However, the advantage is that we can now push rates much deeper into negative territory.

26) Which means its possible to incrementally transition from a lite program to an all-out option like a crawling peg or total abolishment of cash. Lazy central bankers may prefer to stick their toes in the water before jumping all the way in.

27) By the way, I've mentioned the first lite technique here, here, here, and here. I mentioned the second lite technique here. I haven't mentioned the third before.

28) If I was a lazy central banker, of the three lite programs I'd be partial to the second one; the closing of the high denomination conversion window. Removing high denomination notes from circulation would probably have messy political implications and draw the public's wrath. Levying a fee is an assertive, some might say aggressive stance necessitating the creation of new processes and administration expenses. Simply closing the $100 window seems like it would take the least amount of effort. It doesn't require that any new infrastructure or the decommissioning of existing machinery. As for the pricing of high denomination notes, this gets outsourced to the market.

Thursday, February 12, 2015

Slow greenbacks, fast loonies

Canada trails the U.S. is a common refrain, but not when it comes to payments. Courtesy of the Interac e-transfer service, Canadians have been able to make person-to-person (P2P) payments in real-time as early as 2002. By person-to-person, think email or mobile phone payments to friends, family, or your landlord, and by real-time, the receiver of a payment can immediately turn around and use those funds to buy something. By contrast, most Americans are still stuck in the nebula of three day delays when it comes to P2P. 

Why this incredible lag? I think it's for the same reason why the U.S. banking system is so much more unstable than the Canadian banking system. Whereas Canada has a small number of strong national banks, the U.S. has a large population of weak undiversified regional. This lack of size and strength renders U.S. banks prone to failure while simultaneously making it difficult for them to coordinate together in order to create shared-use systems. 

Part of the problem in providing a real time P2P solution to consumers is that U.S. banks can't use the Federal Reserve's existing small payment network, ACH, to do the job. ACH is a forty year old system that transfers funds with delays sometimes lasting as long as 3-4 days. That being said, the Canadian equivalent small payment system, the ACSS (run by the Canadian Payments Association) isn't much better, with settlement occurring the next business day. Yet somehow we Canadians enjoy real time P2P. 

Over a decade ago, Canadian banks decided to avoid ACSS altogether and set up their own proprietary network to provide real time P2P capability. Run by Interac, a bank-governed non-profit, the network processes P2P payments, nets them out across all banks, and provides instant communications among participants. At the end of the day, the banks settle balances owing and owed by trading Bank of Canada clearing deposits via the CPA's Large Value Transfer System (LVTS). Even before the banks settle among each other, Canadians will have instant access to funds they have received either via email or their smart phone (or, if they have been debited, lose access to these funds). Heck, Royal Bank even has real-time payments via Facebook. [1]

As anyone who has read the free banking literature knows, the U.S. has an awful history of bank regulation. Until recently, law makers forbade banks from setting up national branch systems, with unit banking being the norm. (Here is George Selgin on the topic). As a result, the U.S. is characterized by a patchwork quilt of banks, 6,891 in fact, with the top five banks accounting for only 56% of all deposits. Canadian law, on the other hand, never discouraged national branch banking. As a result, Canada has five dominant banks with broad exposure to all provinces and maybe two dozen smaller banks, the "big 5" accounting for at least 80% of Canadian deposits. 

You can understand now why it would be difficult for U.S banks to set up their own real-time payments system. In Canada, only a handful of bankers needed to be convinced that the time and effort to build a mutually beneficial system was worthwhile before the remaining minority followed. A much larger expense must be incurred in herding U.S. banks towards that same equilibrium. It's sort of like fax machine adoption. A single fax machine is useless, but the value of every fax machine increases as the installed base of fax machines grows, since the total number of people with whom each user can send/receive faxes rises. Ideally, everyone just agrees ahead of time to get a fax, or in the case of P2P, all bank decide to jointly build a shared network. Tough to do when you're a thousand squabbling voices. Enlightened cooperation among a few large banks, the Canadian solution, gets you there quicker.

The result is that in the U.S. most of the P2P solutions haven't been developed by banks, but by technology companies. Finance tech giants Fiserv and Fidelity National Information Services have developed their own networks; Popmoney and People Pay. Upstart Dwolla is trying to convince financial institutions to adopt its FiSync real-time service. This plethora of competing networks reminds me of what I've read about the early days of electrical utilities in the U.S., with multiple competing wire systems running down the streets. To avoid this sort of redundancy, some might say that the best option is to have a regulated monopoly like the Fed take the baton, say by upgrading ACH to real time. And with so many different competing systems, I can't help but wonder how they 'talk' to each other. If there were three or four brands of fax machines, and each brand could only receive its own faxes, how much less useful is the fax network?

So we Canadians have ubiquitous real time P2P and the Americans don't. However, the dark side to the Canadian system is that cooperation among the few needn't always be so enlightened. Just as the chiefs of the big 5 banks can get together in a back room and cobble together a mutually beneficial shared network, it's just as easy for them to set up a mutually beneficial pricing schemeat the expense of consumers. It costs $1.50 to do an Interac e-transfer. Sounds suspiciously high to me. 

[1] My source for information on the Interac e-payments system is the CD Howe's Mati Dubrovinsky, who briefly describes how the system works here.  

Sunday, February 1, 2015

The zero problem

The price of bitcoin is a capricious thing. Imagine that you've saved enough bitcoin to take your significant other out to a fancy restaurant. When the bill comes you discover to your horror that the price of bitcoin has crashed sometime between main course and desert. For the next few hours you're both stuck doing the restaurant's dishes. Far less embarrassing to choose dollars as your payment media at the outset given the unlikelihood of a dollar crash. This has always been one of bitcoin's main problems. The burden that a consumer must endure in absorbing bitcoin's incredible volatility until the time of payment outweighs any reduction in transaction fees that they might enjoy.

Or maybe not. Marc Andreessen recently posted a number of thoughts on twitter. The most interesting ones are #9 to 17, namely that bitcoin's fabled volatility needn't deter regular folks from using it as a cheap and fast payments mechanism.

In effect, it's possible to enjoy all of bitcoin's benefits without having to hold an inventory of the schizophrenic stuff. Consider that when merchants currently receive bitcoin in exchange for their product, their payments processor (say Bitpay or Coinbase) will instantaneously convert those coins into US dollars, thus sparing the merchant the risk of holding volatile bitcoin. As for shoppers, a service that allows them to purchase bitcoin in the instant prior to paying for a good would preclude them from having to bear the risk of a bitcoin crash. (1)

It's the "never-hold" approach to bitcoin. As long as just-in-time bitcoin purchases and sales are possible, shoppers and merchants can avoid bitcoin's worst feature, its volatility, while enjoying all of its best features, low fees and speed. These just-in-time services aren't free. Bitpay and Coinbase extract a fee for providing merchants with protection from bitcoin hyper-volatility, and a provider of shopper volatility protection would also expect to be compensated. Now I'm not sure how large these two fees would come out to. However, as long as the total cost is less than the fees levied by competing mechanisms like credit card networks, then bitcoin provides a net benefit to society. (2)

Touché, Andreessen. En Garde!

I've been talking about the potential for bitcoin to be displaced by stable-value cryptocoins as media of exchange for a while now. But if Andreessen is right (and I'm inclined to think he is) then who really cares if bitcoin suffers from +/-50% daily price changes? Whether it's worth $100 or $100,000, either way it serves regular folks as a superior last-second value transfer mechanism (subject to the above cost condition). We may not need stable-value cryptocoins after all.

But Andreessen is missing one of the larger points of the volatility criticism, which I'll call the zero value problem. Granted, we needn't care whether bitcoin is worth $100 or $100,000, but we do care if it is worth $0. While no categorical difference exists between any two given positive bitcoin prices, a categorical difference *does* exist between a positive price and a price of zero. Bitcoin works smoothly at any positive price, but it breaks down as value-transfer mechanism when it's worth nothing.

A key pillar of the volatility critique is that bitcoin's price earthquakes arise because the only players in the market are speculators. A more fancy way to say this is that bitcoin has no non-monetary demand. By non-monetary demand, I'm referring to that portion of an asset's total demand that would remain if prospective owners were notified that they could never sell that asset after purchasing it. Given this imposition, I sincerely doubt anyone would be willing to buy bitcoin. By and large, people only want the stuff because it can be got rid of in the future.

By way of comparison, gold has both monetary and non-monetary demand. There are consumers who will purchase the yellow metal knowing that they can never sell it again, say as jewelry or ornamentation. Same with an IOU like a stock or banknote. Because an IOU offers dividends (or a promise of cancellation at an attractive price), investors will be content to hold that IOU knowing that they can never resell it. This is the Warren Buffett approach to holding an asset, whereby one's favorite holding period is forever.

With the only folks holding bitcoin being future sellers, i.e. speculators, enter the zero value problem. An object whose value is purely speculative has no equilibrium price. In economics-speak, its price level is indeterminate. A $10,000 price is as good as a $10 price, or a $0 price. And that last price will inevitably arrivemaybe in 2015, maybe in 2020, maybe not till 2025when for some reason or other speculators all begin to get antsy at the same time. It could be something as innocuous as the belief that everyone else is about to sell (because they expect everyone else to sell, because they expect everyone to sell, etc). When a reflexive process like this begins, the only way for the bitcoin market to accommodate everyone's desire for an exit is for the price of bitcoin to hit zero.

For illustrative purposes, gold isn't subject to the zero value problem because if a speculative panic begins, consumption demand kicks in to anchor gold's price at some lower level. The same goes for central bank money.

At $1, bitcoin still works. But at zero, bitcoin breaks down as a payment mechanism. Since bitcoin no longer has a positive purchasing power, regular shoppers can no longer make just-in-time bitcoin purchases in order to consummate a transaction. Merchants will quickly pull bitcoin price quotes from their websites, unwilling to trade something (their wares) for nothing (bitcoin). Since the financial reward to mining will have disappeared, the process for verifying the blockchain may become tenuous. All the hard work put into building a payments mechanism will be gone in a few moments of speculative fervour. And what happens to all of the other "use cases" that Andreessen describes, like bitcoin apps and sidechains, when bitcoin hits zero?

Even if bitcoin hits $0 for an hour or two, won't the inevitable dead cat bounce fix the problem? Not necessarily. The best theory for how bitcoin rose above zero back in 2009 is the 'bootstrapping theory.' A small clique of insiders conspired to trade what was then an intrinsically-useless token among each other, generating a long enough history of positive prices so that bitcoin began to be accepted by naive outsiders at a non-zero price. From nothing, otherwise worthless tokens had pulled themselves up by their own bootstraps. Andreessen admit as much in his eleventh tweet.

The point I'm trying to make here is that if bitcoin were to fall to zero, a dead cat bounce isn't the natural next step. Rather, as in 2009, an outlay of time and resources would be required to fabricate a positive price. In essence, bitcoin would need to be re-boostrapped. But how to go about this process? Who would be willing to join the front line and risk their capital trying to trick the market into valuing bitcoin at a positive price again? Surely not all the former bitcoin millionaires. Keep in mind that it might take multiple efforts to jump start the system. And with bitcoin being so much more widely known than before, the re-bootstrapping process might take significantly more resources than it did in 2009. Finally, even if the system is successfully kickstarted after a few days, what about the damage that is incurred in the interim thanks to a period of inactivity? Could it do irreparable damage to bitcoin's reputation as a payments mechanism, in the same way that Visa would suffer if it went down for a few days?

How to overcome the zero problem

Luckily, there's a pretty easy fix to the zero problem. Set bitcoin's minimum price at US$1 so that it never has to go through a re-bootstrapping process.

To set a minimum price, bitcoin believers like Andreessen should consider donating US$21 million to a bitcoin stabilization fund. The fund will have a standing bid to purchase all bitcoin at $1. Since there will never be more than 21 million bitcoin in existence, the fund will have the financial resources to credibly support this price. In an extreme scenario in which all speculators run to the exits, the stabilization fund will be left holding 21 million bitcoin and no dollars. The good thing is that no harm will be done to bitcoin as a just-in-time value transfer system. The fund will make a market in bitcoin at $1, providing shoppers with an avenue to acquire the requisite bitcoin from the fund (say at $1.01) just prior to consummating a purchase, and providing merchants with a right to sell bitcoin at $1 in the moment after a sale.

The core idea here is that if speculators are for the moment unwilling to set a positive value for bitcoin, then someone else needs to. At some point after hitting the fund's $1 floor, speculators will likely gain enough confidence to once again take up the baton and drive bitcoin's price above $1. The roller coaster ride begins anew. If so, the stabilization fund's job is done, for the time being at least. It can start selling its hoard of bitcoin into the rally, replenishing its dollar reserves so that it can once again enforce the $1 floor should that necessity arise.

Think of the provisioning of a bitcoin stability fund as a public service. If bitcoin's promise is as enormous as folks like Andreessen believe, then the fund's $21 million price tag is a small cost to ensure that said promise isn't destroyed in a zero-value bitcoin scenario.

(1) I was going to point out that I don't think anyone is offering this service, but now I see that one is: Cryptosigma.
(2) If the combined cost of merchant protection and shopper protection +  bitcoin transfer fees are higher than the costs that banks and the card networks earn on transactions, then bitcoin may not be the panacea that everyone makes it out to be.