Crypto-Currency and the Problem of Intermediation
Political Economy9

Crypto-Currency and the Problem of Intermediation

This paper is also available on SSRN as a PDF.1 An earlier version was titled “Bitcoin and the Denationalization of Money”. This paper assumes a working knowledge of monetarism and unfamiliarity with Bitcoin. For a summary which assumes a working knowledge of Bitcoin and unfamiliarity with monetarism, see “How ‘Bitbanks’ Could Solve Bitcoin’s Volatility Problem” on Coindesk.

All theories aim at predicting the unknown, but there are few better tests of a theory’s mettle than its ability to accommodate the totally unexpected. Such has been the effect of the advent of crypto-currency on the various answers to the question, “What is money?” No economist before the dawn of the computer age could have predicted its emergence, and given the tiny overlap between economics and computer science, few could have predicted it even afterward. The novel form these monies take sheds clear light on certain aspects of the problem which hitherto were able to be evaded by clever economists – in particular the process by which a currency comes into widespread use, and the question of what, if any, monetary arrangements might come to pass in a laissez-faire setting.

To the common objection that crypto-currency has no “intrinsic value,” little more needs to be said than that this fact is true of every economic good, including gold. ‘Unique cryptographic hash’ has no more or less intrinsic value than ‘shiny metal rock’: the value of each is determined by familiar forces of supply and demand. Indeed, though it has no representation in physical space, many crypto-currencies (Bitcoin among them) exhibit properties more similar to gold than to similarly abstract fiat currency. This connection will be used to illuminate potential paths forward for Bitcoin, the current crop of “alt-coins”, and crypto-currencies in general.

The paper will begin with a brief technical overview of crypto-currency which will be necessary to understand its economically relevant properties. From there, drawing on the historical experience of gold, it will explore the technical, legal, and economic hurdles which crypto-currencies face in the future, focusing on the unique problems of financial intermediation for these currencies. Finally, some protocol-level alternatives to intermediation are considered, which might be able to mimic the salutary effects of financial intermediation on the money supply.

What Is Crypto-Currency?

A crypto-currency is a method of constituting virtual “coins”, and providing for their secure ownership and transaction using a cryptographic problem. This problem is designed to be easy to verify, but computationally difficult to arrive at a solution. Various crypto-currencies use different functions for this purpose, the most common being a hash-target, by which hashes are calculated so as to come in below a certain value.2 This hash target (i.e. the difficulty of the problem) is adjusted every so often based on the total computing power on the network. This has the advantage of keeping the time between solutions more or less constant. Other protocols, such as Primecoin, provide for the problem by the calculation of large prime numbers. In theory any hard-to-calculate but easy-to-verify function with easily adjustable difficulty would do.

This “proof-of-work”, though computationally intensive, is the method by which transactions are verified as unique and trustworthy. To incentivize participation, transactors can include a transaction fee which goes to the first user to successfully verify it. This fee is optional in Bitcoin, but mandatory in some others.

In addition, verifiers are rewarded by the network with a certain number of coins after successfully verifying a block of transactions. This process, called mining, is the means by which the supply of coins on a network is expanded, and the adjustable difficulty ensures that computing advances will not affect the rate of expansion. As might be expected, the marginal cost of mining (mainly electricity) tends to equilibrate to the marginal benefit.3 In Bitcoin’s case, the reward for mining halves every 210,000 blocks verified, leading to a supply path over time with a positive first derivative which diminishes discontinuously to zero.4 A great variety of supply schemes have been implemented by alternative crypto-currencies, some of which will be discussed in the following sections

A coin itself is constituted by its transaction history on the network, going back to the block from which it was mined. Each input into a transaction points to the output of a previous transaction. This history is kept track of by every computer on the network in a continuously updating record called the block chain – literally, a chain of transaction blocks to which newly verified blocks are added. As transaction records are public, anonymity is maintained only by keeping the account owners private. If there are competing block chains among different users – for example if two transactions are received in a different order by different users, or if someone attempts to forge a transaction – the protocol defines rules by which only one is accepted. This method is quite secure, the more so as the protocol gains wider currency. Bitcoin and its close relatives will prefer the longest block chain – that is, the one with the most computing power behind it. Thus, to forge a transaction, an attacker would have to make sure that his own block chain was longer than the legitimate one, requiring him to have at his disposal more computing power than the total of the honest nodes.5

Here we may briefly put to rest several recurring fears surrounding crypto-currencies. First, the proliferation of copycat currencies (“altcoins”) cannot be inflationary unless any protocol is a perfect substitute for any other.6 Whether or not they are substitutable in some technical sense, entrenched network benefits mean that copycat protocols will not displace or rival existing protocols without clear feature advantages. Where physical notes from one bank or another may fit equally well in a wallet (and might even both be Dollars), holding multiple crypto-currencies involves the technical inconvenience of operating on multiple disjunct protocols, plus the additional calculational inconvenience that the currencies float against one another in value.7

Second, though the protocol is indeed defined arbitrarily in software, it cannot be changed arbitrarily once created. Once a protocol comes into use, the control of its constitution depends entirely on continued trust in the developers. Each user must be persuaded to upgrade. Thus, contra Grinberg (2012), the fact that “most users would use [a] new version of the software because of their trust in the development team” does not make the development team “the de facto central bank of Bitcoin”. “Bank of issue”, in fact, would be a broader category (private, non-central banks have also issued currency throughout history), a safer claim, and still wrong. The power of a bank of issue consists in its ability to issue new currency indistinguishable from (or in the case of a central bank, with the same legal tender status as) the old, and hence to gather seigniorage. The Bitcoin development team has no such power.8 Given the precedent of open-source crypto-currency protocols, trust in the developers is reasonably attributed to their continuing trustworthiness: a malicious update will be easily spotted and ignored.

The Money-ness of Crypto-Currency

What makes crypto-currency money? Mises’ regression theorem (1996, pp. 408ff) explaining the emergence of money, by which a particular commodity gradually overcomes network hurdles and becomes accepted as money by virtue of its increasing liquidity, has lately been taken by some armchair Austrians to imply that crypto-currency cannot be money, because it was never accepted as something useful of its own account. This is a curious argument, considering Mises saw money as a category of human action. In other words, something is money when people use it as money – that is, as a medium of indirect exchange. And in this sense, crypto-currency clearly qualifies. The regression theorem begins with the fact of a money and reconstructs the history by which it became such. It cannot be used in the other direction, starting with the observed history of a commodity and passing judgement on its moneyness.

If the theorem is interpreted strictly so as to demand some non-monetary terminus at the end of the regression, we could say that the anti-authoritarian quality of a decentralized currency was a consumption good for its initial adopters. It would be less of a stretch, however, to say that non-monetary (industrial) use is a possible (and so far the most historically significant) terminus, but not the only conceivable one. There are many ways that a commodity might gain wide enough initial acceptance to function as money. History, in this case, rules out any theory which denies to crypto-currencies the possibility of being money.

As for why crypto-currencies might be used as money, it is easy to see how they fit the textbook qualities of a useful commodity for indirect exchange:

  1. Portability. Crypto-currencies excel here because they have no extension in physical space. They can be exchanged using any device on which you can carry your “wallet file”, and it is no more difficult to send an amount across the world than across the street.
  2. Durability. Crypto-coins can be “lost”, but will not get worn out or depreciate.
  3. Divisibility. Bitcoins are divisible to eight decimal places. In principle there is no technical limit to the divisibility a protocol might allow.
  4. Security. As noted above, protocol-level theft and counterfeiting is extremely difficult, though – obviously – the protocol has no special way to prevent more traditional types of theft and fraud such as social engineering.

In addition to these “intrinsic” characteristics, money commodities will tend to possess some economic characteristics as well, such as liquidity (ready acceptance), saleability (wide acceptance), and stability of value. It is on these points, rather than the first four, that crypto-currencies have borne the most criticism.

On acceptance, though Bitcoins cannot yet be spent at the grocery store, there is a significant and increasing number of online merchants who do accept them, and they have already broken into the physical world with the advent of Bitcoin ATM kiosks in various large cities in Canada and the US. A vibrant and growing niche ensures that Bitcoins remain quite liquid. With large Bitcoin exchanges accessible online, Bitcoins can be bought and sold nearly instantly at the market exchange rate from anywhere with internet access.

On stability of value, however, crypto-currencies reveal their inadequacy as day-to-day currency. Bitcoin, despite the advantage of size, has suffered from frequent jumps and crashes since its inception in 2010.

volatility fig 1. Daily percent change in the Bitcoin-Dollar exchange rate, 2010-20149

The daily change in the USD/BTC exchange rate has reached nearly 50% in both directions, and frequently hits 20%. By contrast, the daily change in the USD/EUR exchange rate over the same period never exceeded 2.5% in either direction. This is not simply a problem of scale which can be expected to diminish as the volume of transactions grows: the volatility during Bitcoin’s time in the $100-$1,000 range (roughly the right third) is hardly less than its volatility in the range of a few cents (roughly the left third).

Milton Friedman (1951, p. 207) characterized the countercyclical effects of a strict commodity currency (i.e. a system, like the Bitcoin ecosystem currently, in which only the commodity itself is used as money) as depending primarily on an elastic supply of the money commodity. The supply of Bitcoins, however, is very nearly invariant to anything except time.10 In terms of the equation of exchange MV=PT,11 M is perfectly exogenous and predictable. Changes in the demand for cash balances, then, having no outlet in the money supply, must affect nominal spending until prices adjust. As this demand increases (i.e. V falls), nominal spending will likewise fall, hence the much-lamented “deflationary” aspect of Bitcoin.

In addition to a secular upward trend as the ecosystem grows, the demand for Bitcoins is also highly volatile. Though difficult to measure, it is well accepted that a very large portion of all Bitcoin transactions are made for speculative purposes – up to 90% by some estimates. It is true that investment spending is inherently more volatile than consumption spending, but more importantly in this case, demand shocks are self-reinforcing. Frankel & Rose (1995) describe the mechanism:

Expectations can be described as stabilizing when the effect of an appreciation today – relative to some long-run path or mean – is to induce market participants to forecast depreciation in the future. . . . Expectations can be described as destabilizing, on the other hand, when the effect of an appreciation is to induce market participants to forecast more appreciation in the future.

For currencies with a credible anchor – which does not necessarily imply a fixed exchange rate12 – speculation will tend to be a stabilizing force, pushing the actual value back toward that target. But on the other hand, without such an anchor, “the variability of exchanges will tend to multiply its magnitude and may turn what originally might have been a minor inconvenience into a major disturbance” (Hayek 1937, p. 64). More prescient words could hardly have been spoken of Bitcoin’s current situation.

For this reason, Bitcoin is by no means a unit of account, a situation which bears a striking resemblance to episodes of hyperinflation. Eichengreen (1994, p. 135), for example, quotes Schacht’s (1927) account of the German hyperinflation, noting that merchants “calculated prices with reference to the exchange rate and converted mark receipts into foreign currency as quickly as possible.” Because prices become nearly perfectly flexible, hyperinflations are remarkable as the most consistent example of short-run purchasing power parity (Rogoff 1996). Much the same occurs with Bitcoin today, and with even less friction. Merchants can convert their dollar prices into bitcoin using up-to-the-second data feeds, and services like Bitpay allow immediate exchange for local currency following a transaction, minimizing the risk to the merchant of holding a balance of bitcoins.

Though Bitcoin is by no means hyperinflationary, evidently the “safety valves” that spare the volume of transactions much of the brunt of fluctuations in the exchange rate come into operation in response to any sufficiently violent and sustained price or exchange rate paroxysms. Should merchants begin to adopt Bitcoin as a unit of account, however (let us suppose there is a long enough quiet period of relative stability for this to happen), prices will not be able to adjust as quickly to changes in nominal spending, and the familiar output effects of monetary disturbances will manifest themselves. The extant crop of crypto-currencies, then, will hardly be able to supplant the currencies on which they depend as an anchor for the speedy adjustment of prices.

Bitcoin and Gold

This is, however, no death knell for crypto-currency. It will be illuminating to consider the historical development of another successful money commodity with a mostly exogenous global supply: gold. It is true that fluctuations the price of gold did sometimes spur prospecting and more intensive output from gold mining, but this effect was both unpredictable and slow.13 Likewise, a jump in the price of Bitcoins brings more computing power to bear on the network, but because of the adjustable hash target, this affects only the distribution of coins, not the long-term rate of coin generation.

To answer how Bitcoin and other crypto-currencies with a rigidly capped nominal supply might overcome the problems associated with it, we ask then: how did gold (and other metals) overcome these problems?

Before the explosion of economic growth in the West, rigidity of the supply of precious metals was not a serious problem,14 and people could get by carrying actual metal coins with which to transact. Much local trade was in barter, and without well-integrated international markets, those money prices that did exist did not vary much from day to day in normal circumstances. Economic growth was slow and mining not all that productive, so “natural” inflation and deflation occurred over very long periods when they occurred at all. Even the so-called Price Revolution, during which prices in Europe more than doubled following the discovery and importation of New World silver and gold, took place over such a long period that the annualized inflation rate stayed in the low-by-modern-standards range of 1 to 1.5% (Kugler & Bernholz 2007).

But the explosion of economic growth starting with the industrial revolution was accompanied by a banking revolution as well. Moneylending lost its stigma, and with the development of financial and actuarial innovations, trade and industry were able to flourish. The observation that financial development generally precedes industrialization can, at least for this era, be explained by the happy circumstance that these financial innovations – fractional reserve banking in particular – were able to overcome gold’s nominal supply problem. As Mises (1953, p. 298f) noted of this time period,

The development of the clearing system and of fiduciary media [i.e. circulating bank liabilities] has at least kept pace with the potential increase of the demand for money brought about by the extension of the money economy, so that the tremendous increase in the exchange value of money, which otherwise would have occurred as a consequence of the extension of the use of money, has been completely avoided.

Rather than affecting prices in the long run and the volume of transactions in the meantime, changes in the demand for cash balances, which were for the first time rapidly growing, could for the first time be accommodated by changes in the money supply. By using loans and credit to pyramid a larger stock of banknotes and deposits on top of a smaller and fixed supply of physical gold, the money supply was, to varying extents across different countries, able to roughly stabilize nominal spending even in the face of wide variations in the demand for money.15

We might expect similar financial innovations to precede an explosion in the use of crypto-currencies. They face, however, a number of unique hurdles for which the history of gold provides little guidance.

Hurdles to Intermediation

At this point it will be useful to distinguish the currency itself from its method of exchange. For most of the world’s monetary history, the method of exchange was generally hand-to-hand. The currency transition from gold coins (base money) to paper banknotes (inside money)16 that marked the advent of fractional-reserve banking was much facilitated by the fact that the same method of exchange that was suitable for one was just as suitable for the other. The two are, of course, wholly different commodities, each entailing a different sort of claim, even if they are denominated equivalently and exchangeable one for the other at a fixed rate. But besides the usually small risk of bank collapse, accepting and spending gold-backed banknotes had no serious drawbacks compared to gold, and required no special investment to begin. Indeed, an important advantage of banknotes is that they were less costly (more convenient) to use than gold coins with the same method.

The more recent advent of electronic payment, on the other hand, was a method of exchange transition. Though a similar process had existed for centuries with checks, electronic payment no longer involved the hand-to-hand transfer of an asset (such as a check). Telephone lines and bank software were now the method of exchange. Even though the currency was the same as before (dollar-denominated deposit balances), the change in method entailed a network hurdle and a fixed cost to join the network, as merchants had to install card readers next to their cash registers. Nevertheless, the benefits were clear enough that by now most merchants have joined the network: as of 2011, nearly two thirds of sales in the US were made with a credit or debit card.

In both cases, the transition was much aided by being limited to one aspect or the other. Those deciding whether to accept a new currency faced no initial cost to joining the network, and those deciding whether to change their method of exchange did not have to worry about whether to accept a new currency. Crypto-currencies had the misfortune of having to effect both transitions at once, and a unit of account transition on top of it all! The success of Bitcoin in spite of this triple hurdle is remarkable. But the hurdles are not all behind: the establishment of bank liabilities redeemable in crypto-currency will again have to make both transitions at once.

The necessity of a currency transition is implied in the establishment of fractional-reserve banking, in exactly the same way as the original transition from coins to banknotes. The essential problem here is trust. The necessity of a method transition, however, deserves a few more words. Because a crypto-currency protocol defines both the coinage and exchange of the base money, issuing banknotes on a fractional reserve basis requires more than simply adding parameters to coins. This would enable at best the establishment of a 100% reserve bank (which is somewhat the idea behind Open Transactions’ voting pools). A bank that wanted to vary its issue with demand would need to create its own (centralized) coinage and exchange mechanism – a new protocol – which would not be compatible with the original, even if its processing took place on the same block chain.17 Nor would one issuer’s notes be compatible with another’s. Merchants would need to implement the new protocol, facing a similar sort of cost to those merchants installing PIN pads to accept credit cards.

Technical innovations can, however, collapse the two method transitions. Issuers would find it in their interest to provide an abstraction layer which allowed the same apparatus to transact in both base money and banknotes. Open Transactions (OT) already facilitates secure cross-block-chain exchange of different crypto-currencies. If the issuer’s protocol can be administered on a block chain, OT can provide a framework for the relatively transparent use of multiple inside monies and base monies, and even the issue of notes backed by and denominated in a basket of crypto-currencies.

In fact, such issues could be denominated in any unit at all. Again, credit card networks are an illuminating precedent. These perform a similar function to our abstraction layer: merchants do not have to implement separate pads to accept Visa and Master Card, because there is a network that transparently routes the transaction to the appropriate firm. Merchants can even choose which cards they are willing to accept (Many refuse American Express, for example, which charges the merchant more per transaction in order to fund its rewards program). And the process is rendered transparent to consumers by being denominationally agnostic: each network routes Dollars just as well as Euros, and even automatically exchanges them, based on what the merchant takes.

Nonetheless, though an abstraction layer can lower the network costs of adopting crypto-notes, even if the transition can successfully be made, important qualities of the base money are inevitably lost by using inside money. OT can minimize the trust users have to put in the bank – for example, the possibility of absconding with deposits can be ruled out. But users must still trust it:

  1. To maintain redeemability at par, i.e. not to inflate. This possibility is the price of quantity-stabilizing intermediation. Against it there is no recourse but market competition, which would be easily subverted by the incautious application of any modern economy’s byzantine banking regulations.
  2. With their information. This exposes users to hackers or subpoenas. As Cavalcanti (2010) put it: “Although anonymity preserves money, it rules out all forms of credit.”

Considering the ethos of crypto-currency as a trustless ecosystem, the banknotes of a crypto-intermediary will succeed in the market only by sacrificing those qualities to which crypto-currencies owed their initial success.

In addition to the intrinsic network hurdles of establishing crypto-intermediaries, regulatory uncertainty and hostility also constitute an extrinsic hurdle in a way that they do not for the protocols themselves. Though governments around the world have targeted crypto-currency users (often under money-laundering regulations for individuals, and financial regulations for would-be intermediaries), their success has been mixed. The protocols cannot be targeted or shut down; the best that can be done is to pinpoint prominent individual users, a drawn-out and expensive process. This is the reason that Bitcoin has not succumbed to the same fate as the Liberty Dollar, and it is precisely the protection that will vanish with the rise of financial intermediation. OT can prevent the seizure of users’ assets, but it cannot prevent the forcible closure of any intermediary that becomes large enough to attract attention.

No doubt some degree of intermediation and crypto-note distribution could survive anonymously, even under the harshest of legal climates. But legal hostility severely limits the scale of such an undertaking. This highlights another potential economic hurdle, the closest parallel to which might be the “free-banking” era in the United States, where legal restrictions on branch banking limited the capitalization and diversification of the numerous note-issuing intermediaries (Dowd 1992). This left the banks highly vulnerable to seasonal fluctuations, and made the system as a whole notoriously unstable. Should concerted legal hostility lead to a similar market structure of crypto-intermediaries, the crypto-currency market could well become even more unstable than it is at present.

Finally, should a crypto-currency come to serve primarily as a reserve currency on top of which banknotes are pyramided, the centralization of transactions that would accompany this development increases the risk of protocol fraud. As noted earlier, the protocol’s security lies in the assumption that an attacker will never be able to out-work all the honest computers on the network. The use of notes has the effect of taking people off the original network and putting them on the bank’s own services, a process helped along by diminishing rewards to mining over time. If the majority of computing power on the base currency’s protocol comes to be controlled by a few banks, the generation of a fraudulent block chain would require only the hijacking of the servers of a few major intermediaries, or collusion among them – still a tall order, but an order of magnitude easier than hijacking enough individual miners to reach 50%.

More worryingly, if a bank were to at any point constitute more than 50% of the computing power on the network, it could in principle “manage” the currency by a combination of double spending and rejection of legitimate transactions. To this fear, two comforts may be offered. First, in historical situations where intermediaries were allowed most freely to compete, “there was no evidence of natural monopoly in [note] issue, nor even of a natural number of firms that could be called ‘small’” (White 1995, p. 146; cf. Selgin 1988, pp. 150ff). Nevertheless, history also shows that incautious or malevolent regulation, especially if coordinated globally, might easily create a situation in which one reserve bank held enough power to monopolize the block chain’s computing power. Here, though, the ease with which new protocols can be created works as a safety valve. When the original protocol is attended with such disadvantages as a rogue reserve bank, dissatisfied users can always flock not only to a new bank, but to a new protocol. And if the rise of intermediation had by this point been accompanied by the development of an abstraction layer, the costs of switching will be very low indeed.

Alternatives to Intermediation

Intermediation will determine the value of base money more than vice versa. Thus Hayek (1979) notes,

[I]t is an erroneous belief that the value of gold or any metallic basis determines directly the value of money. The gold standard is a mechanism which was intended and for a long time did successfully force governments to control the quantity of money in an appropriate manner so as to keep its value equal with that of gold. But there are many historical instances which prove that it is certainly possible, if it is in the self-interest of the issuer, to control the quantity even of a token money in such a manner as to keep its value constant.

Similarly with crypto-currency as base money, intermediation will tend to stabilize the its day-to-day value by stabilizing the demand for it, allowing fluctuations to be borne by changes in the supply of banknotes rather than by the volume of transactions in base money.18 The base money, on the other hand, influences the value of the banknotes only so far as convertibility (as, we presume, will be demanded of private intermediaries) disciplines them to maintain such a quantity of notes that the two values stay roughly on par.

What this means is that we might be able to design a protocol so that the total supply of coins behaves similarly to competitive coin-backed banknotes, without intermediation at all. In other words, a crypto-currency might be designed so as to mimic a monetary base other than gold. Bitcoin’s periodic halving of the mining reward is arbitrary, and several alternative mining schemes have already been devised. Given the hurdles discussed above – at least some of which are likely to be permanently intractable – it will be worth considering alternatives by which the flexibility of intermediation might be baked into the protocol, thereby evading the legal hurdles.

One idea which has been suggested is linking the proliferation of coins to some macroeconomic variable – for example unemployment or an exchange rate – as a central bank might target these variables. The relative ease with which new crypto-currencies can be created might suggest a wide scope for experimentation here. However, to link the proliferation of coins to some exogenous macroeconomic variable would be to give up decentralization and subject it to somebody’s will. Its dependence on data entered from outside means that in principle it can be modified arbitrarily, and targeted easily by hostile or well-intentioned governments.19 Such an arrangement would not differ essentially from rule-based central banking: the rule can be broken, as it has been in every historical example.

The transactions velocity of money, however – VT in the equation of exchange – could be calculated endogenously. Coins exchanged per hour, reckoned either as a rolling or a cumulative-decay average, would be very close to the textbook definition, and would involve little more than a series of queries on the block chain. We need not concern ourselves with the Sisyphean econometrics and data collection which central banks find necessary to guide their behavior, and without the financial sophistication which we are trying to obviate, there is no worry about different velocities corresponding to M0, M1, etc.20 The only problem will be to set the optimal period over which the average is calculated. Too long and the monetary base will be rigid in supply; too short and M will fluctuate chaotically with transient changes in V.

Having a protocol-endogenous value for velocity would allow us to target MV at the protocol level, a money-supply norm with growing support (e.g. Selgin 1997, McCallum & Nelson 1998).21 Though in principle any behavior of the money stock is compatible with monetary equilibrium so long as it is perfectly anticipated and adjusted for in prices (Gilbert 1957), Selgin argues that an MV target will require the fewest discrete price adjustments, and for this reason is best suited to engender monetary equilibrium in a world where adjustment is piecemeal and discontinuous.

Thus we set our sights on the elusive goal of a neutral money, one in which changes in the money supply cause no relative price changes. The problem, then, will be to set the net rate of proliferation of coins (ΔM) to offset changes in velocity. This can be done easily enough by a protocol that provides both for the proliferation and deletion of coins,22 varying the rates of each to match the ΔM implied by the current measure of velocity, in a mechanism similar to the dynamic adjustment of Bitcoin’s hash target to computing power on the network.

I elide the details of implementation, however, because the focus on macroeconomic aggregates gets us in fact no closer to a solution. Such a scheme would be form without substance – a macroeconomic cargo cult, lacking the cardinal function of intermediation: the channeling of liquidity to its most valued uses. Intermediation ensures a relatively quick diffusion of new money through the economy – otherwise we are left with the transmission mechanism of the idiosyncratic spending habits of the miners to whom the new money goes as a reward. This will necessitate spurious and self-reversing price adjustments as miners spend their funds.

This, of course, vitiates completely the vaunted neutrality of our currency. Historically, equilibrating changes in the quantity of money have propagated through financial intermediaries. A currency which adjusts the quantity without intermediation, by contrast, highlights the fact that money approaches neutrality only when quantity changes are borne through a loanable funds market, and only to the extent of the depth of that market. And, on the other hand, where changes in the quantity of money enter elsewhere than through a loanable funds market, transitional disequilibria will arise with greater severity, even if the quantity change is in the equilibrating direction.

A loanable funds market, where intermediaries can take bids from those most willing to bear those monetary changes, is more than a merely automatic act of channelling funds. Even if the protocol should also provide for a central clearinghouse to channel savings to those who might bid for it, individual to individual, the specialization and economies of scale that constitute the chief functional justification of banking as a separate industry is still lost. Individual entrepreneurs would scarce be able to take on the administrative functions of an intermediary, and still less miners, who are by no means the same sort of person who would excel as an entrepreneur.

Intermediation, then – and in particular, intermediation carried out by large-scale specialists – is not merely a means to achieve MV stability in pursuit of monetary equilibrium. Rather, intermediation appears to be necessary to take advantage of any benefits that MV stability ostensibly offers.

A Decentral Bank

Perhaps then, rather than mimicking the supply effects of intermediation at the protocol level, we could mimic intermediation itself at the protocol level. A number of protocols are capable of “smart contracts”, virtual addresses that can operate automatically based code written in this language.23 Such platforms are flexible enough to accommodate numerous concurrent uses, both monetary and non, including sophisticated financial instruments such as futures contracts, allowing a wide area for experimentation without forking the protocol. To what extent, then, could smart contracts facilitate or mimic financial intermediation?

The most straightforward method would be to build our economy as an application within the existing ecosystem of such a protocol. Coins would be colored by adding various parameters to it (as with the Mastercoin extension discussed above) to indicate its use in our particular application. The supply of such coins would be managed according to nominal spending stability by a privileged smart contract – let us call it the “decentral bank”. In addition to the code embodied in the bank itself as a smart contract, the protocol would also need two additional features for its successful operation:

  1. The rolling transactions velocity must be available as a variable to smart contracts, so as to calculate a target ΔM
  2. The decentral bank must be privileged so that no other smart contract can color or decolor coins so as to bring them into or out of the ecosystem.

The decentral bank would then stand as if overseeing a “small open economy” with a fixed exchange rate. It would have to hold reserves of uncolored coins, coloring and lending them (more on which below), or on the other hand purchasing colored with uncolored coins, as it aimed for its monetary target.

This, of course, is problematic without the unconstrained ability to “mint” coins. As colored coins are literally converted (and not merely exchanged) from the “foreign market” of uncolored coins, there is no seignorage, and in a growing economy, the maintenance of nominal spending would require the continual drawing down of reserves.

Thus, a more practical solution is to dispense with the idea of reserves and to have users maintain a nominal balance with the decentral bank, which may or may not be composed of actual coins (the decentralized maintenance of transferrable ownership records, as for example with Namecoin, is not fundamentally different in a technical sense than the decentralized constitution of spendable coins). This is similar to the approach taken with Zerocoin, which in the interest of enhanced anonymity has users spend from a system-wide pool of coins, rather than spend particular, traceable coins. This obviates the second (and more onerous) required feature above, and gives the decentral bank the ability to stabilize nominal spending without worrying about reserve balances.

Our decentral bank might be seen, then, as a central bank with a mechanical rule: though its rule is more complex than Friedman’s proposed k-percent rule, it will literally be, as he proposed, replacing the central bank with a computer (or more accurately, an algorithm, executed across many computers), and with the added advantage of an accurate and up-to-date parameter for transaction velocity. The bank would “lend” by distributing any positive target ΔM in return for bids, i.e. future commitments to pay. The simplest rule would be to publicly broadcast some interest rate schedule (the price of loanable funds), which would then adjust based on demand in order to reach the target ΔM calculated as a rolling average over some period of time – perhaps a week, in order to prevent interest rates from reacting too violently to individual borrowers.

The purpose of the decentral bank is to overcome the limitation that additions to the money stock must go to miners. Miners in this system must still be incentivized to keep the gears of the protocol turning, but there is no reason the sums accruing to miners couldn’t be dwarfed entirely by sums lent out to bidders, and the aforementioned injection effects largely avoided. Any price changes brought about by their spending will be sustainable; they will not need to be undone as their effects work their way through the economy. And second, our intermediary will not have to create a new and incompatible inside money to pyramid on top of the coins themselves (the monetary base; “outside money”). Because the protocol itself is highly abstract, the aforementioned abstraction layer is rendered largely unnecessary.

Nevertheless, an automated system lacks the ability to evaluate the profitability of projects for which loans are to be made. Accordingly there exists a tradeoff between usefulness and vulnerability: the more impossible default becomes, the less useful intermediation becomes. The decentral bank can take measures to ensure repayment using auxiliary smart contracts, or (since it maintains the balances itself) enforce payment absolutely. Ideally the bank would constitute only a pure intertemporal market, leaving risk unintermediated in order to protect itself from default risk, which could be systematically exploited by malicious debtors. But because risk is an inextricable part of any intertemporal transaction, it is unclear that automated intertemporal trade could (in conjunction with freedom of account creation) exist at all without commensurate vulnerability to exploitation. Control of balances can eliminate default risk to the bank, but beyond some margin, it can do this only by shifting the risk onto some other party, for example the recipients of borrowed funds, or to the economy at large.

This situation may seem similar to the problem of futures contracts and other financial instruments which are accomplished easily enough through smart-contracts. The differences will be instructive. Financial instruments are created and enforced through multi-signature transactions. The instrument is essentially a smart contract that holds a definite number of coins in escrow until some number of the initial signatories vouch that such and such state of the world has indeed come to pass, at which point the coins are disbursed appropriately. The contract allows for a trusted third (or fourth, or…) party to enforce the terms of the contract.

There are two important differences here. First, in these financial instruments, the coins are held in escrow, thus there is no default risk. At the end of the transaction somebody gets the coins. At no point are coins created or destroyed, in contrast to the gratuity with which a true bank must be able to create and liquidate money through loans. Second, and more important given the fact of default risk, is that the third signatory must be trusted by both parties to the transaction. This only pushes the problem back one step for the decentral bank, for how will it know which third parties to trust?

Financial instruments, as important as they are, do not constitute financial intermediation in the sense necessary to stabilize the value of the currency. Ultimately, formal anonymity removes any tool by which the decentral bank might prevent the market from unravelling.


The problem of automated, protocol-level intermediation, like all those considered heretofore, is a technical problem, and by no means insurmountable. Nevertheless, borrowing and lending introduces risk – and thus moral hazard – which cannot be mitigated without personal judgement, something that an automatic protocol would find it difficult, if not impossible, to emulate. This appears to put an upper limit on the financial sophistication a crypto-currency can support without supportive legal institutions. As it is, true intermediation on a scale sufficient to stabilize the value of the currency will have to fight not only in the marketplace for general acceptance separately from the base money, but also in the political realm for the privilege of operating unmolested.

While a case can be made that crypto-currencies will be able to alter political dynamics enough to gain the requisite freedoms, that analysis is far outside the scope of this paper. As it stands, though it would be an utterly quixotic task for governments to try to stamp out crypto-currencies, they are well-positioned to prevent the emergence of stabilizing financial institutions around crypto-currency ecosystems. Without these institutions, crypto-currencies will no doubt continue to innovate on a technical level, but will not be able to supersede more legally privileged and centrally issued currencies.

  1. I am grateful for helpful comments from Raymond Niles and Andrea Castillo. Errors and bad ideas are my own.
  2. What follows is a brief account of the proof-of-work paradigm. For the sake of brevity, many technical details are elided. These are unimportant for the purposes of the paper, but can be found in Nakamoto (2008). The details of the less common proof-of-stake paradigm (King & Nadal 2012) are somewhat different, but the security implications are largely similar. The economic implications are discussed in subsequent sections.
  3. The marginal benefit of mining will include expectations of the crypto-currency’s future price, so when its value is expected to rise, the electricity costs can exceed the value of the coins mined. That this comes to a very large amount of electricity is illustrated by the 2011 police raid of a Canadian Bitcoin miner’s home because his energy usage was consistent with the high-powered lamps used for growing marijuana.
  4. Starting with a reward of 50 coins per block, the supply path is described by the infinite sum Σ(210,000 × 50/2i), which comes to 21 million. Because the hash target adjusts to keep the time between verified blocks more or less constant, each period i is about four years.
  5. This is known as a 51% attack. While no attack has so far ever been successfully executed, the potential is a matter of some debate. Dowd & Hutchinson (2015) argue that Bitcoin’s mining algorithm is characterized by significant economies of scale, due to the fact that the cryptographic problem can be more efficiently solved by specialized GPU-based machines known as ASICs (Application Specific Integrated Circuits). In addition, because it is common for miners to join a “pool” in order to decrease the variance of their returns, the argument is that a malicious pool operator would be in a better position to pull off an attack. In 2014, many users abandoned the pool in order to prevent it from reaching 51% of the computing power on the network. Thus, the very publicity of the problem seems to have stymied it. Ghash responded by implementing measures to lower its own share. But if a competitive solution is not satisfactory, there are technical solutions as well. Other crypto-currencies (Litecoin, for example) implement a modified proof-of-work function that significantly diminishes the potential economies of scale to mining. Others, using proof-of-stake, dispense with arbitrarily intensive computation entirely.
  6. This is analogous to the discussion in Selgin (1988, pp. 42ff) in which even minimal note-brand discrimination will prevent undue credit expansion by any single competitive note-issuing bank.
  7. Crypto-currencies could become much closer substitutes to one another if intermediation arises with the help an abstraction layer through which various protocols can be treated similarly by buyers and sellers. This possibility is discussed below.
  8. An update that allowed for the issue of new Bitcoins beyond the 21 million limit would require changes to what clients consider to be valid blocks. Such backward-incompatible updates (or “hard forks”) have in fact been issued, though not entirely smoothly. But given the importance attached by existing users to a non-inflationary currency, any attempt to fudge the 21 million limit would in all likelihood fail to be adopted on a sufficiently wide scale to alter the blockchain. Even if the team did succeed in doing so, it would still be unable to collect seigniorage from the expansion without further changes involving the rewards to mining.
  9. Data is from the largest exchange at each point in time, so the collapse of one exchange (e.g. Mt. Gox in February 2014) does not introduce spurious volatility.
  10. The hash target adjusts about every two weeks (more precisely, every 2016 blocks) based on the time taken since the last adjustment, so sudden changes in computing power on the network can have transient effects on the rate of coin generation.
  11. Discussion at the level of the protocol makes total transactions, T, more appropriate (and, conveniently, more measurable) than sales of final goods, y. V will then refer to the more expansive VT rather than Vy.
  12. For currencies on the gold standard, the “gold bands” served as a credible focal point beyond which speculators would not ordinarily let the currency’s exchange rate move (Bordo & MacDonald 2012). In addition, Engel and Frankel (1984) find evidence that the U.S. was able to take advantage of this sort of speculation during the Federal Reserve’s brief stint with money supply targeting: increases in M1 were actually associated with appreciation of the Dollar, indicating a consensus expectation of compensating contraction in the future.
  13. Bordo (1984, p. 218 fn) finds the strongest supply response to deviations from the purchasing power trend of gold at a lag of 16 years. While this response is indeed in the equilibrating direction, it is practically useless for the short-term fluctuations with which we are here concerned.
  14. I will ignore the problem of debasement except to note that this is an important aspect where crypto-currencies are more resilient to political exigencies than precious metals.
  15. See Selgin (1988) for a theoretical account of the mechanics of supply adjustment through fractional reserve banking under a gold standard. Where this process was impeded, periodic crashes were the rule – see also Selgin (2010, p. 494) for a suggestive graph comparing the responsiveness of US and Canadian money supply to money demand. It is nonetheless a testament to the salutary effects of financial development that, impeded as it was in various ways in different places, it was able to set off a historically unprecedented explosion in economic growth throughout Europe and North America.
  16. For the sake of concision, I will refer hereafter to bank liabilities in general, including checkable deposits, simply as banknotes. Though the two gain different macroeconomic significance under a central banking regime where banknotes are base money, we are analyzing a hypothetical crypto-banking system as a free-banking regime in which movement from one to the other has no macroeconomically significant effects (cf. Selgin 1988, ch. 8).
  17. A block chain is essentially a decentralized database, so multiple protocols can operate side-by-side on the same block chain if they speak a sufficiently similar language. This allows smaller enterprises to take advantage of the volume of mining done on Bitcoin’s block chain. By itself, this does not diminish the hurdle of protocol incompatibility. However, see below on Open Transactions as an abstraction layer.
  18. Note, on the other hand, the dramatic increase in the volatility of the purchasing power of gold following its final demonetization at the collapse of the Bretton Woods system (Erb & Harvey 2013).
  19. There is an interesting proposal in the Ethereum white paper to use a decentralized data feed for this purpose. This feed would guarantee accuracy in a manner similar to a prediction market, by rewarding all entries within the 25th and 75th percentile – the assumption being that the true value is a natural focal point, and the median entry will be very close to it. This is certainly a more robust solution than simply trusting a single data source. Still, an attack to skew the median will be much easier to pull off than an analogous 51% attack on a crypto-currency’s block chain. And more importantly, even if successful, this would be tantamount to pegging its value to another currency, making it vulnerable to monetary disequilibrium exported from the currency to which it is pegged.
  20. Off-chain transactions (i.e. Bitcoin derivatives) have already made this somewhat untrue, i.e. a (shallow) pyramid already exists with on-chain Bitcoins at the base. But, for the political reasons mentioned above, their importance is still small relative to on-chain transactions.
  21. McCallum and Nelson actually advocate an NGDP target, which differs from our MV target only in practical considerations of measurement, 1) using the narrower income velocity Vy rather than transactions velocity VT, and 2) using the other side of the equation of exchange (Py) as the target. A price level target (as advocated by, for example, Yeager (1990)) would, unfortunately, require the same data-feed apparatus as targeting other external variables: the protocol can calculate the product PT endogenously (because it equals MV), but calculating a value of T in order to separate out a targetable value of P would require the construction of a price index – a difficult task without an existing taxation apparatus, fraught with the same specification and measurement issues that come with the construction of any other price index.
  22. Peercoin, for example, takes Bitcoin’s variable transaction fee and makes it a mandatory, fixed amount, which is then destroyed rather than going to the miner. Alternatively, a rate of nominal depreciation could be built into the protocol, amounting to something like a continuously stamped Gesell currency, as discussed by Keynes (1936, ch. 23, §VI).
  23. Ethereum, for example, is a protocol whose smart contracts are written in a turing-complete scripting language. It is, at the time of writing, still in the proof of concept stage, and its future as an actual platform uncertain. Open Transactions, on the other hand, is a blockchain-agnostic extension that brings more limited smart contracts to existing protocols, including Bitcoin.


  • 1


    Jun 08, 2011 at 18:26 | Reply

    Bitcoins are specie. They are things desired in and of themselves. We have so much mental baggage in the way that we try to categorize money (token, commodity, fiat,..) that it becomes hard to recognize things like bitcoins that challenge our initial assumptions (digital is token, hard is commodity).

    Bitcoins are a particular use pattern of numbers that emerge from the use of a cryptographic framework (Bitcoin). They are not tokens or symbols of something else. They are discreet entities with cryptographic rather than physical boundaries.

    I think both men would have ultimately welcomed Bitcoin. Rothbard especially so because of what it may foreshadow for the state.

    • 2

      Cameron Harwick

      Jun 08, 2011 at 19:16

      If you mean that the distinction between token and specie is “specious” (har har), then absolutely. If the subjective theory of value means anything, then “cryptographic electronic signature” is not inherently less valuable than “ooh, it’s shiny”. The important thing is just that it’s hard to create more.

      You’re absolutely right that there’s no reason Bitcoins couldn’t be used economically in exactly the same way as gold. If government pressure eases to the point where we can get stable formal institutions around it, I’ll predict fractional-reserve banks which issue their own “Bitcoin-backed” currency. This would operate along the lines of Selgin’s free-banking theory, and would provide a measure of price stability for prices in that currency, even if Bitcoins themselves remain deflationary. Bitcoin itself I would imagine then being used mainly for reserves, for the same reasons that banknotes eventually superseded gold coins in daily use (not just because they’re easier to carry). And considering all the hoarding worries around Bitcoin, it should be a relatively small leap from major hoarder to moneylender, and then bank.

      You may be right about Rothbard welcoming Bitcoin. But his followers at least, from what I’ve seen, have been pretty down on the idea. Mental baggage sounds about right, hah.

  • 3


    Jun 08, 2011 at 19:39 | Reply

    It’s good to see that some Austrians are taking the rise of Bitcoin seriously. This is a fine article on the subject, and that is saying a lot since so many of your peers are filling the Internet with rushed Bitcoin commentaries littered with error.

  • 4


    Jun 09, 2011 at 0:38 | Reply

    “it wouldn’t be very fair to fault any mid-century economist for failing to predict its possibility”

    True we can’t fault them for not predicting Bitcoin. However Bitcoin isn’t that different from a pure decentralized peer-to-peer system of valuing and trading rocks for example is it? The one distinction is that the inflation of Bitcoins are definitely capped at 21.

    Bitcoin is technically a more advanced currency, but in principle doesn’t seem that different than anything Rothbard or Hayek had seen historically.

    Am I wrong?

    • 5

      Cameron Harwick

      Jun 09, 2011 at 16:22

      You’re right, economically it should behave similarly to gold. As far as this is true, all their insights should remain valid. But it has other unique features as well, especially as regard politics, which would definitely influence the path we take towards a free market in money (which they both wrote about). Rothbard’s insistence that we need a gold dollar, for example, is looking less and less relevant as the dollar itself becomes less and less relevant.

      The main difference is, gold is easy to monopolize, and has few alternatives (silver? copper?). There can exist any number of equally good cryptocurrencies, however, and this is the practical strength of the idea. Particular cryptocurrencies can be monopolized, but it’s very simple to just use a different, more or less identical one. They can spring up faster than they can be monopolized.

      (I elaborate on this in the next post)

      So yeah, it’s a new form, perhaps more general, which still conforms to old economic laws, but has very different political implications.

  • 6


    Jun 25, 2011 at 4:52 | Reply

    Thank you for this great article.

    At last, someone with in depth knowledge about Austrian School Economics is taking Bitcoin seriously and approaching it with rigour and references to theory.

    This article also goes some way to explain why some ‘Rothbardians’ have been so reflexively hostile to the idea of Bitcoin.

    Bitcoin really is the best current option in the fight to destroy the State’s monopoly on money, and everyone should adopt it, even on the scale of a few tens of dollars, for that reason alone.

    If all the hundreds of millions of disaffected people out there all put $20 into buying a bitcoin, and ran the client full time the networks worth and integrity would be strengthened to a level that would put it way beyond the reach of the State, and any lie its media outlets tried to spread about it.

  • 7

    Jon Matonis

    Jun 26, 2011 at 11:59 | Reply

    Great focus on bitcoin from an Austrian economist. I particularly like your comment above about the subjective theory of value in relation to a cryptographic math puzzle versus a shiny element atomic number 79. It seems to me that both would have the same value on a deserted island.

Leave a Reply

More Content

About »

Hi, I'm C. Harwick, an economics PhD student in Virginia with an interest in monetary theory, web development, and folk music.

Care to know more? Read on »

Twitter »

May280:56"It was, in fact, a fairly typical study for the field of diet research. Which is to say: It was terrible."…
May2712:50.@dylanmatt Speculation is not zero-sum, so it’s not rent-seeking.

Design By Thrica Powered By Wordpress Hosted By Nearlyfreespeech No Copyrights
Veritas Pulchritudo Est