The Bank for International Settlements (BIS), the guys who are responsible for Basel regulations and act as the global central bank to all the central banks, caught global headlines yesterday by saying that bitcoin will break the internet. Wowser. I couldn’t ignore this and neither should you so here’s a summary of what they said in a 24-page PDF, part of their Annual Economic Report.
The chapter starts with a review of why and how money works from a historical perspective.
The quest for solid institutional underpinning for trust in money eventually culminated in the emergence of today’s central banks. An early step was the establishment of chartered public banks in European city-states during the period 1400–1600. These emerged to improve trading by providing a high-quality, efficient means of payment and centralising a number of clearing and settlement operations. Such banks, set up in trading hubs such as Amsterdam, Barcelona, Genoa, Hamburg and Venice, were instrumental in stimulating international trade and economic activity more generally. Over time, many of these chartered banks functioned in ways similar to current central banks. Formal central banks, as we know them today, also often emerged in direct response to poor experiences with decentralised money. For example, the failures of wildcat banking in the United States eventually led to the creation of the Federal Reserve System.
The tried, trusted and resilient way to provide confidence in money in modern times is the independent central bank. This means agreed goals: clear monetary policy and financial stability objectives; operational, instrument and administrative independence; and democratic accountability, so as to ensure broad-based political support and legitimacy. Independent central banks have largely achieved the goal of safeguarding society’s economic and political interest in a stable currency. With this setup, money can be accurately defined as an “indispensable social convention backed by an accountable institution within the state that enjoys public trust”. In almost all modern-day economies, money is provided through a joint public-private venture between the central bank and private banks, with the central bank at the system’s core.
And, after explaining how cryptocurrencies work and reviewing their features and capabilities as digital money, concludes:
Cryptocurrencies such as Bitcoin promise to deliver not only a convenient payment means based on digital technology, but also a novel model of trust. Yet delivering on this promise hinges on a set of assumptions: that honest miners control the vast majority of computing power, that users verify the history of all transactions and that the supply of the currency is predetermined by a protocol. Understanding these assumptions is important, for they give rise to two basic questions regarding the usefulness of cryptocurrencies. First, does this cumbersome way of trying to achieve trust come at the expense of efficiency? Second, can trust truly and always be achieved?
As the first question implies, a key potential limitation in terms of efficiency is the enormous cost of generating decentralised trust. One would expect miners to compete to add new blocks to the ledger through the proof-of-work until their anticipated profits fall to zero. Individual facilities operated by miners can host computing power equivalent to that of millions of personal computers. At the time of writing, the total electricity use of bitcoin mining equalled that of mid-sized economies such as Switzerland, and other cryptocurrencies also use ample electricity. Put in the simplest terms, the quest for decentralised trust has quickly become an environmental disaster.
But the underlying economic problems go well beyond the energy issue. They relate to the signature property of money: to promote “network externalities” among users and thereby serve as a coordination device for economic activity. The shortcomings of cryptocurrencies in this respect lie in three areas: scalability, stability of value and trust in the finality of payments.
First, cryptocurrencies simply do not scale like sovereign moneys. At the most basic level, to live up to their promise of decentralised trust cryptocurrencies require each and every user to download and verify the history of all transactions ever made, including amount paid, payer, payee and other details. With every transaction adding a few hundred bytes, the ledger grows substantially over time. For example, at the time of writing, the Bitcoin blockchain was growing at around 50 GB per year and stood at roughly 170 GB. Thus, to keep the ledger’s size and the time needed to verify all transactions (which increases with block size) manageable, cryptocurrencies have hard limits on the throughput of transactions. To process the number of digital retail transactions currently handled by selected national retail payment systems, even under optimistic assumptions, the size of the ledger would swell well beyond the storage capacity of a typical smartphone in a matter of days, beyond that of a typical personal computer in a matter of weeks and beyond that of servers in a matter of months. But the issue goes well beyond storage capacity, and extends to processing capacity: only supercomputers could keep up with verification of the incoming transactions. The associated communication volumes could bring the internet to a halt, as millions of users exchanged files on the order of magnitude of a terabyte.
Another aspect of the scalability issue is that updating the ledger is subject to congestion. For example, in blockchain-based cryptocurrencies, in order to limit the number of transactions added to the ledger at any given point in time, new blocks can only be added at pre-specified intervals. Once the number of incoming transactions is such that newly added blocks are already at the maximum size permitted by the protocol, the system congests and many transactions go into a queue. With capacity capped, fees soar whenever transaction demand reaches the capacity limit. And transactions have at times remained in a queue for several hours, interrupting the payment process. This limits cryptocurrencies’ usefulness for day-to-day transactions such as paying for a coffee or a conference fee, not to mention for wholesale payments. Thus, the more people use a cryptocurrency, the more cumbersome payments become. This negates an essential property of present-day money: the more people use it, the stronger the incentive to use it.
The second key issue with cryptocurrencies is their unstable value. This arises from the absence of a central issuer with a mandate to guarantee the currency’s stability. Well run central banks succeed in stabilising the domestic value of their sovereign currency by adjusting the supply of the means of payment in line with transaction demand. They do so at high frequency, in particular during times of market stress but also during normal times.
This contrasts with a cryptocurrency, where generating some confidence in its value requires that supply be predetermined by a protocol. This prevents it from being supplied elastically. Therefore, any fluctuation in demand translates into changes in valuation. This means that cryptocurrencies’ valuations are extremely volatile. And the inherent instability is unlikely to be fully overcome by better protocols or financial engineering, as exemplified by the experience of the Dai cryptocurrency. While engineered to be fixed to the US dollar at a rate of one to one, it reached a low of $0.72 just a few weeks after its launch in late 2017. Other cryptocurrencies designed to have a stable value have also fluctuated substantially.
This outcome is not coincidental. Keeping the supply of the means of payment in line with transaction demand requires a central authority, typically the central bank, which can expand or contract its balance sheet. The authority needs to be willing at times to trade against the market, even if this means taking risk onto its balance sheet and absorbing a loss. In a decentralised network of cryptocurrency users, there is no central agent with the obligation or the incentives to stabilise the value of the currency: whenever demand for the cryptocurrency decreases, so does its price.
Further contributing to unstable valuations is the speed at which new cryptocurrencies – all tending to be very closely substitutable with one another – come into existence. At the time of writing, several thousand existed, though proliferation makes reliable estimates of the number of outstanding cryptocurrencies impossible. Recalling the private banking experiences of the past, the outcome of such liberal issuance of new moneys is rarely stability.
The third issue concerns the fragile foundation of the trust in cryptocurrencies. This relates to uncertainty about the finality of individual payments, as well as trust in the value of individual cryptocurrencies.
In mainstream payment systems, once an individual payment makes its way through the national payment system and ultimately through the central bank books, it cannot be revoked. In contrast, permissionless cryptocurrencies cannot guarantee the finality of individual payments. One reason is that although users can verify that a specific transaction is included in a ledger, unbeknownst to them there can be rival versions of the ledger. This can result in transaction rollbacks, for example when two miners update the ledger almost simultaneously. Since only one of the two updates can ultimately survive, the finality of payments made in each ledger version is probabilistic.
The lack of payment finality is exacerbated by the fact that cryptocurrencies can be manipulated by miners controlling substantial computing power, a real possibility given the concentration of mining for many cryptocurrencies. One cannot tell if a strategic attack is under way because an attacker would reveal the (forged) ledger only once they were sure of success. This implies that finality will always remain uncertain. For cryptocurrencies, each update of the ledger comes with an additional proof-of-work that an attacker would have to reproduce. Yet while the probability that a payment is final increases with the number of subsequent ledger updates, it never reaches 100%.
Not only is the trust in individual payments uncertain, but the underpinning of trust in each cryptocurrency is also fragile. This is due to “forking”. This is a process whereby a subset of cryptocurrency holders coordinate on using a new version of the ledger and protocol, while others stick to the original one. In this way, a cryptocurrency can split into two subnetworks of users. While there are many recent examples, an episode on 11 March 2013 is noteworthy because – counter to the idea of achieving trust by decentralised means – it was undone by centralised coordination of the miners. On that day, an erroneous software update led to incompatibilities between one part of the Bitcoin network mining on the legacy protocol and another part mining using an updated one. For several hours, two separate blockchains grew; once news of this fork spread, the price of bitcoin tumbled by almost a third. The fork was ultimately rolled back by a coordinated effort whereby miners temporarily departed from protocol and ignored the longest chain. But many transactions were voided hours after users had believed them to be final. This episode shows just how easily cryptocurrencies can split, leading to significant valuation losses.
An even more worrying aspect underlying such episodes is that forking may only be symptomatic of a fundamental shortcoming: the fragility of the decentralised consensus involved in updating the ledger and, with it, of the underlying trust in the cryptocurrency. Theoretical analysis suggests that coordination on how the ledger is updated could break down at any time, resulting in a complete loss of value.
Overall, decentralised cryptocurrencies suffer from a range of shortcomings. The main inefficiencies arise from the extreme degree of decentralisation: creating the required trust in such a setting wastes huge amounts of computing *power, decentralised storage of a transaction ledger is inefficient and the decentralised consensus is vulnerable. Some of these issues might be addressed by novel protocols and other advances. But others seem inherently linked to the fragility and limited scalability of such decentralised systems. Ultimately, this points to the lack of an adequate institutional arrangement at the national level as the fundamental shortcoming.
So I guess they don’t rate cryptocurrencies for much in the overall scheme of things. The report proceeds to talk about blockchain and distributed ledger technologies, and says they’re good, and concludes with a debate around whether central banks should issue digital currencies. All in all a good read, but the bitcoinistas aren’t going to like it.