© Philipp Paech
Full version, including references: The Modern Law Review (2017) 80(6) MLR 1073-1110
Where blockchain networks are used to hold and transfer financial assets, such as securities, money or derivatives, they could become systemically important in the future, quite similar to traditional market infrastructures such as clearing and settlement systems or central counterparties. Their function in the market might at some future point in time place them among these critical infrastructures. Blockchain financial networks would provide a service that would not be easy to replace should they fail to function properly, as they would provide for the constitutive records of financial asset holdings, act as a repository for a variety of important data and as the platform on which smart-contract-based derivatives are executed. As networks linking a multitude of financial market actors, potentially of different types, they are also highly interconnected. For all these reasons, such networks are destined to become important in terms of financial stability once they have attracted a certain volume of assets and a critical number of users. It might, therefore, be necessary to regulate blockchain financial networks in order to ensure that they are resilient and do not contribute to systemic risk but, ideally, help to reduce it. There are a number of relevant aspects which I will address in turn below.
Operational soundness and software loopholes
The first concern is about the operational soundness and continuity of the relevant processes. Uncertainty as to the accuracy or availability of records or the correct execution of smart contracts could have significant repercussions for financial stability. The relevant hardware, ie the node-servers, and the individuals operating it are ‘distributed’ throughout the network, independent from each other and not centrally controlled – hence any concerns regarding integrity, availability, continuity, safety and accuracy relate to the software platform. Given its crucial importance for the nodes and their clients and for the market as a whole, there is a need for relevant regulation.
However, this issue extends far beyond the operational functioning of the network. A matter of equal importance is that the processing of transactions and the execution of smart contracts must result in the ‘correct’ or ‘true’ outcome. What is correct or true is not defined objectively according to absolute criteria obtained outside the network. Rather, the yardstick is a consensus among nodes on how transactions should be processed and records kept. This consensus is typically established when nodes join the network and thereby adhere to the rules determining the acquisition and disposition of assets and the execution of smart contracts on the network (hereafter referred to as ‘internal rules’). These rules are laid down directly in the form of a computer code; there are no ‘bylaws’ or similar documents in human language. The internal rules may also be changed following the relevant internal governance procedures.
However, there is significant room for trouble. The software programming and user expectations may diverge, either because of an unintended loophole, ie a planned software functionality that produces, in combination with other functionalities, unexpected results, has been created due to the sheer complexity of the software platform (as was the case with the Ethereum-DAO 50m USD ‘theft’, which did not, technically speaking, occur because of an illegal intrusion into the software but as a result of the exploitation of a previously undetected loophole in the software), or because of a ‘bug’, ie a straightforward programming error. Ex-ante regulatory measures to avoid such loopholes or bugs are important, also to ensure the transparency of the internal rules. However, loopholes and bugs can never be entirely avoided and they might affect all or significant parts of the assets held in the network. Therefore, systemic stability requires that ‘incorrect’ results in a blockchain financial network be prevented before they materialise or that there is at least a possibility to reverse such results. The programmers of Ethereum, to the surprise of many, were able to ‘reset’ past transactions and undo the abusive transfers. This approach obviously contradicts the original concept of immutable outcomes of blockchain-based transactions; however, as Ethereum has shown, it is necessary to protect the market at large from becoming hostage to a programming bug or loophole. Hence, it is questionable whether the original ideas regarding immutability will prove practicable as soon as blockchain technology is adapted for use in financial markets.
In any case, independently of the question of whether a blockchain financial network provides for the correct outcomes, it can contribute to systemic risk. Blockchain networks record the assets of their users. These assets are part of a highly complex risk management process in which every significant financial market participant is constantly engaged. Risk management is a central, integral part of capital requirements regulation, and hence a centrepiece of the framework that governs financial markets.
The main mechanisms used to mitigate risk are delivery-versus-payment, security or collateral, set-off, closeout netting and multilateral clearing of exposures. In addition, financial institutions hedge their market risks using derivatives such as interest rate swaps. In principle, all these mechanisms could be programmed into the functionality of a blockchain financial network as smart contracts. However, in practice the technical hurdles are immense.
The key difficulty is that risk mitigation spans different classes of asset: for instance, a simple delivery-versus-payment mechanism keeps a performance (eg a transfer of securities) on hold until the other party has likewise performed its part (ie made the corresponding cash payment), in order to release both at the same time, thereby eliminating the settlement risk. However, to do so requires both the securities leg and the cash leg of the transaction to occur in the same blockchain financial network, on pains of not being able to enforce the necessary interdependency with any certainty. Alternatively, if securities and cash were held in two different networks, both networks would need to be linked in operational terms.
The risk management of a financial institution is a highly complex thicket typically managed with the assistance of computer algorithms. Cash, securities, claims and derivatives are all inextricably connected through the mechanisms mentioned above, ie delivery-versus-payment, security, collateral, closeout netting, clearing and hedging. Therefore, modern risk management requires all these asset types and mechanisms to be available on a single network. Such a universal network would obviously raise questions of systemic risk in itself. The alternative to such a ‘leviathan’ would be to have several networks where different asset types could be perfectly and unalterably linked through these risk mitigation functions—however, the resulting set-up would probably be extremely complex, requiring a high degree of standardisation and interoperability so that ultimately such a meta-network of blockchain financial networks would resemble the current situation in terms of complexity and proneness to error.
Herding, flash crashes and supervisory stays
The unstoppable execution of transactions and smart contracts in blockchain financial networks might also have effects akin to systemic dangers provoked by the phenomena of ‘herding’ or ‘flash crash’. The term ‘herding’ describes the synchronised behaviour of significant parts of the market as a reaction to certain market events. For instance, all hedge funds active in a given market segment may sell assets in the event of a sharply falling market, thereby amplifying the offending price movement. In extreme cases, herding may be one of the causes of so-called flash crashes, where extreme devaluation of an asset occurs in a very short period of time without any change in the underlying economic parameters. This phenomenon is typically due to identical behavioural patterns of the decision-makers or, where investment or risk mitigation decisions are outsourced to machines, to the use of algorithms that produce identical outcomes.
The autonomous and unstoppable execution of transactions and smart contracts in blockchain financial networks may aggravate this phenomenon. Removing the human element entirely eliminates the last vestiges of elasticity in the behaviour of parties, which does to some degree exist in wholesale financial markets due to the generally relational character of contracting prevailing in this environment. Eliminating elasticity may be advantageous from a market efficiency point of view in good times, but may also amplify market distortions in times of crisis. Blockchain technology takes the ‘immediateness’ of market reactions to an extreme and may combine it with a high degree of interdependency of the various processes involved. This could, in addition, cause unwanted feedback loops, especially in relation to the operation of smart contracts that execute autonomously on the basis of market data automatically retrieved from data sources. As a result, a single significant change in the market may immediately trigger another strong market move, which may, in turn, set off the third one, and so on. Hence, there is a need to assess blockchain financial networks and the potential of smart contracts in the light of rules addressing flash crashes and algorithmic trading.
An additional issue is relevant in this respect: in order to be better prepared to prevent systemic risk caused by failures of banks or investment firms, recent legislation on bank resolution has established an administrative framework which is applied instead of judicial insolvency proceedings. Under this framework, supervisors are equipped with the authority to halt the execution of certain contract terms under certain circumstances and for a short period of time. The application of this ‘supervisory stay’ is intended to prevent the mass termination of derivatives and repurchase agreements in the event of failure of a bank or investment firm. Automatic, unstoppable execution of blockchain-based transactions would produce the exact opposite. In order to maintain the effectiveness of this administrative supervisory stay, the relevant authority would need to be provided with an ‘emergency stop’ function, enabling them to halt the automatic termination of contracts recorded in a blockchain financial network. Such functionality would need to be built into the smart contract itself, making the stay dependent on data input triggered by the relevant administrative decision. By contrast, it would not be possible to allow the termination to happen and then afterwards ‘reverse’ it. First, because the termination would wipe out many derivatives and repurchase agreements that were important for the relevant bank’s risk management and secondly, because it would be difficult to find counterparties prepared to offer new contracts to the near-insolvent party on economically viable terms.
Shadow banking risks and bubbles
The emergence of blockchain financial networks and smart contracts may also influence the investment decisions made by market participants. Individuals or corporations may use the blockchain financial networks to store value, exchanging financial assets held with intermediaries for financial assets held in a blockchain network, in particular because of a perceived smaller risk, lower cost or better return as compared to more traditional ways of holding. As such, a blockchain financial network could also assume functions resembling those typically performed by banks, notably that of storing money. However, only the banks’ clients benefit from the relevant safety nets, such as deposit guarantees and access to central bank money for liquidity support. Blockchain financial network nodes do not benefit from these safety nets. If they act as intermediaries for clients outside the network, these clients are only protected if the node is a bank and the clients’ holdings are deposits or assimilated to deposits. The negative impact of adverse events on the market as a whole may be amplified by the fact that retail customers could withdraw their savings from the traditional banking sector, thereby diminishing their liquidity base. Both phenomena may cause risks comparable to those produced by so-called shadow banking.
Taking this thought a step further, the use of blockchain technology and smart contracts may cause a false impression of zero credit risk, because smart contracts allow for the immediate and autonomous enforcement of collateral should the obligor fail to perform. Collateral takers might assume that they are free to take on higher exposures, for example, to lend more money, as there seems to be no risk of unenforceability of the collateral.This assumed certainty is risky in itself. Risk-takers will decrease their buffers in terms of reserves if they perceive a collateralised obligation to amount to zero risk. However, the use of more efficient technology alone is poor justification for increased leverage in the financial system or, in other words, for stretching the liquidity cover of financial institutions even more thinly.