Journal of Financial Market Infrastructures

Five years after the publication of the genesis block of the bitcoin (BTC) algorithm(1) is a good moment to assess the implications of such virtual currencies for financial market infrastructures. When one wants to get acquainted with a financial market infrastructure (FMI), it is customary to look at the average volume and value of daily transactions settled. For bitcoin those numbers were, in December 2013, 7 x 10^4 transactions/day and 2 x10^5 BTC/day, respectively. That is four to five orders of magnitude less than the figures for mature currencies such as the euro, where the equivalent statistics in 2012 were 4 x 10^8 transactions/day and €4 x 10^12 (assuming a conversion rate of 1 BTC D €102)/day, respectively.

The structure of the bitcoin infrastructure, however, is very simple and elegant: one uniform network of nodes where each node contains all the transactions up to the point in time when the bitcoin algorithm was last run by the operator of that node. This also means that there may be many operators/participants. The links between the nodes can be viewed as preserving the integrity of transactions: there is a link between two nodes if there is agreement about the validity of all transactions contained in those nodes. There is only one payment order type with four elements: the address of the payer, the address of the payee, the amount of bitcoins involved and a transaction fee, possibly zero (denoted in BTC as well). The effect of that "payment order" is that the payer pushes the amount of BTC to the payee, resembling the traditional credit transfer. The address of the payee can be created especially for each transaction with the purpose of preserving anonymity, although existing addresses may be reused.

In contrast, the euro payments infrastructure (or any other established currency worldwide) is very complex. The network consists of many different node types as there are many different substructures we usually call FMIs. A general network property of FMIs is that the network is scale-free, meaning that there are relatively few core nodes (nodes with many connections) and many periphery nodes (only a few connections). Such networks are vulnerable in the sense that, if a core node is taken out, the network may change drastically. This is the case if a large participant experiences an operational outage. The bitcoin network, however, is much more uniform in its distribution. Any node may be temporarily down without having any influence on the performance of the system.

Also in terms of availability (24/7 and worldwide open access) the bitcoin network trumps traditional infrastructures. At the time of writing this letter, block #289027 has been published, containing 131 transactions with a total amount of 142 BTC (or €95 000) and a transaction fee of 0.03 BTC. The main problem with bitcoin is, of course, its (mal)function as a store of value. One week ago those 142 BTC would have cost you only €78 000. That is too much of a variation to qualify as a stable currency but it will be very interesting to see how virtual currencies develop in the coming years.

In the meantime we keep focusing on the infrastructures that process real money. The topic of the first paper in this issue, "Is this bank ill? The diagnosis of doctor Target2" by Ronald Heijmans and Richard Heuver, is transaction level data inTARGET2. The aim of the paper is to develop monitoring tools that shed light on the liquidity situation of individual banks and of the money market in euros as a whole. They show that stressful events such as the default of a participating bank can be identified rapidly in the data, which is available with a lag of one business day. Turnover and price developments on the money market can also be monitored successfully.

The issue's second paper, "Network indicators for monitoring intraday liquidity in BOK-Wire+" by Seungjin Baek, Kimmo Soramäki and Jaeho Yoon, describes the network properties of an FMI as mentioned above in the bitcoin comparison. The FMI discussed here is the Korean interbank payment system (BOK-Wire+). The authors develop a new intraday liquidity indicator that compares banks' expected resources for settling payments in the remainder of the day with their expected liquidity requirements. They find that the payment system of the Bank of Korea has relatively more evenly distributed payment flows than interbank payment systems in other countries. The tools presented are especially suited for the day-to-day oversight of intraday liquidity in large-value payment systems.

In our third paper, "Assessing financial market infrastructures' systemic importance with authority and hub centrality", Carlos León and Jhonatan Pérez perform network analysis on Colombian financial market infrastructures. They consider the relative systemic importance of FMIs by implementing so-called centrality measures. Unlike standard centrality measures (such as degree and strength), authority and hub centrality allow for assessing financial market infrastructures' global importance despite the strictly hierarchical (ie, directed and acyclic) nature of the networks they comprise. In the Colombian case the systemic importance of the large-value payment system (CUD) and the sovereign securities' main settlement system and central depository (DCV) is verified because they are the foremost authority central and hub-central financial market infrastructures, respectively.

The fourth paper in the issue, "Central counterparty links and clearing system exposures" by Nathanael Cox, Nicholas Garvin and Gerard Kelly, focuses on links between central counterparties (CCPs). These links are still a relatively scarce property in the industry worldwide. Such links enable participants to clear positions in any linked CCP without needing to maintain multiple CCP memberships. A link may introduce two opposite effects on the exposures: if netting across CCPs is permitted, it will reduce exposures, but at the same time CCPs' exposures could increase as they become exposed to one another through the possibility of a CCP default. The authors model these two effects and assess them quantitatively in a number of plausible scenarios. Their models show that CCP links can reduce overall system exposure.

I hope you enjoy reading this issue of The Journal of Financial Market Infrastructures.

(Footnote 1)
This happened on January 3, 2009 establishing the first 50 bitcoins (see, for example, https://blockchain.info/en/block-height/0). The first transfer of bitcoins took place nine days later, as registered in block 170 (https://blockchain.info/en/block-height/170).

Ron Berndsen
De Nederlandsche Bank and Tilburg University