Operational problems in banks--effects on the settlement of payments in Norges Bank.

• Author: Berge, Tor Oddvar
• Position: Report

1. Introduction

   Norges Bank is the central bank and the ultimate settlement bank in Norway. Interbank payments are settled in Norges Bank's settlement system (NBO), a real time gross settlement (RTGS) system. Such payments are often time-critical and of high value. Operational disruptions that impair the ability of participants to execute payments may therefore pose a threat to financial stability. In this article, we want to quantify the potential effects of such disruptions.

   In the past 10 years, the development of computational tools for simulating key functions of payment systems has made it easier to quantitatively assess the robustness of these systems. A number of recent international studies have employed simulation tools to measure the effects of operational failures (see Manning et al. (2009) for an overview). However, owing to differences in payment patterns, liquidity level and infrastructure design across countries, the vulnerability of the Norwegian system cannot be assessed simply by employing results from studies in other countries. The ability of a payment system to withstand an interruption is not only dependent on the amount of liquidity in the system as a whole, but also on the distribution of liquidity among banks. In addition, because of market-specific calendar effects, operational problems on certain days during the year will have greater consequences than on days with normal transaction flows.

   In this article, we use Norwegian payments data to simulate operational problems in each of the 21 banks that are active in the daily payment settlements in Norges Bank. By simulating the settlement of actual payment flows in the Norwegian RTGS system, we are able to quantitatively assess the systemic effects should operational problems render a bank unable to execute outgoing payments over a prolonged period of time.

   The rest of this article is organised as follows: Section 2 discusses operational risk and relates it to the key features of the Norwegian system for interbank payments. Section 3 presents the data and explains the simulation methodology employed in this article. The results of the simulations are presented in Sections 4 to 6. In Section 4, we describe the direct effects of an operational problem in one of the participating banks. As such direct effects do not take into consideration the potential chain reactions of problems for other banks, a system-wide perspective is taken in Section 5 with an analysis of the full systemic effects following an operational problem in one bank. This analysis is complemented in Section 6 and 7, first by the introduction of a reaction pattern for other banks and then with a discussion of the effects on the net settlements. The article's conclusions are presented in Section 8.

2. Operational risk and the Norwegian interbank system

   The literature on risks in payment systems has traditionally focused on credit and liquidity risk, cf. Manning et al. (2009). As most modern central banks have put in place measures to limit such risks (e.g. by the introduction of RTGS systems to limit credit risk, or throughput rules to limit liquidity risk), focus has also moved to operational risk. Operational risk is the risk of loss resulting from failed internal systems, from human error or from external events such as deliberate attacks or natural disasters. As large-value payment systems allow financial institutions to settle obligations stemming from financial market transactions, any disruption to normal payment settlement processing could constitute a threat to financial stability; cf. Bedford et al. (2004).

   Norges Bank provides an RTGS system for the settlement of interbank payments. (2) In this system, large value payments between banks and specially marked transactions are settled individually (this is referred to as gross settlement). Small value payments, such as consumer payments, are first collected and netted in an auxiliary system (NICS--Norwegian Interbank Clearing System) before being sent to Norges Bank for settlement. Payments for trades in shares, certificates, bonds and derivatives are also handled in auxiliary systems dedicated to such transactions (VPS--The Norwegian Central Securities Depository and Oslo Clearing) before the net positions are sent to Norges Bank for settlement.

   Potential disruptions to the normal settlement processes may stem from operational problems in the RTGS system itself, in any of the auxiliary systems, or among one or several of the individual participants. This article follows the literature on simulations of participant-level operational outages which render a bank unable to send payments. Such outages may be due to IT failures, administrative errors (a shortage of trained staff), lack of contingency measures, or external factors such as power or telecommunication failures, including disruptions stemming from natural disasters or from deliberate attacks.

   As liquidity conditions in RTGS systems are often benign, it is generally considered unlikely that an operational problem in a participating bank will lead to liquidity constraints for other banks, see e.g. Bedford et al. (2004), Schmitz et al. (2006) and Ledrut (2007) for the case in the UK, Austria and Holland respectively. However, in the case of the Swiss RTGS system, Glaser and Haene (2008) find that the effects can be large in systems with a high degree of concentration, such as in Norway. Lubloy and Tanay (2007) find that this may be the case even in liquidity-rich environments. Although the likelihood of operational disruptions remains small, it is difficult a priori to predict the adverse effects such a disruption may have. In this article, we therefore investigate the robustness of the daily settlements in Norges Bank to operational problems in each of the banks that are active in the system on a daily basis.

   [GRAPHIC 1 OMITTED]

3. Data and simulation methodology

   3.1 Data

   We have extracted a record of all transactions settled in Norges Bank's RTGS system between 4 January 2010 and 31 December 2010, 252 settlement days and 289 999 transactions in total. (3) Our dataset contains all gross transactions between banks and between banks and Norges Bank which were settled directly in Norges Bank. In the case of transactions which were settled on a net basis, we only consider the final net amounts transferred across bank's accounts in Norges Bank. Chart 1 shows that the daily value of all gross transactions was generally between NOK 100 and 200 billion, with an average daily value of NOK 175 billion. However, on several days in 2010, the value was much larger, reaching more than NOK 500 billion on five occasions. (4)

   The Norwegian interbank system is tiered in the sense that most large banks (21 banks) settle their net positions and gross payments directly at Norges Bank. Most small banks, however, participate via a private settlement bank. (5) Banks that use a private settlement bank may also settle payments directly through Norges Bank, but the volumes of such payments are low. The daily turnover in NBO shown in Chart 1 is almost exclusively driven by the 21 large banks which constitute the focus of this article.

   [GRAPHIC 2 OMITTED]

   In order to settle their payments in the RTGS system, banks need liquidity either in the form of deposits or unutilised borrowing facility at the central bank. We have therefore also extracted a record of individual banks' deposits and credit limits during the same period. This data is essential in order to investigate how quickly banks exhaust their available liquidity, and the extent of contagion in the system will depend critically on these variables.

   Chart 2 shows how the total level of liquidity on each day of 2010 is split between deposits and intraday credit. Norges Bank targets an overall level of deposits which is positive and sufficient to keep the short-term money market interest rates close to the key policy rate, see e.g. Akram and Christophersen (2011). To this end, it provides loans with a fixed maturity, normally between a few days and six weeks, and with interest rates determined through central bank auctions. (6) The distribution of deposits among banks thus depends on their bids in the auctions and the payment flows stemming from day-to-day activities and interbank borrowing and lending. As all loans are collateralised, the intraday credit facility is defined by the value of collateral pledged by the banks, after haircuts have been applied, minus the value of already utilised collateral for loans from the central bank. (7)

   [GRAPHIC 3 OMITTED]

   Throughout the day, banks need to resort to the overdraft facility in cases where their remaining deposits are not large enough to cover an outgoing payment. Chart 3 shows that this occurs routinely, with average intraday credit taken by banks at NOK 19 billion each day. On average, 7 banks resorted to this facility each day.

   3.2 Simulation methodology

   In our analysis, we have used a payment system simulator to replicate the functioning of the RTGS settlement process. (8) As input we use the historical record of all the settled transactions in the payment system each day along with the balance account data for each bank. After regenerating each settlement day with the historical transactions in the order they were settled, we are able to explore what would have happened if the flow of funds was interrupted by an operational problem in a single bank. We are also able to measure the consequences for the other banks if this bank fails to submit its payments into the system.

   Although our data reflects the structure of the Norwegian interbank system, we are not simulating an RTGS system with exactly the same functions as that of NBO. The main difference between the simulated RTGS system and NBO...