RDP 2012-05: Payment System Design and Participant Operational Disruptions 1. Introduction

High-value payment systems are critical infrastructure for financial markets. To mitigate the systemic impact of a participant's default, most high-value payment systems now settle on a real-time gross settlement (RTGS) basis (Bech, Preisig and Soramäki 2008). But while RTGS eliminates credit risk between participants, it requires more liquidity since payments are settled individually. To limit the call on participants' collateral to secure additional intraday liquidity it is important that liquidity is recycled through the system efficiently. If an operational disruption results in a participant being unable to send payment instructions to the RTGS system for settlement, liquidity accumulates in that participant's account, forming what is known as a ‘liquidity sink’. Such a disruption in liquidity recycling can prevent other participants from settling their payments.

The design of RTGS systems varies significantly around the world. Many RTGS systems incorporate elements of net settlement systems to economise on liquidity. Such hybrid features have the potential to mitigate the systemic effect of participants' operational disruptions. Glaser and Haene (2009) suggest that a central queue, which stores transactions that have been submitted to the RTGS system until they can be settled, can reduce the size of the liquidity sink that results from a participant's operational disruption because the payments already queued by that participant can still settle.^[1] Liquidity-saving algorithms – such as the bilateral-offset algorithm in Australia's RTGS system, the Reserve Bank Information and Transfer System (RITS) – can also potentially reduce the value of unsettled payments that result from any liquidity shortage caused by a participant's operational disruption. This is because such a feature means that less liquidity is needed to settle payments. Furthermore, features that reserve liquidity for certain types of payments by limiting the liquidity available to settle other types of payments (such as the ‘sub-limits’ in RITS) tend to slow the flow of liquidity into the liquidity sink, which gives other participants more time to react to the operational disruption.

However, if the liquidity-reservation feature does not specifically target the participant with the operational disruption (i.e. if there are no bilateral limits) it can slow payments between all participants. By restricting the flow of liquidity, broad-based liquidity reservation may increase the value of unsettled payments. In addition, the presence of a liquidity-saving algorithm may result in participants committing less liquidity to the RTGS system, thus negating the benefit that these mechanisms might have during an operational disruption.

This paper analyses the effect of system design on the systemic impact of participant operational disruptions using a simulator developed by the Bank of Finland (‘the simulator’). These simulations use data from Australia's RTGS system, RITS. As RITS features a central queue with a bilateral-offset algorithm, as well as a sub-limits feature, it provides a rich dataset with which to analyse the effects of system design. The paper also investigates how hybrid features interact with participant reaction times, and how they may alter the relationship between the size of the participant with the operational disruption and the systemic impact of that disruption.

As with all simulation studies, the lack of an endogenous behavioural response means that the results should be interpreted with care. In particular, simplifying assumptions are made regarding different participant behaviours in response to a variation in system design.

The remainder of the paper is structured as follows. Section 2 provides an overview of the literature on system design and operational disruptions. Section 3 describes RITS and its hybrid features. Section 4 presents the methodology used to analyse the effect of system design on operational disruptions in RITS. Section 5 presents the results of the simulation and Section 6 concludes.

Footnote

In contrast, in an RTGS system that does not have a central queue (referred to in this paper as a ‘pure’ RTGS system), transactions that cannot be settled immediately are rejected and must be resubmitted by the payer institution, and therefore would be affected by an operational disruption at that participant. [1]