RDP 2012-06: The Impact of Payment System Design on Tiering Incentives 1. Introduction

Most high-value payment systems settle payments on a real-time gross settlement (RTGS) basis. This prevents the build-up of large interbank exposures, which would otherwise occur if high-value payments were settled on a deferred net basis. However, RTGS systems require participants to hold substantial liquidity in order to make payments. Central banks generally make liquidity available to RTGS participants on a collateralised basis, in which case participants incur an opportunity cost in obtaining liquidity because the securities posted as collateral cannot be used for other purposes.

Tiering – where an institution does not participate directly in the central payment system but settles its payments indirectly through an agent that does – is a significant issue for payment system regulators. On the one hand, tiering can reduce system liquidity needs because:

• payments between a tiered participant (client) and its settlement bank are settled across the settlement bank's books rather than sent to the central system; and
• combining the payment flows of the client(s) with those of the settlement bank may allow the settlement bank to fund more payments from receipts rather than from liquidity provided by the central bank.

On the other hand, tiering can increase both credit and concentration risk. Credit risk arises because the settlement bank and its client(s) are exposed to the failure of each other. Tiering, by definition, increases concentration in the RTGS system as more payment activity occurs through a smaller number of direct participants.

The degree of tiering varies across payment systems. The Clearing House Automated Payment System (CHAPS) system in the United Kingdom, for instance, is relatively highly tiered, with only 17 direct participants (not including the Bank of England) making payments on behalf of several hundred other institutions (CPSS 2012). In contrast, the US Fedwire system has a fairly flat payments structure, with several thousand direct participants. Australia's RTGS system, the Reserve Bank Information and Transfer System (RITS), also has a low level of tiering. While in the early days of RITS this was due to restrictions on tiering, these restrictions were relaxed in 2003 to allow institutions whose RTGS payments are less than 0.25 per cent of the total value of RTGS payments to settle through an agent.^[1] Since then, however, very few institutions have opted to settle indirectly. In 2008, around half of RITS's 67 participants were below the 0.25 per cent threshold and therefore eligible to settle indirectly, yet only 6 chose to do so. Given that the vast majority of eligible participants joined RITS prior to the relaxation of tiering restrictions, this may be because the fixed costs associated with becoming a direct participant have already been paid, or simple organisational inertia. However, the low level of tiering does raise the question of what drives participants' incentives to tier and whether aspects of system design reduce the incentive to tier in RITS.

This paper uses simulation analysis to explore the impact of payment system design on institutions' incentives to tier. Specifically, it tests the hypothesis that including liquidity-saving mechanisms in the design of an RTGS system reduces the incentives to use tiering to save liquidity. It also attempts to quantify the increases in credit and concentration risk that would occur if there were an increase in tiering in RITS from current low levels, and the effect of system design on credit risk. Finally, it discusses the relevant considerations in weighing-up estimates of the benefits and costs of tiering. This analysis is intended to shed light on the present level of tiering in RITS, as well as inform policymakers in regard to rules that restrict tiering.

The remainder of the paper is structured as follows. Section 2 briefly reviews the literature on the costs and benefits of tiering in payment systems. Section 3 provides an overview of RITS and Section 4 outlines the simulation methodology. Based on these simulations, Section 5 presents estimates of liquidity savings from tiering under different system designs. Section 6 presents estimates of the increases in credit and concentration risk that would occur if there was to be an increase in tiering. Section 7 discusses how the benefits and costs of tiering might be weighed. Section 8 concludes.

Footnote

See Australian Prudential Regulation Authority and Reserve Bank of Australia (2003) for more information. Tiering has always been allowed for low-value payments that are settled on a deferred net basis. [1]