By Neville Arjani, Principal Research at Payments Canada
The research team at Payments Canada has been deeply involved in Modernization since the outset and it has been interesting to study the various initiatives going on around the world and their meaning for our journey in Canada. The team’s focus lately has been on the design of Lynx, our new core clearing and settlement system, and, in particular, how to strike an appropriate balance between efficiency and risk management. To assist with this line of research, we recently engaged the analytics team at Financial Network Analytics Ltd (FNA).
To fulfill the oversight expectations of the Bank of Canada, Lynx will adopt a real-time gross settlement (RTGS) model. While this model eliminates settlement risk between system participants, it is also expected to increase liquidity costs. To help offset this increased cost, we are exploring the use of Liquidity Savings Mechanisms (LSMs) to augment the Lynx model. Our work with FNA involves simulation-based research, drawing on actual and artificial Canadian payments data, to understand pending liquidity implications from Lynx and to experiment with LSM designs used in high-value payments systems around the world. Use of artificial data is intended to address potential behavioral adjustments on the part of participants in moving to a RTGS model.
Before I dive into the findings it’s important to understand the notion of intraday liquidity in high-value payments systems. Intraday liquidity refers to the ability of a system participant to meet its payment obligations to other participants in a timely manner as they become due. While some payment obligations are time-sensitive and must be met by a specific point during the day, others are non-time-sensitive and are usually just expected to be met sometime on the due date. There are two main sources of intraday liquidity for a participant — funds received in the form of incoming payments from others, or an intraday credit line. With the move to RTGS, the volume of required collateral and overall cost to secure intraday credit are anticipated to rise relative to current arrangements. Absent appropriate controls, capabilities and incentives, this could engender adverse behavioral outcomes based on a collective incentive for participants to delay their non-time-sensitive payments as long as possible to await incoming payments, rather than incurring the necessary intraday credit cost.
LSMs, such as central queues that perform matching of payment obligations between participants, are intended to address this concern. These queues employ algorithms that perform routine or ad hoc offsetting of queued payments so that, instead of having to fund each payment individually, only funds equal to participants’ net obligation amounts from the offsetting are required to process the entire batch of queued payments but all payments are settled individually. This contributes to a smoother intraday flow of payments, reduces the cost for participants, and creates an incentive to submit non-time-sensitive payments earlier in the day. Indeed, Bank of England research signals daily liquidity savings of 15-20 per cent through use of LSMs such as central queuing with payment matching.
The results from the research with FNA are quite interesting. An initial finding is that, absent a central queue, moving to a RTGS arrangement could increase participants’ aggregate daily collateral requirements by more than 40 per cent, on average, based on current payment submission habits in LVTS. The research demonstrates that a central queue employing one or more matching algorithms can substantially reduce this additional collateral requirement, without introducing material payments delay over current submission times. This is not surprising, and is why the vast majority of the high-value settlement systems around the world implement these and other LSMs to accompany the RTGS model.
That said, a second finding from the research is that LSMs are not a panacea. That is, greater intraday volatility of time-sensitive payments — which could emerge, for example, by introducing risk management measures like multiple settlement periods in other arrangements that utilize Lynx for settlement purposes — will certainly limit the capacity of any LSM to introduce efficiency for participants. Third, simpler queue-release algorithms such as bilateral offsetting and bypass-FIFO (‘First-in First-out’) appear to produce more significant improvements in the delay-liquidity profile of the system compared to more complex (and more computationally intensive) optimization algorithms. Fourth, and despite the above result, complex optimization algorithms should still be employed, e.g., to solve gridlock situations.
These insights have already proven to be beneficial in our ongoing dialogue with prospective vendors for Lynx. Looking ahead, our research team is keen to continue supporting the Modernization effort in this and other capacities, and to continue engaging with strong partners like FNA. Stay tuned for further insights!