Strategic Review of Innovation in the Payments System: Conclusions
4. Payments System Architecture
Payments system architecture is the set of physical and logical structures that allow institutions to exchange payment instructions, initiate settlement and perform any additional functions associated with a payment. Payments system architecture is of interest to this Review for two reasons. First, it is relevant to how well and how easily the types of gaps identified in the previous section can be addressed. Second, some forms of architecture appear to be more conducive to innovation than others.
Traditionally Australian payment systems, such as Direct Entry, ATMs and eftpos, have been thought of as bilateral, with physical bilateral lines between pairs of participants, bilateral agreements about how messages are secured, transmitted and processed through those lines, and bilateral business agreements to exchange a given type of payment and potentially also pricing arrangements. By contrast, many systems around the world (and some operating in Australia) operate via a central hub, with all messages passing via the hub to the recipient. This is more efficient, as a new entrant needs only to connect to the hub, rather than establishing – a task that increases with the size of the industry.
Recently the distinctions between these models have been blurring as internet protocol (IP) technology allows communication to multiple counterparties via a single connection to a network. A recent example of this was the implementation of the industry Community of Interest Network (COIN) in Australia. While this reduced the number of physical network connections (from one for each partner, to one set for the COIN), the overhead of separate bilateral system connections (e.g. file transfer protocols, naming conventions, security connections etc.) between all participants remains. Business arrangements also need to be agreed, so governance becomes relevant. These elements could be agreed bilaterally, standardised through industry agreement, or governed by a central entity.
The Summary of Consultation paper identified three possible roles the Bank sees for centralised architecture in Australia:
- Simplifying connections. The greater the extent to which physical, logical and business arrangements are centralised, the simpler will be access for new participants and the better the environment for competition and innovation.
- Enabling additional processing functions, to provide more sophisticated payment products or greater interoperability. A hub could perform functions as varied as: splitting and distributing bulk files; conversion between message formats; fraud detection; the association of identifiers with account details (to facilitate easier addressing of payments or account switching); or holding data centrally that can be referenced in payment messages.
- Facilitating real-time, or near real-time settlement of retail transactions. Real-time or near real-time settlement of retail transactions would likely require centralised architecture to process large volumes of payments for settlement in RITS.
The Board believes that there should be a general presumption in favour of establishing more centralised systems when the opportunity arises, particularly in the establishment of new systems. However, for the purposes of the current Review its focus is on any architecture solutions that are necessary to address the gaps identified in the previous section. The Board considers that the possibility of advances is most apparent in two areas – the DE system and real-time payments.
4.1 Direct Entry Arrangements
The DE system is the key system for most business payments, providing bulk payments, such as salaries and regular bill payments, along with one-off payments. The system is well-established and inexpensive, but in most cases has attracted little investment in upkeep over the years.
Submissions to the Strategic Review have typically recognised the case for addressing the gaps identified in the preceding section and many have focused on the role the DE system would play in doing so. A number of submissions suggested that the DE system would not be the right starting point and that an entirely new system, dubbed DE2, was needed. They argued that this would be the least disruptive approach to both system participants and businesses, and would allow those businesses that had a need for new services to access them.
The Board concurs that addressing some of the gaps discussed above would require the establishment of a new system, as discussed in the next section, but it also believes that there is scope to enhance the services that can be offered by the existing DE system. The architecture that has been developed by the Reserve Bank to link the COIN with the SWIFT network for the exchange of bulk files means that all participants exchanging files link to the Reserve Bank. The Reserve Bank is considering enhancements to this ‘Low Value Clearing Service’ (LVCS) to perform additional ‘hub’ functions, including those that would allow the DE system to better meet some of the gaps identified above. For instance, the system could in the future be capable of accepting a single file that contains payments to be cleared with multiple institutions; the LVCS could also split and sort this into individual files for exchange and generate settlement instructions to RITS. If a new ISO 20022 message standard was adopted, the Bank would provide a translation service that converts ISO 20022 messages to the existing DE standard and vice versa. This has the potential to allow progressive migration to richer standards, reducing the cost of migration. Consideration would need to be given to how this migration could be managed, for both system participants and end users.
There would be potential over time for the Reserve Bank to deliver further processing functions for bulk files, as appropriate.
One issue that has been the subject of some discussion of late is the potential for same-day settlement of DE files. The industry is currently working towards same day settlement for the first three file exchanges each day. This does not cover the majority of the value processed through the system. In the Board's view, the industry should move expeditiously to same day settlement of all five exchanges. This will reduce risk and allow end users to receive access to funds without the need for an extension of credit.
4.2 Real-time Payments
The Board considers that enhancements to the DE system could deliver important benefits for the payments system at a relatively modest cost. However, this solution is only appropriate for bulk files, which by their nature are exchanged in relatively low volumes and are not time critical (e.g. payroll payments). The Board believes that an adequate solution for real-time payments will require a new system, based on real-time clearing of payment instructions. To complement this, the Reserve Bank is prepared to contemplate establishing a system for real-time settlement of retail transactions.
As discussed in Section 3, real-time payments require at a minimum real-time clearing, the capability for real-time posting of funds by institutions, and the ability for participants to provide credit. The latter constraint can be removed if real-time settlement is available. As noted in section 3, the Reserve Bank is prepared to extend interbank settlement hours as required to support payment system innovation.
The Board believes that the critical element of real-time messaging should logically be provided via a payments messaging hub (‘payments hub’), which could readily accommodate a large number of participants in an efficient manner. Importantly, a hub would also be able to provide additional processing services, including holding a database of account identifiers and potentially accommodating remittance data. Ideally, the governance arrangements for such a system would provide open access and allow decisions about the design and operation of the system to be made in an efficient manner.
The Reserve Bank believes that such a system could be linked to a system providing real-time settlement of retail transactions. The ability to do so would remove the need for participants to provide credit in order to provide funds in real-time to end users. This implies a reduction in settlement risk for participants compared with a deferred settlement system. A second benefit of the removal of the need to provide credit would be to facilitate participation by entities that would not be in a position to provide credit.
A real-time retail settlement system would be established as a separate module of RITS, receiving settlement instructions from the payment hub, which would process individual payment instructions from financial institutions. The system might settle using liquidity reserved by institutions for retail settlements, with exchange settlement account balances and the funds available for retail settlements updated periodically. This system would be available for the same hours of operation as the message hub.
There are potentially a number of ways that a payments hub or hubs could be delivered, which might encourage competition and innovation in different ways. A single hub could be provided, either as a commercial entity or as a utility, or multiple hubs could be provided commercially, linking to the settlement hub and competing with one another. The Board considers that universal access by account holders should be an important principle underpinning the system. Therefore, if multiple hubs were provided, this would need to occur in a way that provided interoperability. The Board's presumption is that a single hub would be most efficient.
For a single hub, the objective would be to provide open access to the hub's core functions at a reasonable cost. Participants would then be able to compete over the services that can be offered to end users through a variety of delivery channels. The hub could potentially support several different payment systems, all achieving real-time settlement. While it would be desirable for a basic credit transfer message to be available to all customers, the hub could potentially accept multiple message types to support different types of activities, provided that some core elements were present. This means that more specialised payment systems could also make use of the hub.
A single hub could be run on a commercial basis or as a utility. As suggested in one submission, a commercial entity might be more responsive to development opportunities over time, but might be inclined to exercise market power if it became successful. It is also possible that the establishment costs and the need to achieve sufficient network size to be viable might make the business case for a commercial entity difficult. This suggests Reserve Bank ownership and operation as one possible model. A hybrid approach suggested in some submissions is for the Reserve Bank to initially establish the system, which could subsequently be transferred to a private entity once it reached sufficient scale. The Board's preliminary view is that a hub of this type would be best operated as a utility, owned either by the industry or the Reserve Bank (with appropriate mechanisms for regulation and industry consultation, respectively). This provides the greatest opportunity for vigorous competition between financial institutions/payment providers over the services that can be offered to end users. Recent history suggests that this is where the commercial drivers for innovation are strongest. It would nonetheless be important under such an approach to put in place systems to ensure that the central architecture continued to evolve and innovate as required.
The Payments System Board recognises that the cost of establishing a payments hub and settlement hub may be significant and that expenditure by individual financial institutions would be required in addition to this to implement real-time payments. This needs to be considered in establishing a process and timetable for achieving real-time payments. The Board is also aware that consideration will need to be given to processes for detecting and preventing fraud, money laundering and terrorist financing given that these processes have the potential to compromise the speed of payment processing.
In summary, the Payments System Board believes that the payments industry should be aiming to provide a system of real-time retail payments within the next several years. A logical extension at some point would be the capacity to associate identifiers with account numbers for easier addressing of payments. Further, the Board's view is that the system would best be provided by the establishment of real-time payment and settlement hubs. Not only would this place Australia at the leading edge of retail payments, but it is likely in itself to promote significant further proprietary innovation, using the capabilities provided by the system.
A settlement hub would need to be provided by the Reserve Bank, but the Board has an open mind about how a payments hub would be provided. Accordingly, it is seeking views on how real-time retail payments should best be delivered. However, the Board does not wish to see Australia lag significantly behind in the provision of real-time retail payments and therefore will provide the industry with a relatively short window to consider the issue. It is seeking industry views by the end of August and would like to be in a position to have a clear path forward later in 2012. An attachment to this paper sets out a possible model for a real-time retail payment system provided via a hub, along with some indicative timelines.
The Board does not see the establishment of a system for real-time payments as eliminating the need for DE payments to be settled in a more timely fashion. The DE system is of pivotal importance for business payments and it is important that its users are able to harness the economic benefits associated with more timely payments. Ongoing investment is also required to ensure its continuing reliability and utility for users.