This article is adapted from an address at the recent Payments '96 Conference in San Francisco.
Central banks are increasingly concerned about payments system risk and have actively encouraged private-sector operators to reduce their exposures.
A key question is whether the risk management standards that have been promulgated for large-value networks should also be applied to retail- oriented systems, such as credit cards.
The central banks of the most important countries in the financial world, the Group of 10, mutually agreed upon minimum standards in the 1990 "Lamfalussy Report." In 1994, the Federal Reserve Board issued a policy statement adopting these standards and instituted a further set of risk management principles that are applicable to large-value payments systems like Fed Wire and Chips.
The Fed defines a "large-value" system as one that routinely settles more than $500 million a day. This is regularly exceeded by the large card- based retail payments systems. Furthermore, the Federal Reserve made clear that the established standards should apply to all networks, though batch- type systems like the cards and automated clearing house were exempted for the time being.
The central question is whether the differences between large-value and small-value systems are so significant that they constitute a difference in kind rather than in degree, therefore exempting retail systems from the strict Lamfalussy standards.
There tends to be an inverse relationship between the number of transactions in a payments system and their total value. Visa and MasterCard executed about 15 billion transactions in 1994, while Chips and Fed Wire exceeded 100 million.
The total value moved over Chips and Fed Wire exceeded $500 trillion in 1994, compared with Visa's and MasterCard's half a trillion dollars. The large-value payments systems move more funds in a day than the retail systems do in a year.
These disparities are also reflected in the average transaction size: about $6 million in Chips, $3 million on Fed Wire, but only $70 in the bank card systems. Would one follow the same safety rules on a conquest of Mount Everest as on a molehill? Hardly.
To make the same point in still a different way: In no country for which the Bank for International Settlements reports the relevant data do card payments constitute more than one-third of 1% of the total value of cashless payments.
It can therefore hardly be argued that retail payments systems present the same kinds of systemic problems as the large-value systems. Nevertheless, retail systems may represent some moderate risks that should be monitored and contained.
Can the retail systems meet the Lamfalussy standards? If yes, this would certainly be beneficial; if no, we must ask ourselves what, if anything, should be done about it.
One of the standards retail payments systems can meet, Lamfalussy IV, calls for objective and publicly disclosed criteria for membership and fair and open access to the system. The bylaws of the various retail payments organizations spell out these conditions in detail and most should be in compliance with this standard.
Another standard, Lamfalussy VI, calls for the operational reliability of technical systems and availability of backup facilities.
Many retail systems have set and met standards of 99.99% operational reliability and should have no difficulty in meeting the Lamfalussy criterion. But 0.01% downtime amounts to 52 minutes during the year. If that occurs during the peak Christmas season, consumers might be very unhappy. However, this is not a regulatory issue but a service quality issue that can be left to the free market.
Lamfalussy II calls for participants in a netting system to fully understand the various risks to which they might be exposed. This is certainly a reasonable requirement that any financial institution can and should meet. The various retail payments systems have done much to educate their members through seminars and the publication of appropriate manuals.
Lamfalussy IV specifies that multilateral settlement systems should be capable of ensuring the completion of the daily settlement cycle in the event that the participant with the largest net-debit position is unable to settle. This requirement calls for the provision of adequate liquidity in the form of cash, credit lines, and collateral.
While this standard is reasonable for a payments system of any size, one can conceive of alternative ways to meet this requirement.
Participants in a netting system may formally agree to a partial settlement or an instant assessment in the event of a failure that exceeds the available liquid resources. The question is whether the liquid resources should be available beforehand in a centralized location or afterward from the decentralized resources of the participating financial institutions.
Which solution is better is a matter of cost and preference - as long as there is the assurance that the settlement will take place and does not represent any systemic risks to the financial system as a whole.
Perhaps the most troublesome requirements are posed by Lamfalussy III and the risk management rules in the Federal Reserve Board's 1994 policy directive.
These call for the establishment of bilateral net credit limits by each participant versus every other participant in a netting system; the monitoring and establishment of systemwide net debit limits for each participating financial institution; and the establishment of real-time reject or hold controls if the bilateral or systemwide limits are exceeded.
Such controls are simply not operationally feasible in retail batch- based systems.
Visa and MasterCard have about 20,000 member financial institutions worldwide. The bilateral rule would require that some 400 million sets of limits be established and monitored. To add insult to injury, there would be only one transaction every three days between any possible pair of participants.
To accommodate the occasional large transaction of a wealthy cardholder, the limits would have to be set so high as to be meaningless. A bilateral limit of $20,000 for all of the 400 million possible pairs of institutions would result in aggregate limits of $8 trillion - hardly reasonable for a system that moves only $500 billion a year.
Furthermore, adherence to the limits in a batch system can be monitored only after the fact - and there is no way to revoke a transaction once the customer has left a store with goods in hand.
It may be reasonable to comply with the spirit of Lamfalussy III and relevant Federal Reserve rules by putting a cap on card-based transactions. An individual transaction limit of $20,000 and a daily limit of $100,000 per card might reasonably meet Lamfalussy requirements while limiting the fraud potential for the banks participating in the system. Much more modest limits are being used successfully by automated teller machine networks.
A system of transaction limits would also discourage use of retail systems for large corporate payments - a purpose for which the retail systems were simply not designed.
As far as counterparty limits are concerned, presumptive limits equal to a bank's corporate lending limit should certainly be acceptable for all parties concerned. A set of systemwide net-debit caps should present no problem for virtually all banks. If these limits should be exceeded, additional collateral could be posted.
Finally, Lamfalussy Standard I states that netting systems should have a well-founded legal basis under all relevant jurisdictions. While this principle sounds very reasonable, it is highly questionable whether the bylaws and operating regulations could be enforced uniformly in all 200 countries where the retail payments systems now do business.
Given the variety of legal systems around the world, even a modest compliance effort would require enormous legal staffs whose output might be of questionable value.
Certainly, the associations' rules regarding jurisdiction could be strengthened and the systems should not do business where they don't have some degree of comfort about the enforceability of contracts. Yet, when dealing with bankruptcy courts in far corners of the world, even the best- drawn contracts may be difficult to enforce.
If central banks insist on iron-clad legal guarantees, one may attempt to introduce appropriate covenants into international treaty law. The World Trade Organization may be a useful venue for such an effort.
These suggestions for cross-border retail payments systems would go a long way toward meeting the intent of the Lamfalussy standards - making participation in payments systems so risk-free that the safety and soundness of the participating financial institutions cannot and will not be endangered.
Mr. Heller is head of International Payments Institute, based in Tiburon, Calif. He is a former Federal Reserve Board governor and former president of Visa U.S.A.