Introduction and overview
That the development of electronic commerce on Internet cannot take place without appropriate payment mechanisms appears practically self-evident. Yet, progress in this area has been far from harmonious. Disagreements on basic definitions and implications of electronic payments persist. While number of standard initiatives this development has been launched, very few of those have been implemented and those, who have, such as SET, have been received less than enthusiastically by market participants. Pioneering payment systems and applications, some of which were truly innovative, failed to gain widespread market acceptance.
As electronic commerce continues to grow at a breathtaking place, the need for secure electronic payment systems becomes more and more acute. In order facilitate their emergence it is essential to understand to reasons for the slow progress to date. This papers thus starts by looking at problems of definition. It then looks at recent standardisation initiative and payment system experience. In the conclusion, we seek to draw broad lessons from recent experience and suggest key areas for further analysis and experimentation. The main lesson is that the discussions of secure payments cannot be limited to technical aspects (authentication methods, PKI, etc.) but need to take into consideration economic and institutional aspects of payment systems such as customer attitudes and payment system participant strategies.
Electronic money
Defining electronic payment
What is the meaning of electronic payment ? For some people, these are payments linked to e-commerce transactions. This definition raises two problems: many, if not the bulk, of e-commerce transactions are neither paid nor settled over the Internet.
Since the development of Internet-based electronic commerce, new payment systems has appeared (E-cash, CyberCash, Kleline…), with the objective to allow merchants and customers to settle their transactions directly on the web. However, today, the most frequently used payment approach, via credit card payment, uses Internet only in the initial stage, when a customer communicates his card to the merchant’s server. Once the number is communicated, the merchants sends the information to the proprietary network of the card company and the Internet transaction is aggregated with other customer transactions. A variant of this approach uses secure server, to receive card numbers without merchant’s knowledge. The server authenticates and validates the number and confirms its validity to the merchant. Upon merchant’s acknowledgement, the number and the related transaction are sent to the credit card company for processing and settlement.
Second problems of the Internet-centred definition is that the electronic payments encompass a much larger universe. The best known forms of electronic payments, such as interbank transfer via SWIFT or credit card clearing, have been in operation long before the advent of Internet-based electronic commerce, which only began to take off in the late 1990s. SWIFT network was launched in the mid-1970s and clearing systems for both interbank and customer transactions were introduced in the early 1980s.
Thus, there is no one-to-one correspondence between electronic commerce over the Internet and electronic payments: not all e-commerce transactions are paid and settled via Internet-based payment systems and not all electronic payments are Internet related.
If one limits the scope of definition to electronic payments over the Internet, a standard approach is to identify four categories of electronic payment products:
* Electronic cash
* Electronic cheques
* Credit cards
* Other Methods.
It is the last, residual, category that is the most interesting but it clearly needs further elaboration.
A more generic approach is to distinguish between value-based and information-based systems. In the value-based systems, there is actual electronic transfer of value of between the parties to the transaction. E-cash is an example of such a system. In information systems, only data circulates on the Internet and actual transactions take place either off-line or on proprietary networks. These data provide transaction amount and account details. They are extremely valuable. As Walter Wriston stated in 1985, "information about money is more valuable than money itself ". The bulk of payment systems are information-based in a double sense of the term:
* they use transaction data
* .. to generate movement between accounts.
Security requirements vary between the two types of systems. Value-based ones require considerably higher degree of security (which may explain the difficulties these systems have experienced) than the information-based systems. This however does mean imply that the latter are not subject to stringent requirements in terms of authentication, non-repudiation and privacy.
Defining electronic money
The rapid development of electronic payment systems raises a question: is a new category of money, electronic money (or e-money) emerging ? Academic, business and regulatory experts appear deeply divided over the question. A generally accepted definition of money is that of an universal instrument accepted for the settlement of all commercial and financial obligations within a given economic system. The traditional approach to is to distinguish between two categories of money : fiduciary money (cash) and scriptural money. The first one is issued by Central Banks, while the second one is created by private banks.
Electronic payment instruments cannot be easily or exclusively assigned to one or the other category.
Some electronic instruments appear to function like cash, in terms of ease of transaction and anonymity, except that such electronic cash is not necessarily issued by central banks and or by financial institutions. Indeed, one of the main reasons to speak about a new kind of money is that in the emerging e-commerce environment, the electronic value can be created by a non-financial institutions and can be transacted outside the traditional banking systems.
Regulators and public authorities view such a possibility with some concern and therefore prefer to restrict electronic money handling to financial institutions. Under this approach, electronic money is assimilated to scriptural money and as such does not appear to represent a qualitative leap, requiring a fundamental overhaul of regulatory and institutional framework of monetary systems. According to many central banks within European Union such as Banque de France [reference], e-money resembles traveller checks, except that the latter are not divisible. No new status or regulation were required for traveller checks, and therefore no new status is necessary for e-money. Under this view, E-money is best defined a form of scriptural money stored on an electronic support and does not require separate treatment.
This is not an universally accepted view. For many business and public policy analysts, the development of electronic payment systems and instruments fundamentally modifies the traditional architecture and relationships within the financial value system and therefore warrants a specific framework.
European Commission position (directive proposal for regulation of electronic money issuers)
This is the view which has been adopted by the European Commission. In September 1998, the European Commission has issued a proposal for European Parliament and Council Directives on the taking up, the pursuit and the prudential supervision of the business of electronic money institutions. This proposal is a result of several years of discussion between official bodies and between the public and private sector. It proposes a legal framework for regulation of e-money issuance by potential non-bank actors.
Until now, the supervisory and regulatory approaches to the issuance of e-money have developed on an ad hoc, country-per-country basis throughout the Union. There is no uniform EC-wide legal framework for electronic money issuance. This has two adverse consequences:
* If regulatory issues are not addressed, innovative payment schemes cannot be implemented, in turn adversely impacting the take-off of electronic commerce within Europe. Even for systems that are launched, regulatory confusion inhibits consumer acceptance.
* The absence of regulatory framework increases potential risks (massive fraud, systemic loss of control or even failure) that new payment systems are likely to generate. It makes it more difficult to address them preventively and to properly cope with their fallout.
It has therefore essential to define common rules and to ensure the soundness of electronic money issuers as well as integrity of electronic payment systems.
The European Commission also insists on the importance of e-money in connection with the introduction of Euro. Electronic money can significantly contribute to the customer acceptance of the single currency during and after the transition period and to the growth and development of cross-border retail payments.
In the directive proposal, European Commission adopts a definition of e-money somewhat different from the traditional one. It defines e-money as a multi-purpose instrument. In other words, commission define e-money as a payment instrument helping to settle more than one kind of transaction, while the widely accepted definition of money stresses an universal dimension to money.
This definition leads to a broader and more ambiguous definition of the issuer of electronic money. A non-financial institution, a retailer, an ISP or an Internet portal, that issues an electronic instrument, which can be used in several types of transactions (buying physical goods with selected merchants, buying intangible goods such as information, participating in an auction, etc) can thus be considered as an electronic money issuer.
As such instances are becoming more likely and more prevalent, the earlier view that
only credit institutions as defined in Article 1 of the First Banking Co-ordination Directive should be allowed to issue e-money is no longer sustainable and the proposed E-money directive explicitly acknowledges the possibility of non-banking e-money issuers.
The directive postulates that such issuers should be subject to prudential supervision to ensure their soundness and integrity and is largely modelled on the existing First and Second Banking Co-ordination directives. Thus, e-money issuers should be subject to same conditions as credit institutions: prior authorisation - minimum capital requirement - fit and proper management - sound and prudent operations - initial and ongoing owner control.
However, the supervisory regime currently applied to banks is deemed too heavy, onerous and convoluted and thus potentially harmful to the development of new payment instruments. European Commission considers that a limited prudential control should be sufficient to ensure the soundness of e-money issuing activities. It therefore recommends lower lies in the initial capital and on-going own funds requirements for institutions which are only involved in the issuance of electronic money. The initial capital requirement for banks is 5 million EUR while that proposed for electronic money institutions is set at 500,000 EUR. On an on-going basis banks are required to maintain a minimum own-funds requirement of 8% of total assets while the figure proposed for electronic money institutions is set at 2%.
An option is left to Member States allowing for a waiver of the provisions of the proposed requirements for small (under 10 million EUR) e-money schemes.
The logical implication of the proposed directive is to create a new category of financial intermediaries: non-bank electronic money issuers.
The proposed directive is still under discussion. As mentioned before, it is controversial and afflicted by the middle-of-the-road syndrome. For e-commerce enthusiasts, it may create an additional burden and deter innovation. For regulators such as Central Banks, it may be too light. Thus the European Central Bank (ECB) would prefer the issuance of electronic money to be limited to credit institutions and to address the question of new entrants through an amendment to the First Banking Co-ordination Directive which would enlarge the definition of credit institution to include all issuers of electronic money.
ECB is also concerned about the absence of treatment of redeemability in the proposed directive. For ECB, it is essential to clearly affirm the general redeemability of e-money instruments not only from the merchant but also from the customer. Systems which do not accept to redeem its e-money instruments from the customer should not be authorised. Furthermore, the e-money should be redeemable at par. In absence of such stipulations, which aligns e-money with fiduciary money and put limits on its creation, e-money may trigger serious risks of customer confusion and aggravates the problems of control of monetary aggregates.
Critical role of clearing and settlement
The redeemability question highlights another significant gap in the scope of the proposed Directive. It focuses on the issuance of e-money and does not address the settlement and clearing: the processes and systems to settle transactions and resulting payment obligations between those who issue the money and those who accept it.
Yet, clearing and settlement are as essential in the determination of the scope of acceptability and universality of money (whether fiduciary, scriptural or electronic) as the issuance. The analysis and treatment of electronic money are incomplete without the consideration of its settlement and clearing aspects. Furthermore, it is in this area that the widespread use of Information Technology has had the strongest impact. It is the back-office automation that facilitated and stimulated the explosive growth in volume and scope of electronic payments, wholesale and retail, national and global.
Clearing and settlement activity is closely supervised by Central banks, which are often directly involved in operating such systems. This is particularly the case for large-value interbank systems such as TARGET, which is used to settle large cross-border in euros and which is operated by the European Central Bank. Indeed it can be argued that for central banks, maintaining the integrity of clearing and settlement system has become a mission-critical function, on the par with its "lender of last resort" responsibility.
In a competitive market situation, there are several issuers of electronic money, using various technologies for issuing their "currencies". These systems will succeed or fail as a function of their acceptance by merchants and customers. It is the underlying clearing and settlement system that determine the acceptance level of a given payment instrument. A related issue is one of the fungibility, the extent to which e-money instruments are fungible with more traditional forms of money. Can a customer convert his e-money balance back to scriptural money. Does a merchant, who receives an e-money payment need to maintain a separate account, where e-money will accumulate or can he/she mingle them with other funds ?
From a perspective of overall e-money development, it is advantageous that various systems be interoperable in order to facilitate cross-system transactions and ensure redeemability. However, while highly desirable in theory, such interoperability is quite difficult to achieve in practice. For instance, global banking card networks, Visa and Mastercard, have only limited, terminal level, interoperability. Within the Euroland area, despite common currency, acceptance of Euro-denominated payment instruments is far from universal. Obstacles are not only technical, dealing with network interconnection and common payment standards, but also and primarily economic, having to do with interchange fees for accepting and processing payments from another payment systems and networks.
Initial dynamics of e-money do not appear very encouraging in this respect. While proponents of different systems proclaim their belief in the virtue of open systems and common standards, most of them view such standards as based on their solutions.
Furthermore, many of those systems pay limited attention to banking and settlement issues. The primary emphasis of promoters of e-payment has been on user interface and overall architecture. They paid scant attention to settlement and clearing aspects. As the reason, most of current electronic payment systems, for instance Cybercash or Kleline, rely on existing banking clearing systems. SET makes the reliance on existing clearing and settlement relations its major selling point.
Yet, such reliance is not without its drawbacks. The overall payment clearing structure is complex and hierarchical, with various systems for various payment categories (check/card, retail/wholesale). Those systems are interconnected by common membership and interoperability agreements but the degree of interconnection varies considerably among systems and instruments. More importantly, the existing systems were not designed for e-commerce transactions. They are cumbersome, fragmented and often costly to use. There is a broad consensus that they are not well adapted to handle large volume of e-commerce transactions.
There is clearly a significant scope for alternative transactions settlement and clearing systems, which could be created outside the traditional banking structure.
A model for such as system can be found in mobile telecommunications sector with companies such as MACH, a Luxembourg-based organisation. MACH core business is the clearing of call data generated by the international roaming activities of GSM phone users. Each roaming call creates a liability of a network of a GSM terminal holder toward the network used by this holder. MACH facilitates settling of these liabilities on bilateral and multilateral basis. Clearing services include comprehensive data checks and all related management reports. MACH processes, converts and reconciles between Transferred Account Procedure (TAP) and Cellular Intercarrier Billing Exchange Roamer (CIBER) Record formats to facilitate transactions between roaming partners using different technical standards.
If more such clearing systems were to be created by entities other than financial institutions, it would be interesting for European Commission and the monetary authorities to evaluate the implication of their development of such systems. Do they create significant new risks ? Who would guarantee their performance and integrity ? Banking clearing systems have a link to Central Banks. In non-bank clearing and settlement systems, Central Bank would normally not get involved. Is there a need for new regulation in this area ? If yes, what would be its principles.
The need for common rules concerning clearing and settlement appears paramount. Such rules should cover redeemability and fungibility of money, as well as operational and financial requirements to ensure proper operations of the system. The questions of who should be authorised to create and manage a settlement system and who should have direct access to the system also need to be addressed.
Payment systems and products: summary overview
Useful distinctions
The term "electronic payment systems" covers a variety of payment mechanisms. Furthermore, this is domain is in a constant state of flux, with new schemes emerging and existing schemes being modified, often substantially.
Two other important distinctions are:
* Account-based e-money, when a transaction involves a double entry into accounts of user, merchant and corresponding financial institutions vs. token-based e-money, where value is embedded in a software, linked to an electronic device (card or hard disk, for instance) which can circulate without any direct reference to a bank account. In this case, a transaction is carried out by transfer of the digital value from one device to another.
Token e-money is often referred as "digital cash". However, e-cash does not present all the features of fiduciary money. For one thing, its issuance is not limited to Central banks. Another distinction is the need for specific infrastructure of readers and application software. This limits the potential acceptance of e-cash to IT users. Another key difference is that e-cash allow the traceability of the owner. The degree of anonymity in e-cash is a hotly debated issue.
* Stored-value products vs. secure access products. Electronic purse or e-cash envelope are stored value products. On the other hand, the French smart debit card does not store value only a secure key allowing the user to access the payment systems which then effects the transaction. Many Internet banking products are in fact the security applications allowing to transmit the necessary instructions to make funds transfers between various bank accounts overt the Internet. Access products may use similar mathematical methods and electronic support as stored value products. However, they do not raise the same level of regulatory and security concerns as e-money schemes. One aspect of stored value systems, which generates considerable controversy is their disintermediation potential. Schemes such as Digicash or Mondex allow transfers of electronic value directly between two parties to a transaction without any involvement of a third party such as the issuer of the electronic value or a clearing bank. While this may sound attractive to advocates of consumer sovereignty, it creates serious concerns about fragmentation of payment systems and possible weakening of integrity of the whole monetary system.
Not surprisingly, secure access products generate less controversy and are more widespread.
* Software vs. Hardware. Some electronic payment systems are entirely software-based and are using hardware, PC or network server, as a generic support. Other systems include at their core specialised and dedicated hardware, such as microprocessor chip embedded in a plastic card. Hardware-based implementation provides high performance, at the cost of lower flexibility. In contrast software implementation, while typically offering lower levels of performance, provides a high level of design flexibility and hence the ability to evolve a design in response to real world needs. While partisans of software-only solutions proclaim that these are highly secure, there is a noticeable preference, particularly among regulators to rely on hardware-based solutions, which are thought to reduce the risk of unauthorised entry and duplication. Thus, smart cards are rapidly becoming an integral and critical part of digital signature and certificates systems.
Internet e-payments: Road kills and hard slogs
While electronic commerce has been growing more rapidly than the practically all the market forecasts, the development of Internet-based payment systems has been somewhat disappointing. Despite considerable media coverage and excitement among the digerati, practically all the systems run into difficulties, which sometimes appear fatal. Thus, First Virtual Holdings, went into bankruptcy in July 1998. company Digicash, tireless and highly visible promoter of E-Cash, which moved from Amsterdam to the promised land of Silicon Valley in April 1997, acquiring substantial funding and prestigious investors, including Nicholas Negroponte, has been liquidated in September 1998 after failing to obtain additional round of financing Even the acknowledged Internet payments market leader, Cybercash, is struggling and its offerings have not really caught on. This is also the case for French Kleline, which at present de-emphasises its technology and stresses its services to merchants as an e-acquirer.
Micro-payments, which were considered in 1996 as a strong candidate for a killer application, have so far failed to take off. One of the most promising approaches Millicent, based on light encryption and innovative institutional arrangements, still remains in pilot stages.
Broad standardisation initiatives are making little progress. JEPI (Joint Electronic Payment Initiatiative), despite a strong support from CommerceNet and World Wide Web Consortium was abandoned due to the lack of interest of business participants in the initiative.
Another broad-based standardisation initiative, SET, supported by the credit card giants, Visa and Mastercard, as well as such IT heavyweights as IBM and Microsoft, faces significant and growing problems of market acceptance by merchants, end customers and banks themselves.
What are the reasons for what appears as a general lack of success ?
Let us first note that practically all the Internet projects are oriented toward retail customers. These are primarily concerned by the ease of use and cost-effectiveness of payment systems. They are not particularly seduced by technological sophistication and conceptual breakthroughs. Furthermore, while they may zip and zap and surf on the Internet, in the payment domain, they are not inclined to rapid shifts in preferences and brusque departures from existing solutions that work.
The main problem of the first generation of Internet payment systems is that they have not focused enough on their customer behaviour and attitudes. As a result, most of these systems appear as solutions in search of a problem, combining considerable technological sophistication with a degree of marketing and business naivety. To demonstrate this point, let us consider two examples: micropayments and SET.
Micropayments
Micropayments have been seen by many as a preferred mechanism for transactions for intangible goods - information or on-line entertainment -, for which fair value and willingness to pay are notoriously difficult to determine. In the digital world, intangible goods take form of digital streams, which, it is argued, can be metered and priced by small increments in way similar to electricity or water.
Yet, development of micropayments faces at least three significant roadblocks.
First such a roadblock is the functionality-cost trade-off. Anonymity and security impose heavy computational overhead in processing speed and disk access. For instance, The original DigiCash model only allows processing of 10 transactions per second. This is a general problem for all secure electronic systems (public key encryption is computation intensive be definition) but it is particularly critical for high volume-low value transactions.
A related roadblock is created by the economics of micropayments. One of the main reasons while bank cards are not used for micropayments is the difficulty of allocating processing and payment commissions for payment amounts under 20 dollars. Those commissions, of the order of 2 or 3% of the face amount, are split between 3 or 4 participants (issuing bank, paying bank, card processor...). How to manage these commissions and the split allocation process for transactions of 1 cent or less ? Is it at all practical ? If it is not, how to tackle the critical issue of business interoperability, without which micropayment schemes will remain highly fragmented ?
A related issue is that of the cost of support and exception handling. Setting up the infrastructure for those functions requires heavy technological and human investment, which entails a high unit cost, For traditional payments, exception handling costs can be as high as 50 dollars per transaction. Furthermore, it is not certain that the high transaction volume reduces these costs significantly.
Promoters of micropayments have not paid enough attention to two basic and, I believe incontrovertible, propositions:
- Micropayments are not a downsized version of normal and large-scale payments. They require a different functional design and specific business arrangements.
- Transactions for intangible goods are different in nature from that for tangible goods. Intangible market specificity calls for new approaches and business models.
Other, and potentially more promising, approaches emerge to tackle both small amount payments and intangible goods transactions. Such approaches are based on two well-known pricing principles: bundling and third party payments. Interestingly, many analysts believed that the greater use of Information Technology will eliminate these principles, considered as relics of the world of insufficient information and high transaction costs. Yet, they are becoming more rather than less important and this trend is likely to continue. In the new universe of abundant and cheap information, the reduction of unit transaction costs entails an explosion of the number of transactions and thus raises the systemwide transaction costs. In turn, this increases the complexity, the variability as well as the system operation and maintenance costs. Furthermore, the availability of time becomes a scarce resource, which means that the opportunity cost, both to consumers and producers, of monitoring every single transaction increases significantly. Bundling reduces the need for such monitoring and thus reduces the pricing system variety and transaction costs. Banks are using bundling extensively to price their services: for instance, customer pays a single price for securities research, transaction execution and portfolio monitoring. Furthermore, in case of mutual funds, he can, for a fixed price, switch a certain number among funds. In information services, whether on-line or off-line, bundling through subscription is widespread. Let us also note that the bulk of Internet services is offered for a fixed price and the trend is toward more rather than less widespread use of fixed pricing. According to two US academics, Yannis Bakos and Erik Brynjolfsson, bundling pricing strategies (as compared to selling the goods separately) are more profitable to Internet content providers and more interesting to information users.
The extensive use of third party payments in intangible goods transactions reflects the multi-output and multi-user (and multi-audience) nature of those goods. It is not only that the same content can be consumed at the simultaneously by many consumers, but also that this consumption generates by-products, such as an audience of a given size or a new business recognition, which in turn be bought and sold, and are often more valuable than the original good. Information technology, and the Internet particularly, make it easy to generate and track such valuable by-products and to identify various audiences for which these by-products have value. Third-party payment systems, such as various forms of advertising and promotion, provide suitable mechanisms to capture this value.
Given these considerations, micro-payments are unlikely to become the dominant payment mode for Internet consumer commerce. The most optimistic estimates do not go beyond a 10% market share in the year 2000. Nevertheless, micro-payments will become an integral part of a full set of payment mechanisms necessary to enable Internet commerce. They appear however useful for some specific segments such as games or pay-per views. Furthermore, the underlying technology can be used for other purposes such as intellectual property tracking and transactions monitoring. Bakos and Brynjolfsson recognise that reductions in transaction and distribution costs made possible by ubiquitous networking are likely to lead to an increased use of disaggregation-based pricing strategies taking advantage of micro-payment technologies.
SET
SET also demonstrates the lethal effect of technological overkill. It is based on two apparently incontrovertible propositions:
* All parties in electronic commerce transactions do not want to turn upside down their existing institutional relationships. Thus, SET goes to great lengths to transpose existing transactions and information flows into the virtual world of e-commerce.
* Customer and merchants want high level of security, particularly as regards authentication of parties to an e-commerce transactions.
Both of these proposition are however less obvious than they seem. Current credit card payment system are complex and convoluted and may not best adapted to the e-commerce environment. Current SET model comprises five major transactions (Cardholder Registration, Merchant Registration, Purchase Request, Payment Authorisation, Payment Capture), each entailing several data exchange and verification flows.
The institutional complexity is further exacerbated by the security consideration. The key feature of SET is the use of digital certificates, based on public key cryptography, to ensure the highest level of security. While it is true that certificates provide unmatched authentication, flow control and data protection, these attributes come at the expense of heavy computational overhead, which make SET-based system slow and difficult to use. Hence adverse reactions from consumers, merchants and banks.
SET is seen as an overkill, particularly because one of its underlying technologies, Secure Sockets Layer, is already extensively used for electronic commerce transactions and it is deemed highly satisfactory. It is the very success of SSL that reduces the need for SET.
There are also non-technical reasons for banks’ reluctance toward SET, which is card-centric and therefore not directly linked to the customers’ bank account. Such system potentially reduces customers’ loyalty towards their banks and . Furthermore, card-based system increase the visibility of card networks and enhance their, rather than the banks’, ability to capture value-added.
Concerning SET, many banks feel that they do not have an inadequate input into the decision-making process. While they recognise the usefulness of a card-based payment approach, they believe that internet banking should emphasise direct relations between banks and their customers it should not be considered as the best or the only Internet payment systems. They prefer to give priority to account-based payment mechanisms. Because such mechanisms link parties which know each other well, they have lesser need for elaborate certificate and Public Key management arrangements, which are geared primarily for occasional purchases among parties which do not know each other well.
Scalability and proportionality
Thus, the principal lesson from the recent experience of Internet-based payment systems is the need for close integration of technological and business aspects (with particular emphasis on customer behaviour) in the design of such systems. Technological overkill is a fatal trap. The same way that IT system designers worry about scalability, the ability of the system to match substantial increase in scale, payment systems should worry bout proportionality between technology and its use. Ecash technology for instance is impressive, with its patented concept of the blind signature, but, because of its total cost, it is much more appropriate for high value payments such as ones required for business-to-business e-commerce.
And this is another lesson to be drawn from the recent experience: the need to shift the emphasis of the efforts to the business-to-business segment. Growth potential for this segment is considerable and higher than for business-to-consumer one. Payment requirements are critical and offer considerable scope for application of innovative solutions such as digital certificates and more broadly, public key infrastructure.
Not surprisingly, new initiatives in this segment are beginning to emerge. One of the most interesting ones is Identrus, a global organisation created in early 1999 by eight leading financial institutions in Europe and in the United States (ABN Amro, Bank of America, Barclays Bank, Chase Manhattan, Citygroup, Deutsche Bank and Hypo Vereinsbank). Identrus seeks to create a global trust infrastructure, enabling business-to-business commerce among all companies which are using Identrus infrastructure. Identrus will create a trust internetwork, linking in a structured and hierarchical way various security and certification systems created by its members. Identrus itself will operate a root certificate authority (root CA), an entity at the pinnacle of the electronic identity hierarchy. In July, Identrus launched a "proof of concept" pilot of its architecture.
Evolving payment system architecture
The Internet Shift
Since beginning of the 1970s the development of electronic payments was based on proprietary networks owned by banks, other financial institutions or co-operative card companies like Visa or MasterCard. Some of those networks were restricted to financial institutions, while others offered connections to merchants or corporate users. In all cases, they were based on proprietary protocols and dedicated telecommunication infrastructure.
The development of financial activities on Internet changes radically this situation. Internet model dissociates the network from the physical infrastructure. It allows interconnection between heterogeneous networks and provides common standards. The notion of dedicated infrastructure is no longer relevant. With encryption technology, digital certificates and smart cards, it is now possible to use security in a modular and flexible fashion. A highly secure environment can be created on the public networks. ON the other hand, the likelihood of data interception and penetration attacks is increasing more easy. A new philosophy concerning security has been adopted for payment systems on Internet and many efforts have been realised to improve security. To what extent these developments change the prevailing approaches to payment and transaction processing networks, based on dedicated infrastructure and proprietary protocols ? What are the implications for their architecture ? For their access and control policies ?
Does the combination of business and technological trends imply a banalisation of payment systems ? Does it reduce barriers to entry to a point where any network can become a market, any computer a clearing system and anybody can issue electronic money ?
In the new environment, what is the meaning of:
- Financial transaction
- Financial intermediary (bank, broker,broker/dealer)
- Money ?
What are the emerging models of
- Financial intermediation
- Money issuing
- Clearing and settlement ?
Network good and co-opetition
When trying to understand the dynamics of the emerging Internet payments systems, the concept of network good appear highly pertinent.
Economists call "network goods" products and services, whose value is linked to the number of users. Thus network goods derive their value from the connection with other goods. A telephone or a fax is useless unless it allows us to communicate with other telephones and faxes. Wider the connectivity, higher the value of each product and, more importantly, of the network (physical or virtual) that connects them. This rule clearly applies to payment systems: greater the number of transaction points (or terminals), higher the value of the system and greater the incentive to join.
Those who offer network goods face discontinuities and threshold effects. They need to create a critical mass of products and network connections before they can attract customers. Thus network goods have high start up and fixed costs. However, once the critical mass is reached, marginal costs of producing additional goods and/or attracting new customers are low and often declining. Network goods display increasing returns and economies of scale. The growth becomes explosive. According to Robert Metcalfe, inventor of Ethernet, in data networks, the traffic is an exponential function of the number of users.
The notion of the critical mass needs to be well understood. It concerns not only the number of users but also the actual consumption and use of network goods and of the network. It is the use that ultimately determines the profitability. For some networks, critical mass and large size are not sufficient, they have to attain ubiquity at a reasonably low cost. This is clearly the case for payment network: if just one merchant does not have a card terminal, the customer needs to keep cash, which substantially reduces his appreciation of the utility of the card network.
The demand for network goods is discontinuous and marked by a "chicken and egg" dilemma: a user is reluctant to join a network with very few users, which keeps the network size small (and access cost high) and therefore discourages from joining. Consumers have no interest to use a payment system that is not accepted by a certain volume of merchants, while merchants will be reluctant to invest in a new system if customers are not using it. Thus the demand curve is non-linear, low growth in the initial stage, followed either by a steep decline, if critical mass of users is not reached, or a rapid growth, if the critical mass is attained. The decision to join a network is influenced not only by the perceived benefits of the use but also by the opportunity cost of exclusion. For instance, a financial institution cannot function without access to SWIFT payment network or Reuters (or Bloomberg) market data information services. From the customer perspective, the critical mass of a network good can be defined as the point where the opportunity cost of exclusion becomes higher than the cost of joining.
In a system environment, relations between stakeholders constitute a mix of competition and co-operation. We call it, using a term coined by Ray Noorda, founder of Novell, " co-opetition." Thus financial institutions co-operate to design, implement and deploy financial networks, while simultaneously competing to capture the largest share of the value created by these networks.
Standardisation is a relevant example of how the co-opetition has evolved. Initially, it was seen as an epitome of co-operation. As the importance of standards in the market development became more evident, the nature of the process has changed and became more contentious. The decision-making has been accelerated in order to reduce the time-to-market, considered as a critical feature of IT applications. Participants in working groups and review committees have adopted more guarded attitudes and are less willing to share information. More significantly, the focus of the process has shifted from public, formal bodies such as ISO or CEN, to private ad hoc groupings. We entered the era of the rule of de facto standards.
De facto standard setting process can take two major forms. One is a dominant player form, where a firm which has the market power, imposes its standards on the majority of participants. This is the case of Microsoft and Windows operating system. The dominant player form is basically a zero-sum game, where only one standard survives in the end. The rewards are huge (and so are the corresponding losses). The holder of the standard can embed it into his products and charge a royalty on its use. While this model has worked very well for Microsoft, it has generated a strong backlash from market participants who do not want to see it reproduced and resist strongly any standards that appear as a monopoly of a single supplier. As a result, firms, which have developed interesting new technologies, tend to shy away from this model. They prefer to forego potential revenues, which would accrue from the standard's exclusive ownership, in favour of a wide adoption of the standard, from which they reap indirect but often substantial benefits. This is the strategy of Sun for Java.
An alternative de facto standard setting process is through common initiatives of ad hoc groups on a sectoral (vertical) or cross-industry (horizontal) basis. The most successful example of cross-industry process is the TCP/IP standard for Internet. The lack of a dominant player reduces resistance of market participants and makes the process appear as a positive sum game, whose objective is to expand the market.
The de facto standard setting has the advantages of speed, market relevance and adaptability. On the other hand, it leads to standard competition and fragmentation, which in turn creates confusion among market participants and may hamper the speed of adoption of new technologies.
Competing standards create a need for interoperability. Interoperability is a complex and dynamic game. On the one hand, it enlarges the size of the market and reduces development and operating costs for all players. On the other hand, it reduces the switching costs and the differentiation between competing suppliers. It therefore exacerbates competition and often lead to product commoditisation.
The cost of interoperability depends on the timing of its implementation. It is easier and cheaper to build it into the initial system design than to have to re-engineer into an existing system. This consideration has led for instance the telecommunication industry to adopt the principle of interoperability from the outset of the GSM system. Furthermore, while costs of implementing interoperability are clearly identifiable and borne by suppliers of networks goods (or of their components), its benefits are considerably more diffuse and accrue to all stakeholders.
Security in payment systems: evolution
The evolution of Internet payments can be divided in three major steps.
First transactions on Internet were settled using rudimentary payment approach. Customers were communicating their card number directly to merchants. This procedure was extremely unsecured both for customers (risk of card number interception, aborted transactions, …) and for merchants (card number theft, exception handling, erc).
In order to improve transactions security, systems as SET, introduced a few years ago the digital certificates technology. Those certificates guarantee the identity of both customers and merchants, through a systematic use of digital signatures. However this approach relies on a software applet, which only identifies the user's computer but not the cardholder. This may cause serious problems if a third party obtains the password to activate the certificate. For example, many Internet users connect themselves on their working place, benefiting this way from free Internet access. This means that many persons have access to the computer in which the certificate is recorded.
The third step towards a more secure payment mean is the introduction of secured hardware in the process with projects such as Cyber-Comm developed by Groupement des Cartes Bancaires and the French banking sector. This system, based on SET technology, uses a smart card reader connected to the computer. To make a payment, a customer enters his/her bank card in the smart card reader and dials his PIN code on the smart card reader. If the code is correct, the smart card reader activates the digital certificate located in the computer memory. In other words, without the bank card and the code it becomes impossible to make a payment. Such a system allows a level of security similar to that offered in the physical world by to reach the same security level as with EFT POS and ATM infrastructure. However, if the risk level is reduced for the merchant, customer-merchant relationship becomes more imbalanced.
This evolution towards smart card-based secure payment solutions is not limited to Internet: in Europe bank cards are migrating to EMV standards and many large-scale electronic purse experiments have been launched. As PC becomes more integrated into the network, the need for a secure mobile access and stored value device is likely to increase.
Issues
The analysis of current trends in Internet-based payment systems shows that the evolution has so far raised more questions then brought firm and definite responses.
From security to risk management
The notion of security of payment is clearly insufficient to provide appropriate conceptual framework for technical and institutional design of Internet payment systems. There is a need for a broader approach of risk management. Such approach recognises that electronic payment entails a series of interrelated risks: financial risks, technological risks, operational risks, legal risks. Some of those risks are generic to banking business, others are specific to electronic payments, such as interception of messages, break-in into security infrastructure.
Identification, confidentiality and payment integrity
Payments on the Internet need to three major broad conditions:
* Firstly, each party involved in the transaction must be sure that its counterparty is exactly what she tells she is. In other words, people involved must be identified.
* Secondly, data exchanged between buyers and sellers must remain confidential.
* Finally, buyers must be certain that the information they get about the payment (regardless of the underlying value) are reliable.
Those three conditions can be met by the use of encryption technology. The main issue there is the migration from private to public key cryptography. Advantages of the latter are well known: employing a public key system, it is possible for a user to receive encrypted messages from an entity he has not met and with whom he has no on-going relationship.
The public key system also offers the possibility to create unique and hard-to-imitate electronic signatures.
On the other hand, The public key system has disadvantages. The key certificates as standardised in the public key systems take up to 1.5 to 2 Kbytes each. And in order to interact with multiple users, it is necessary to possess multiple keys. The system is heavy and slow. One solution to the performance handicap would be to put key management off-line n a smart card. This solution is currently experimented by Cyber-Comm and Identrus projects
Customer protection
The introduction of secured hardware and of a trusted third party reduces risks but, if not corrected, may create a serious balance between merchants and customers. Using Cyber-Comm system, the merchant has the quasi-certitude that he will be paid. Yet, the proposed solution does not reduce the uncertainty of a customer concerning good and timely execution of his order. The customer has to make the payment when signs his final order, before receiving anything. Customer is also concerned is about after-sales service and how to contact the merchant if a problem occurs. Often no phone number is indicated on web pages and there is no guaranty of receiving an answer if sending an e-mail. Other related issues are: in which language will the customer get the answer, are merchant's legal system compatible with customer's one regarding after-sales service? If customers protection is not improved, electronic commerce on Internet may generate a strong customer backlash. meet serious growth problems in the following years.
One solution would be to enhance the synchronicity of payment and delivery. This could done if banks were to assume a more active role as trusted intermediaries. Banks would keep money in escrow until final delivery of the product. For this solution to work, it is essential that a bond of trust exists between the bank and the customer, on the one hand, and the bank and the merchant, on the other. This may require substantial adjustments in the posture of European banks, which, contrary to American banks, have for the most shied away from involvement in disputes between merchants and customers.
Cross border electronic payments
Electronic banking and electronic money activities are based on technologies designed to extend the geographic reach of banks and customers. Such market expansion, particularly in Europe, will easily extend beyond national borders. Banks need to recognise that cross-border electronic payments are different not only from domestic systems but also from traditional global interbank systems such as VISA or SWIFT.
Banks face different legal and regulatory requirements when they deal with customers across national borders. They are compounded by uncertainties about legal framework for electronic banking, electronic payment and electronic money. Thus, banks active in cross-border electronic payments need to consider legal risk associated with non-compliance with different national laws and regulations, including consumer protection laws, record-keeping and reporting requirements, privacy rules, and money laundering laws.