Part I

The Structure and Economics of

Payment Systems
This page intentionally left blank
1
The Architecture of
Payment Systems

Payment systems are indispensable to our lives as individuals and to

the smooth functioning of the economy. They allow money to fulfil its
role of accepted means of exchange when purchasing goods or services.
As private persons, it is through payment systems that we receive our
salaries and pay our bills. Enterprises use the same payment systems to
settle invoices within the terms of their contractual relationships. Finally,
financial trading activities also result in one cash leg through a payment
system to purchase shares, or two for a foreign exchange deal, one in
each currency.
It is difficult to trace the birth of the first payment system, but Box 1.1
at the end of this chapter describes one of the earliest.
A payment system includes private or corporate customers, financial
intermediaries, generally commercial banks, and central banks, linked by
telecommunication networks transmitting information between com-
puter systems. It is important to understand the roles and responsibilities
of each to reach the optimum balance between speed, efficiency, cost,
security and economic safety. On the other hand, each participant
is driven by its own objectives which are often contradictory. Can
speed be increased without impacting costs? Can costs be reduced with-
out creating opportunities for fraud? Each payment system reflects
therefore a compromise depending on the participants, the speed of
execution and level of security required and, last but not least, the
risk posed by the amounts involved. These will determine the most
appropriate operating mode, legal framework, security level and tech-
nology. Payment systems operate in a competitive environment and
technological innovation is one of the most important drivers in the
evolution of payment, chip cards being one of the most obvious
examples.

3
4 The Structure and Economics of Payment Systems

This opening chapter will attempt to define and describe some funda-
mental concepts which will enable readers to understand the rationale
behind the various payment instruments and systems described.

1 Introduction and basic concepts

‘A payment system consists of a set of instruments, banking proce-
dures and, typically, interbank funds transfer systems that ensure the
circulation of money’1 and normally requires:

• a payment instrument, for example cash, a cheque, an electronic

funds transfer, a credit or debit card;
• scheme rules defining the procedures, practices and standards agreed
between the payment service providers;
• a transfer mechanism; and
• a legal framework to guarantee irrevocable and unconditional finality,
that is the discharge of the obligation between debtor and creditor.

It is particularly important to distinguish between the information rel-

evant to the payment and the final transfer of value. If you settle a
purchase in cash for instance, your debt is immediately extinguished. If,
however, you remit a cheque, the vendor will need to clear the cheque to
ensure that the drawer has sufficient funds (or credit line) on his account
for the cheque to be honoured (that it will not ‘bounce’) which can
sometimes take a few days.
The participants in a non-cash payment system include at least service
providers, generally banks, which effect the payment on behalf of the
debtor and remit the funds to the creditor, and a settlement agent that
discharges the obligation as shown in Figure 1.1
The role of settlement agent is normally assumed by the central bank
for the relevant currency, which transfers funds between accounts held
with it by the commercial banks, known as ‘settlement in central bank
money’ which guarantees unconditional and irrevocable settlement. As
shown in Figure 1.1, we see that the debtor would still be liable if his
bank fails even after debiting his account. He is however no longer liable
if the creditor’s bank fails after its account has been credited by the cen-
tral bank. The time interval between the moment at which the debtor’s
account is debited and the funds are made available to the creditor is
known as the float.
As the volumes of payments grew (and we are talking nowadays about
tens of millions of payments daily in an advanced economy) it was
 The Architecture of Payment Systems 5

Central Bank

Bank D Bank C
⫺1000 ⫹1000

Bank D Bank C

Debtor’s account ⫺1000 Creditor’s account ⫹1000

Utility bill for 1000

Debtor/Payer Creditor/Payee

Figure 1.1 Settlement and discharge of obligation

no longer possible to settle each payment gross in central bank money

as described above. It became therefore necessary to introduce Auto-
mated Clearing Houses (ACH) which perform the clearing, defined as
‘the process of transmitting, reconciling and, in some cases, confirming
payment orders or security transfer instructions prior to settlement, pos-
sibly including the netting of instructions and the establishment of final
positions for settlement’2 (see Figure 1.2).
Banks send batches of payments to the ACH which, after sorting and
merging, sends the banks details of payments for their customers. It
also calculates and transmits the net positions between the banks to the
settlement agent (normally the central bank) which transfers the net
amounts between the settlement accounts of the participating banks,
therefore assuring final settlement for all payments in that cycle. It is
therefore important to distinguish between clearing, which is a set of
processes, and settlement which is an event which guarantees the dis-
charge of the debts which takes place when the central bank transfers
the funds between the accounts of the debtors’ and the creditors’ banks.
Clearing Houses would settle at the end of the day. As the value of
payments grew, reaching trillions of dollars daily (1 trillion = 1 million
millions = 1012 ), central banks became concerned that a bank could
default on its obligation at the end of the day. This intraday (or day-
light) risk could easily generate systemic risk ‘the risk that the failure of
6 The Structure and Economics of Payment Systems

Debtor/Payer
Creditor/Payee

Bill for 1000

Credit advice
Payment
and/or
instruction
statement
Clearing: a Process
Payments to
customers of ACH Payments for
other banks customers of C
Bank D ⫺1000 ⫹1000 Bank C

Net value of all payments

between banks C and D,
e.g. C owes D 10,000
Settlement: an Event

Bank D ⫹10,000 Bank C ⫺10,000

Central Bank

Figure 1.2 The clearing and settlement chain

one participant in a transfer system, or in financial markets generally,

to meet its required obligations will cause other participants or financial
institutions to be unable to meet their obligations (including settlement
obligations in a transfer system) when due’.3 Central banks imposed
therefore that large value transactions had to be settled gross (without
netting) as they are initiated, giving birth to Real-time Gross Settlement
Systems (RTGS) which debit and credit the banks’ accounts at the central
bank in real-time, providing instant irrevocable finality. These systems
normally process a relatively small amount of large-value payments: in
the UK for instance, the CHAPS sterling RTGS system accounted in 2006
for 90 per cent of non-cash payment values (in £s) as opposed to only
0.2 per cent of the volume (number) of non-cash payments!4
This move to RTGS for the large-value payments resulted in dramatic
changes in the way banks manage their treasury and liquidity as shown
in Figure 1.3.
For a net system settling end-of-day, it suffices for the bank to have
enough liquidity to settle its end-of-day net position, its short positions
(amounts it owes to other banks) being reduced by its long positions
(amounts owed to it by other banks). In a RTGS system, the balance on
 The Architecture of Payment Systems 7

Time Collateralized
Balance

Balance
credit line

Cut-off

Time
Net payment system: RTGS: need to maintain
need to fund final position balance within limit throughout the day

Figure 1.3 Intraday liquidity management is essential in a RTGS environment

its settlement account with the central bank will fluctuate according to
the payments it sends and receives. Central banks, no longer willing to
act as ‘lender of last resort’, will only grant credit facilities to the commer-
cial banks against collateral, normally securities on deposit or repurchase
agreements (repos), which represents an opportunity cost to the banks
who cannot trade them. Payments which cannot be effected because they
would breach the credit limit are queued, therefore delayed, awaiting
incoming payments or a further injection of liquidity. Banks must there-
fore manage their liquidity carefully during the day, seeking to minimize
the collateral posted to secure credit facilities. We will see in section 3.3
of this chapter how RTGS systems have evolved to optimize liquidity.
Payment systems generally operate during pre-determined opening
hours; most important is the cut-off time beyond which payments
received will not be processed and are carried over until the next
working day.

2 Participants in a payment system

2.1 The banks
Banks are the compulsory intermediates between users and payment sys-
tems as they hold a license to take deposits and effect payments for which
they are subjected to regulation. They maintain accounts on behalf of
their customers which are debited or credited when a payment is effected
or funds are received.
8 The Structure and Economics of Payment Systems

If a payment is effected between two accounts held by the same bank,

the amount will be transferred between the debtor’s and the creditor’s
accounts. These intrabank payments, known as ‘on us’ or ‘book entry’
payments, do not affect the bank’s overall treasury position. If, however,
the creditor’s account is held by a different financial institution this inter-
bank payment results in a debt between the two banks which has to be
settled through correspondent accounts (accounts held by banks with
each other) or a payment system, which impacts the treasury, liquidity
and risk positions of both banks.
Although access to most payment systems is restricted to banks, several
disintermediation factors have emerged over the past 10 years.

Legal: Deposit taking is a regulated activity subject to minimum capital

requirements, deposit insurance and supervision by a national regula-
tory authority, for instance the Financial Services Authority (FSA) in the
UK, The Federal Reserve System in the US, or the Banque de France in
France. However, the banks’ monopoly on payments is gradually erod-
ing. The recently enacted Payment Services Directive (PSD) in the EU
allows non-banks to offer payments services; these payment institutions
will be subject to much lighter regulatory requirements (see ch. 6 sec. 2).

Functional: Non-banking payment systems operate with or without the

involvement of banks. Most multinational corporations operate inter-
nal payment and netting systems for transfers between their different
national subsidiaries and/or affiliated legal entities to reduce banking
fees and float. They will also operate treasury and financing subsidiaries
in tax-efficient locations, often with limited banking licences, for inter-
nal netting, balance optimization and often, trading in foreign exchange,
debt instruments and derivatives. In the best cases, banks intervene to
provide access to payment systems and/or to manage net balances.

Commercial: Several closed payment systems operate in every country,

particularly stored-value prepaid cards offered by transport authorities
(for instance Oyster in London, MetroCard in New York, Navigo in Paris)
and mobile phone operators. Other organizations (for instance Western
Union and MoneyGram) offer cross-border low value transfers, known
as remittances, to migrant workers sending money to their families back
home (see ch. 3 sec. 4). Store and supermarket chains, which until
recently just offered store cards accepted only by their outlets, now offer
credit cards co-branded with Visa or MasterCard thereby gaining uni-
versal acceptance: banks are losing out on the fees and the interest on
 The Architecture of Payment Systems 9

outstanding balances. Again, it should be noted that the processes and

IT systems required to operate these schemes are often subcontracted to
banks or payments processors.

Technical: The internet and mobile telephony have enabled a host of

parallel payment systems to flourish and disintermediate the banks. Pay-
Pal, building on the success of the on-line auction system eBay, is gaining
its market share in the person-to-person (P2P) segment (see ch. 9. sec.
5). Several mobile phone operators are offering mobile payment services
either independently or in cooperation with banks (see ch. 9 sec. 4). If
they are not totally by-passed the banks are forced, at the very best, to
share the revenues.

2.2 Membership and economics of payment systems

Membership of a clearing house or RTGS system implies meeting spe-
cific eligibility criteria as well as the implementation of procedures and
IT infrastructures to comply with the operating rules and performance
criteria set by the system.
Membership is often two-tiered, differentiating between direct members
and indirect members – generally banks which do not meet the eligibility
criteria or are unwilling to invest to comply with the technical and oper-
ational requirements. Indirect members participate by clearing and/or
settling through a direct member. In this case, the direct member assumes
responsibility for settling and managing the liquidity on behalf of the
indirect member, activities for which direct members compete on ser-
vices offered, credit lines and fees. Direct members can offer a better
service to their customers: they will accept payments later, credit benefi-
ciaries earlier and will be able to offer lower charges. Direct membership
becomes a competitive advantage in addition to the benefits in terms of
float and improved liquidity. These factors need to be taken into account
when banks choose to join as direct or indirect members, assuming of
course that they meet the eligibility criteria.
Financial eligibility criteria revolve around creditworthiness, usually
capitalization and credit rating. The robustness of a payment system is
directly related to that of its weakest direct member, a factor which is
more important in net than in gross systems. As payment systems will
inevitably include banks of varying creditworthiness, they generally offer
facilities by which each bank can set limits, or caps, on other members
to manage its risk towards them. An element of reciprocity can intervene
in these allocations. These caps can be varied throughout the day to take
into account payment volumes . . . or market rumours on a specific bank’s
10 The Structure and Economics of Payment Systems

solvability. Ultimately, a bank can reduce its cap on another bank to zero,
but this drastic measure is only taken under extreme circumstances.
A related criterion is the number of payments a bank will contribute
to the system. Eligibility criteria normally stipulate either a minimum
number of transactions or a minimum market share in terms of value. If
an indirect member’s payment volumes grow, accession to direct mem-
bership will not increase the total number of payments across the system,
except for the small amount which it might settle directly with its former
direct member.
Each payment system sets technical and operational criteria according
to the schemes it operates: messaging and file formatting standards; com-
munication interface specifications; liquidity management; ability to
respect deadlines and response times; and recovery and backup require-
ments in case of technical failure of the payment system itself or of the
members’ back-office systems. These place heavy demands in terms of
investments and human resources which often cause even sizeable banks
to opt for indirect membership.
Payment systems generally operate on a cost-recovery basis: revenues
originating from membership fees, annual charges and transaction fees
should cover operating costs and the funding of new developments.
Some Clearing Houses operate on a cost-plus basis in order to gener-
ate a profit for their owners. RTGS systems operated by central banks are
sometimes subsidized in the interest of risk reduction, the subsidy being
euphemistically referred to as a ‘public good factor’. The annual charges
are usually fixed and therefore to the disadvantage of low-volume mem-
bers. Transaction fees are independent of the value of the payment and
generally reduce in relation to the volume of payments each member
contributes to the system. In the US and particularly Europe since the
advent of SEPA which allows clearing houses to offer services across bor-
ders (see ch. 6 sec. 4), competition is constantly forcing transaction fees
downwards to attract new members and volumes. When a new payment
system is developed, membership fees generally help fund the initial
investments and launch costs. When banks seek to join an existing
system, the fee should theoretically reflect the actual value of the sys-
tem in operation balanced by the fact that the new member contributes
payment volumes which can reduce the transaction fee and improve
liquidity; these valuations are extremely complex and the pursuit of
volumes to achieve economies of scale in the current competitive envi-
ronment generally result in ‘token’ joining fees. Ultimately, the balance
between these pricing components reflects the objective of the payment
system owners: do they wish to expand its use or maintain it as an
 The Architecture of Payment Systems 11

‘exclusive club’? Regulators however demand full transparency of pricing

and participation criteria.

2.3 The settlement agent

The settlement agent manages the settlement accounts of the direct
members and transfers amounts between them to achieve finality. Tech-
nically this role could be undertaken by a commercial bank or a central
bank, but risk management considerations point towards the central
bank which holds the monopoly for issuing legal tender: the credit and
liquidity risk are theoretically nil as only the central bank can issue cur-
rency without limits or security, influenced only by macro-economic
considerations such as money supply, price stability, interest or exchange
rates. This has given rise to the settlement in central bank money doctrine
which dictates that ‘assets used for settlement should preferably be a
claim on the central bank’,5 particularly for systemically important sys-
tems. This categorization will be explained in Chapter 4, suffice it to
state at this point that all large-value systems and national ACHs are
considered to be systemically important.

2.4 The central bank

As we have seen, central banks act generally as settlement agent. They
also, most often, operate the large-value RTGS systems, while the private
sector operates some large-value systems (for instance CHIPS in the US
and the EURO1 system operated by the Euro Banking Association (EBA))
as well as virtually all low-value ACHs and card clearings. Central banks
are however responsible for oversight: ‘a central bank task, principally
intended to promote the smooth functioning of payment systems and to
protect the financial system from possible ”domino effects” which may
occur when one or more participants in the payment system incur credit
or liquidity problems. Payment systems oversight aims at a given sys-
tem (e.g. a funds transfer system) rather than individual participants’.6
The Committee on Payment and Settlement Systems (CPSS) of the Bank
for International Settlements (BIS) in Basel is the main forum where
central banks cooperate internationally to issue common guidelines on
oversight and managing risks in payment systems.
Although bank supervision is not always the responsibility of the cen-
tral bank (for instance in the UK where it has been devolved to the FSA), it
can take action against the entire banking system or individual banks. As
it maintains the settlement accounts, the central bank is well positioned
to monitor each bank’s position in real time: balance, liquidity, number
of payments queued, ability to secure funding in the money markets.
12 The Structure and Economics of Payment Systems

If a bank appears to be in difficulty, the central bank can secure emer-

gency funding or, in extreme cases, suspend the bank from the payment
system; this would however be a very serious decision as it could create
a systemic, or ‘domino’ crisis which would affect the reputation of the
financial marketplace. Whatever action it takes, the central bank will be
open to criticism, either for intervening too late, or too early! In theory,
central banks are not obligated to help a commercial bank in difficulty
according to the ‘moral hazard’ principle; historically however, central
banks have intervened in most cases to prevent a run on banks under
the ‘too big to fail’ principle or to prevent a panic, as witnessed during
the liquidity crisis since mid-2007.
Finally, central banks are also active participants in payment systems
for payments between them and commercial banks and ‘open market’
purchase or sale of government bonds to implement their monetary
policy.

2.5 The money market

The money market is an essential component of payment systems
although it is not, strictly speaking, part of them. An efficient and liquid
intraday (for instance repos) market, offering a variety of instruments
with varied maturities, is essential for the smooth operation of a pay-
ment system as it enables the commercial banks to fund their liquidity
and settlement positions. From a macro-economic viewpoint, a payment
system can only function if those members of the clearing with long
positions accept to lend funds to those with short positions. Some pay-
ment systems even incorporate automatic lending-borrowing facilities
to facilitate settlement.
The money market would be ‘perfect’ if:

• all participants had access to the same information at the same time;
• no participant held a dominant share enabling it to influence liquidity
and pricing (interest rates); and
• the market was sufficiently liquid.

In practice, imperfections in the market are introduced by the bilateral

credit lines which limit the funds a bank is prepared to lend to another.
In addition to the interbank money market, whether directly between
institutions or through brokers, the central bank can also intervene by
granting credit to the commercial banks, generally end-of-day when the
money market closes and dealers have squared their positions. In certain
countries, this facility is known as the Discount Window, referring to the
 The Architecture of Payment Systems 13

Indirect
member

Direct Net
member balances
ACH RTGS
Central
bank

Money
Direct market
member

Figure 1.4 The chain of payment operations

time when banks would send representatives to a teller window at the

central bank to negotiate credit facilities.

3 Payment operations
3.1 Payment initiation
The payment instrument is agreed between the payer and the bene-
ficiary. In the case of a remote (not face-to-face) electronic payment,
for instance to pay a utility bill, it could be a cheque, a credit trans-
fer initiated by the payer or a direct debit where the beneficiary has
a mandate to draw funds from the payer’s account. In the case of a
credit transfer, the payer will issue a payment instruction and send it
to his bank. If this instruction is not in a compatible electronic form
(for instance from internet banking), bank staff will input the payment
details, an operation which inevitably creates the risk of transposition
errors. If the payer’s bank is an indirect member of the system, it will
transmit the instruction to its direct member after checking the bal-
ance or credit limit of its client. The direct member then transmits the
instruction to the clearing house if it is a low-value payment or to the
RTGS payment system if it is a high-value payment. If this sequence of
processes, see Figure 1.4, is entirely electronic with no human inter-
vention likely to cause errors, it is known as STP (Straight Through
Processing).
14 The Structure and Economics of Payment Systems

3.2 Clearing and ACHs

The ACH receives batches of payments from all banks which are first
validated in terms of formatting and non-duplication. Some ACHs will
also receive files of payments directly from corporations, for instance
the payroll, which they will process subject to authorization from the
account holding bank. This authorization can either be for each file, or
against credit limits which the bank sets on each customer registered to
use this corporate access service. Files are opened and the payments are
sorted and merged into files of payments for the banks of the benefi-
ciaries. Procedures exist to handle returns, payments which cannot be
effected because of incorrect account numbers, closed accounts and, in
the case of direct debits, insufficient funds. Some ACHs incorporate risk
management procedures enabling banks to place limits on each other
or imposing net debit caps on each member’s overall position. Once a
payment has entered the ACH, it cannot be cancelled; should this be
necessary, the bank would have to ask the beneficiary’s bank to initiate
a reverse payment.
The ACH also calculates the net positions. These can be either bilat-
eral between each member or multilateral (also known as net/net): the
algebraic sum of the bilateral positions of each bank resulting for each
into one position vis-à-vis the system: short if the bank owes money or
long if it is owed funds (see Figure 1.5). Multilateral netting reduces the
amounts and the number of payments each participant has to handle.
These net positions are then transmitted either to the central bank for
settlement, or to the relevant RTGS system which will settle them with
other high-value and systemic payments. The ACH is financially neutral
as all net positions should algebraically add to zero. Clearing houses also
issue reports for reconciliation and maintain audit trails and historical
data for queries, investigations, billing and statistics.
These payment systems are known as Deferred Net Systems (DNS)
as settlement takes place at some later time. DNS systems would gen-
erally settle end-of-day: batches were transmitted to the banks in the
evening which would process them overnight and credit the beneficia-
ries next day if not later. Several ACHs now run multiple settlement cycles
throughout the day to reduce the window of intraday risk and provide
earlier availability of funds.
In the current competitive climate among ACHs, most now offer
additional value-added services: back office processing for banks and
corporates, queries and investigations, e-billing, mobile payments, etc.
It should be noted that some countries do not operate an ACH (for
instance Australia, Germany, Ireland and Finland) in which case the
 The Architecture of Payment Systems 15

40 40
20 70

20

60
Gross amounts

10
20 30
20 10

40
Bilateral netting Multilateral netting

Figure 1.5 Bilateral and multilateral netting

banks exchange files of payments bilaterally, agree the net amount, and
settle through the local RTGS system. The generic term Clearing and Set-
tlement Mechanism (CSM) is therefore used to describe these operations
irrespective of whether an ACH is involved or not.

3.3 RTGS systems

As briefly explained in sec. 1 of this chapter, RTGS systems, which han-
dle a small amount of large value payments, settle payments one-by-one
gross through the settlement accounts held with the central bank. Pay-
ments, after format and non-duplicate validation, are only processed if
sufficient funds or credit is available at the initiating bank’s settlement
account. Payments which cannot be effected are queued and conse-
quently delayed. Each bank must therefore carefully assess its liquidity
requirements throughout the day:

Liquidity = funds brought in by the bank + incoming payments

+ collateralized credit line negotiated with the
central bank − outgoing payments.

At the beginning of each day, banks will transfer funds into their settle-
ment account and/or post the collateral required to secure the neces-
sary credit facility. Should the queue lengthen because of insufficient
16 The Structure and Economics of Payment Systems

incoming payments, the bank must either top-up its settlement account
by transferring own funds, accessing the money market, or post addi-
tional collateral with the central bank to extend its credit line. Efficient
liquidity management is essential in RTGS systems as the substantially
higher cost of RTGS payments relative to ACH payments is less depen-
dent on the processing charge than on the cost of the liquidity: interest
on money market operations or the opportunity cost of immobilizing
securities for collateral.
For this reason, great efforts have been deployed to implement
liquidity saving features in RTGS systems:

• Priority levels: high level priority payments will always take precedence
over lower priority payments: separate queues are maintained for each
priority. The highest priority is normally reserved for payments related
to operations with the central bank and the settlement of DNS pay-
ment systems, or securities clearing systems, which settle through
the RTGS system (known as ancillary systems). The higher priority
queue(s) are normally processed on a first-in-first-out basis (FIFO).
• Queue management: up until settlement, payments can be re-ordered
within queues, moved between priorities or even cancelled.
• Offsetting payments: a lower priority payment from bank A to bank B
will be delayed until a payment from bank B to bank A is presented:
both payments will be submitted simultaneously so that only the dif-
ference will reduce the liquidity. RTGS systems which also include
such netting facilities are known as hybrid payment systems.
• Liquidity reservation: liquidity can be set aside for high priority
payments and the settlement of ancillary systems.
• Timing of payments: earliest and/or latest submission times can be
allocated to payments, which can be changed before settlement.
• Liquidity pooling across the various subsidiaries and foreign branches
of a multinational bank.

In addition, banks can limit their risk vis-à-vis other direct members by
setting bilateral limits against individual banks and/or multilateral limits
against groups of banks which can be changed throughout the day.
A situation may arise when the system is gridlocked, meaning that
payments are queued because of insufficient funds on some banks’ set-
tlement accounts which, if settled, would lift the balance to allow other
banks’ payments to be settled. In Figure 1.6 we can see that payments
are queued for banks A, B and C which, if released, would allow all
to be settled. Facilities exist therefore for authorized staff at the central
 The Architecture of Payment Systems 17

A to B: 200

Bank A: Bank B:
position 100 B to C: 300 position 200

C to A: 500

Bank C:
position 300

Figure 1.6 Gridlock

bank to release payments to solve the gridlock. Modern systems include

automated gridlock resolution routines which allow payments to flow in
most cases. Information and control facilities are also available to enable
banks and the central bank to monitor balances, liquidity, limits and
track progress of individual payments in real-time. RTGS systems also
issue end-of-day reports for reconciliation and maintain audit trails and
historical data for queries, investigations, billing and statistics.
We will see in later (ch. 6 sec. 3) how these various features operate in
the euro based TARGET2 RTGS.

3.4 Communication networks

The transmission of payments and reports between customers, the banks,
the ACHs, the RTGS systems and the central bank should take place over
secure and resilient transmission networks. SWIFT, a bank-owned global
network for financial messages and service provider (see ch. 3 sec. 2), has
established itself as the preferred transmission network for large-value
systems.
Several routing solutions have evolved over time in response to require-
ments emanating from the scheme owners, designated by the capital
letter they resemble.
18 The Structure and Economics of Payment Systems

Payment
system

Figure 1.7 V routing

Full or partial copy

Payment
system

Figure 1.8 T-copy

The simplest is the V routing, whereby the payment messages are sim-
ply transmitted between the direct members and the payment system
(see Figure 1.7).
In the T-copy routing, messages are sent between banks and copied to
the payment system (see Figure 1.8).
The copy can either contain the full payment message, or only the
information necessary for clearing and/or settlement: essentially identi-
fiers for the payer’s and beneficiary’s bank and the amount; information
such as the originating and beneficiary customers as well as the motive
for the payment, such as an invoice reference, are not required for
settlement.
The most sophisticated is the Y-copy, essentially used for RTGS systems
(see Figure 1.9).
 The Architecture of Payment Systems 19

Send payment

Optional sender notification Add booking time and transmit

with booking time original payment

SWIFT

Settlement request with full or stripped copy,

Settlement confirmation
e.g. only bank identifiers and amount

RTGS system
Settle payment subject to:
validation, liquidity, limits, etc.

Central bank
Simultaneous debit/credit amount accounting
A B

Figure 1.9 Y-copy

The message is copied to the RTGS system and held by SWIFT until set-
tlement confirmation has been received from the RTGS system so that
the receiving bank knows that the funds have been irrevocably settled.
The architectures described in this opening chapter represent the
‘established order’ and the paradigms prevailing until the end of the
twentieth century. We will develop in subsequent chapters how these
models and the payments business have evolved under global competi-
tive pressures and regulation.

4 Standards
Standards are an important element of payment systems. They ensure
that all participants can automate the process by specifying that, within
a message containing payment details, each field (such as name of benefi-
ciary, amount, etc.) can be uniquely identified and that the information
is transmitted using the same format to avoid, for instance, the potential
confusion created by an Englishman writing a100.60 and a French-
man writing a100,60. Each country has developed its own standards for
domestic systems while all international payments use the SWIFT stan-
dards (see ch. 3 sec. 2). The current trend is to move towards the ISO20022
standard for electronic payments, which is a methodology by which
20 The Structure and Economics of Payment Systems

standards can be created and a framework by which standard syntaxes

can be made to coexist. The adoption of the UNIFI ISO 20022 XML syn-
tax enables interoperability between messaging standards, for instance,
between those adopted for payment initiation between the bank and its
customer and those for messages between banks. XML standards include
a naming field (tag) and the characteristics (attributes) for each data
item. The adoption of the UNIFI ISO 20022 XML syntax enables there-
fore interoperability between messaging standards, for instance, between
those adopted for payment initiation between the bank and its customer
and those for messages between banks, each application capturing the
data it requires. Several payment service providers and infrastructures
have developed converters between the various industry and national
formats which, while easing transition, can encourage inertia.
Standards have been developed for all instruments: magnetic stripe
and chip for cards, optical or magnetic character recognition as well as
imaging for cheques and notes and coins for cash!

5 Efficiency criteria for payment systems

The efficiency of a payment system is generally measured by three
criteria: execution time; risk; and cost.
Execution time represents a cost for the customers who do not have
access to funds: this float is dependent on the interest rate and elapsed
time. Risk can be quantified as the cost of the risk management proce-
dures, such as the opportunity cost of assets immobilized by collateral
requirements. Costs reflect not only the processing and infrastructure
costs, but also the cost of liquidity, be it interest charges or collat-
eral. These criteria are interdependent and are ultimately reflected in
the cost, the final choice representing a trade-off: RTGS systems, for
instance, minimize risk but at the cost of collateral. A high-volume
retailer who operates on competition-driven margins will be more sen-
sitive to transaction costs and might refuse to use instruments which
bear high handling charges and operational overheads, such as cheques,
which might wipe out the company profit. Each participant in the value
chain will have its own selection criteria.
Retail customers, or individuals, are most sensitive to transaction
costs, less to execution times except when needing to transfer emergency
funds; the risk management costs are hidden from them.
Small and medium enterprises (SMEs) are also most sensitive to costs
and execution times. Large corporations are more interested in acceler-
ating receivables than transaction costs. This customer segment is also
 The Architecture of Payment Systems 21

best positioned to negotiate fees and execution times with the banks and
issue requests for competitive proposals. We will see in Part IV how the
banks attempt to lock them in with cash management and other value-
added services. Businesses are also, generally speaking, impervious to the
risk management costs.
Commercial banks are most sensitive to the profitability of payment
services: maximizing revenues and reducing overall costs. These include
clearing and settlement fees, investments in infrastructure and operat-
ing expenses, as well as the cost of liquidity and collateral reflecting
risk management. Certain banks specialize in payment services, seek-
ing participation in clearing and settlement systems in several countries
to accelerate execution times. They are also constantly streamlining
processes and upgrading their technology to improve STP and liquid-
ity management (see ch. 14 sec. 1). Faced with increased investments,
several small and medium-sized banks are seeking to outsource their pay-
ment operations to these specialized transaction banks, which are keen
to increase volumes to reduce unit processing costs through economies
of scale.
Central banks are primarily concerned with risk minimization, partic-
ularly systemic risk which could destabilize the entire financial system.
They have responsibility for oversight over payment systems and to
promote technological innovation for risk minimization and smooth
liquidity management. They also require efficient systems as a tool to
implement their monetary policy, to rapidly transmit their interventions
and to measure their impact.
We can see that payment systems are not neutral. They transfer funds
and demand liquidity. They require heavy investments, operational dis-
cipline and resources from the commercial banks, as well as vigilance
from the central banks and regulators.

Box 1.1 An international payments system during the

Middle Ages: the Papacy
For security reasons, the Papacy transferred itself to Avignon, in the
south of France, from 1309 to 1418. This coincided with a period of
turbulence between 1378 and 1417, during which the Church was
ruled by no less than four popes and anti-popes until the Council of
Konstanz ended the Great Western Schism. Substantial funds and new
financial collection and management systems were required to recruit
mercenaries to regain the Papal States, as well as build the sumptuous
22 The Structure and Economics of Payment Systems

Palais des Papes. Italian popes often came from merchant and banking
families and were therefore fully aware of the latest financing tech-
niques. They would appoint a senior Church official, experienced in
banking and payment networks, to manage the Papacy’s finances.
The Church would maintain a network of tax collectors extending to
the most remote regions: Poland, Scandinavia, the Levant, etc., so
funds had to be repatriated to Rome or Avignon. As physical trans-
portation was too dangerous, the Papacy would contract with the
great Italian merchant and banking families such as the Medici or
the Bardi. The pontifical tax collectors would remit the taxes into
the foreign branches of the bankers, who would make them available
(minus a pre-agreed commission) to the Papacy in Rome or Avignon
by using the funds deposited with them by the Church dignitaries and
members of the Curia. They would use the funds collected locally
to purchase goods, such as wool from England which was sold to
the Florentine weavers for the sumptuous cloths and robes we see
today in the portraits by Holbein and Titian. The bankers ensured the
safe availability of funds collected remotely, the settlement and the
foreign exchange, all sources of fees.
2
Payment Instruments

Buyer and Seller must first agree on a payment instrument, be it cash,

cheque, card or electronic. These various instruments are the ‘raw mate-
rials’ of payment systems and have evolved in response to demands for
ease of use, cost reduction, security and more information, as well as
technological progress.

1 Characteristics of payment instruments

The choice of a payment instrument represents a compromise between
the various counterparties based on consideration of the features and
benefits:

• ease of use and convenience for the debtor or the creditor;

• terms, conditions and execution time: the beneficiary, in particular,
wishes to know when the funds are available for him to draw upon;
• ease of automation, not only for processing the payment but also
transmitting the reason for the payment, or remittance information,
to facilitate reconciliation;
• costs, in terms of fees charged to the initiator and/or the beneficiary,
as well as processing costs to the service providers including the cost
of liquidity;
• security, expressed in terms of authenticity, confidentiality and
integrity: the assurance that the declared source is the true source and
that no outside party could have seen and/or changed any of the data:
amount, beneficiary’s name, references, etc. Another factor gaining
importance with internet banking is non-repudiation: the inability
for a counterparty to deny that it has taken a specific action; and

23
24 The Structure and Economics of Payment Systems

• auditability and traceability: the ability to prove that a payment has

been effected and/or received, as well as facilities to track and trace
the payment in case of delayed receipt or queries.

Retail customers mainly base their choice on convenience and execu-

tion times, principally the date at which their account will be debited,
less so as to when the beneficiary will be credited unless for emergen-
cies, for example, transfers to children at university having prematurely
exhausted their monthly allowance!
Corporate customers are mainly concerned with fees and execution
times; they are anxious to optimize their cash flow: accelerated avail-
ability of remitted funds, latest possible debiting of their account when
initiating payments and minimizing idle balances. Fees and execution
times are keenly negotiated between large corporations and financial
institutions as we will see in Chapter 13 (sec. 4). Enterprises also favour
payment instruments which maximize the accuracy of the remittance
information so that they can reconcile, for instance, payments received
against invoices issued.
Banks and service providers will be mainly concerned with processing
costs and security. They are keen to reduce the processing costs of generic
payment services to a minimum so as to devote funds and resources to
value-added services and customer relations. This is achieved by max-
imizing STP and automating the entire chain of processes: receipt of
payment instruction, validation, balance check, debit/credit accounts,
transmission to clearing and settlement mechanism through to final rec-
onciliation; any manual intervention is to be avoided as it will increase
costs (see ch. 14 sec.1).
Central banks are mainly concerned with security and minimizing
risks (see ch. 4). As we have seen, they are responsible for over-
sight and sometimes regulation and operation, but should not inter-
fere in the competition between service providers. They should how-
ever promote efficiency and ensure that technological progress is used
advantageously.
Before we examine the various payment instruments in more detail,
it is worth looking at two other factors which influence the choice of
instruments:

• Circumstances: face-to-face when creditor and debtor are in physical

presence of each other, as opposed to remote payments when mail
and/or electronic transmission must be relied upon; and
 Payment Instruments 25

• Frequency: occasional (or one-off) payments for shopping or, for

instance, professional fees, as opposed to recurring payments such
as mortgage repayments, insurance premiums or utility bills.

2 Cash
Cash – notes and coins – is the oldest payment instrument since mankind
progressed beyond barter. Coins are usually minted by the government
(the Mint in the UK, the Hôtel des Monnaies under the Ministry of
Finance in France), while notes are printed under the authority of the
central bank, either by themselves – the Bank of England – and/or sub-
contracted to other central banks or specialist security printers such as
De la Rue or Giesecke and Devriendt. Several illustrious historical figures
have been in charge of issuing notes and coins, from Thomas Gresham
and Isaac Newton to . . . Che Guevara who was governor of the central
bank after the 1959 revolution in Cuba.
Cash is linked to the concept of seigniorage: ‘In a historical con-
text, the term seigniorage was used to refer to the share, fee or tax
which the seignior, or sovereign, took to cover the expenses of coinage
and for profit. With the introduction of paper money, larger prof-
its could be made because banknotes cost much less to produce than
their face value. When central banks came to be monopoly suppliers
of banknotes, seigniorage came to be reflected in the profits made by
them and ultimately their major or only shareholder, the government.
Seigniorage can be estimated by multiplying notes and coin outstand-
ing (non-interest bearing central bank liabilities) by the long-term rate of
interest on government securities (a proxy for the return on central bank
assets)’.1
Cash has the advantage of providing instant finality and discharge of
debt, but is bulky and expensive to handle in terms of transport, stor-
age, security and counting. For this reason several countries have passed
legislation to ensure that salaries, pensions and social benefits are paid
by cheque and/or electronic credit transfers. The scenarios of gangster
films based on attacking the payroll vans are today obsolete, but cash
still accounts for the largest number of personal payments (63% in the
UK, two-thirds of which being five pounds or less in 20062 ), so hold-ups
on security transport vans and their staff are still common . . . no change
in the scenario from attacking the Wells Fargo (the precursor of the global
Californian bank) stagecoach in westerns!
Cash handling costs are estimated at a45–70 billion in the EU, or
0.4–0.6 per cent of GDP.3 No explicit charge is made to retail customers
26 The Structure and Economics of Payment Systems

for cash handling, but banks attempt to recover transport and handling
costs from large retail outlets such as supermarket chains.
Recent anti-money laundering (AML) legislation also compels mer-
chants to report cash payments in excess of a certain amount which
varies by country.

3 Cheques
A cheque (or check in the US) is a signed written payment instrument
drawn by the debtor (or payer) on his/her bank and presented, either
face-to-face or by mail, to the creditor (or payee). The cheque is a ‘pull’
payment. The theoretical sequence of events should be:

• creditor presents the cheque to his/her bank (collecting bank) who

verifies that amounts in figures and in letters match;
• creditor’s bank sends cheque to debtor’s bank (paying bank), either
directly or via a clearing house;
• clearing house sorts the cheques received from the collecting banks
and sends them to the paying banks;
• payer’s bank verifies debtor’s signature and balance (or credit line) on
the account;
• payer’s bank notifies creditor’s bank that the cheque will be hon-
oured and that the funds can be credited to his/her account, or that
the cheque is refused for insufficient funds (commonly known as
‘bounced’) or suspected fraud, in which case the dishonoured cheque
is returned; and
• payer’s bank returns the cheque to the drawer with the statement of
his/her account.

In cases where immediate acceptance is required, banks will issue a

banker’s cheque (or draft) after debiting the debtor, therefore guarantee-
ing good funds; these drafts can become negotiable instruments, hence
their name assegni circulari (circulating cheques) in Italy.
This is a long, cumbersome and expensive process which has proven
difficult to automate. The magnetic (MICR: magnetic ink character
recognition) or optical (OCR: optical character recognition) encoding or
pre-printing of the drawer’s account identifiers (account number, gener-
ally accompanied by a sort code) and cheque number, allied with progress
in optical recognition of the handwritten amount in figures (but not the
amount in letters or the beneficiary, even less the signature) have greatly
facilitated automation.
 Payment Instruments 27

Legislation was first passed dispensing the banks from returning the
cheques to the drawers, but compelling them to store them – either
physically, or on microfilm or digital image. To increase acceptance by
retailers, cheque guarantee cards were introduced which guaranteed the
cheque up to a specified amount, subject to the creditor verifying the
card number written on the reverse of the cheque, the signature and in
some cases the photograph – this however only being truly effective in
the case of face-to-face payments.
The next step in certain countries was cheque truncation, whereby
the data is captured by the creditor’s bank (who stores the cheque or its
image and charges the debtor’s bank for his efforts) and transmitted elec-
tronically to the debtor’s bank or the clearing house – thereby omitting
signature verification! Cheque imaging is also gaining wider acceptance
(see ch. 7 sec. 5). Under customer pressure, banks have recently cred-
ited the beneficiary immediately (especially if the cheque is guaranteed
by a cheque guarantee card), but reserve the right to recover the funds
should the cheque not be honoured. The system is therefore wide-open
to fraud and counterfeiting, the onus resting squarely on the debtor to
verify his bank statement regularly and report any cheque debit which
appears suspect: this triggers the recovery of the original cheque or its
image for investigation.
The average cost of processing a cheque is estimated at 6.3 cents4 in
the US.
The number of cheque payments is declining regularly (8 per cent in
the UK in 20065 ), which means that the processing costs per item are
rising as the infrastructure costs are largely fixed. Economies of scale are
essential, so cheque processing is generally outsourced – in Great Britain,
all banks entrust cheque processing and clearing to their jointly owned
Cheque and Credit Clearing Company.
Corporations dislike cheques as they require manual handling and
reconciliation is difficult, relying mainly on the drawer scribbling the
invoice number and/or customer reference on the reverse! Many leading
retailers and petrol chains in the UK refuse to accept cheques. Cheques
remain however popular with retail customers and small businesses on
account of convenience (particularly for remote occasional payments),
force of habit and the float – ‘the cheque is in the mail (!)’. France
and the US remain the largest cheque users, while some countries (for
instance Sweden, the Netherlands, and Japan for retail customers) have
withdrawn cheques. Several countries however have taken measures
to proactively reduce the usage of cheques by differential pricing to
encourage the use of more efficient instruments.
28 The Structure and Economics of Payment Systems

Originator’s Beneficiary’s
bank bank

Clearing and
3 Settlement 4
Debit originator’s Mechanism Transmit
account and send credit transfer
credit transfer message message 5
2
Credit
Instruction
beneficiary

1
Invoice
Originator Beneficiary

Figure 2.1 Credit transfer

4 Credit transfers
Credit transfers (or direct credits) are initiated by the debtor (or origi-
nator) who instructs his bank to debit his account; the bank verifies the
instruction and availability of funds prior to transferring the information
to a clearing house or directly to the beneficiary’s bank which credits the
latter’s account after verification (see Figure 2.1). The credit transfer is a
‘push’ payment.
Exceptions are defined as:

• Rejects: credit transfer rejected by the originator’s bank before inter-

bank settlement for incorrect formatting, invalid account numbers or
insufficient funds; the originator will be notified with the reason of
the reject.
• Returns: credit transfer rejected after interbank settlement if the ben-
eficiary cannot be credited, for instance if the account number is
incorrect or has been closed; the creditor’s bank will advise the
originator’s bank which will notify and refund him/her.

The scheme rules will define a maximum execution time following

acceptance by the debtor (or originator’s) bank, expressed in interbank
business days.
 Payment Instruments 29

The majority of credit transfers (payroll, corporate payments, pen-

sions and social benefits, etc.) are initiated electronically from internet
banking portals, accounting software packages or Enterprise Resource
Planning systems (ERP supplied for instance by SAP or Oracle). Paper
forms, optimized for optical reading, are used for transfers initiated man-
ually. Banks will issue to each customer preprinted forms containing their
name, address and account identifier, either as initiator or beneficiary.
Conversely, enterprises will send with the invoice a computer-generated
preprinted form with their account identifier, name and address, amount
and invoice reference: the customer needs only to enter his name and
account details, sign the form and send it to his bank.
The credit transfer offers therefore major advantages in terms of auto-
mated processing; reconciliation is good if the reference is provided by
the creditor, less reliable if entered by the debtor. Processing costs are
low as even paper forms achieve a very high (over 98 per cent) STP rate
thanks to progress in optical character recognition (OCR).
For regular payments for the same amount (for instance rents), stand-
ing orders (STO) are a repetitive credit transfer whereby the debtor
instructs his bank to transfer the sum at regular intervals – monthly,
quarterly, annually.
For completeness sake, we should also mention the giro transfers
offered by the post office banks. These were mainly established during
the late nineteenth and early twentieth centuries to provide deposit and
payment services to rural populations, as banks were then mainly con-
centrated in cities. The first was created by the Austro-Hungarian empire
(the Kaiser Franz-Joseph even symbolically opened an account), followed
by Switzerland, Japan, Germany, Benelux and France. Customers opened
accounts from which they could initiate credit transfers, but no over-
drafts or credit facilities were allowed. Transfers can also be initiated by
paying in cash at the post office counter, which remains the preferred
way for paying taxes and bills in many countries. The ability to initiate
payments in cash provides anonymity, which is sometimes taken advan-
tage of for illicit transfers. The terms ‘postal cheque, chèque postal’ are
therefore largely a misnomer as the payment is effectively a credit trans-
fer. It should be noted that several of these postal banks have today
become powerful financial institutions – for instance Die Postbank in
Germany and the Post Office Bank in Japan – leveraging the large number
of accounts opened by retail customers and small businesses and compet-
ing aggressively in those customer segments with commercial banks, by
offering a full range of financial services including loans, cards, foreign
exchange and even insurance.
30 The Structure and Economics of Payment Systems

Debtor’s Creditor’s
bank bank

Clearing and
5 Settlement 4
Check Mechanism Instruction
6 mandate, to collect, 3
Check transmit including Collection,
mandate, collection mandate including
debit details mandate
account details

2
Pre-notification (e.g. invoice) with date of collection

1
Mandate

Debtor Creditor

Figure 2.2 Direct debit

5 Direct debits
Direct debits are payments initiated by the creditor through its bank,
which collects (draws) the funds from the debtor’s account at his/her
bank, subject to a legally binding mandate agreed by the debtor (see
Figure 2.2). Direct debits generally rely upon a guarantee to the debtor
that he will be able to recover the funds collected in case of error or
dispute within a specified time limit. The direct debit is a ‘pull’ payment
and can be used for one-off or regularly occurring payments. In certain
countries the creditor’s bank charges an interchange fee, also known as
a Multilateral Interchange Fee (MIF), to the debtor’s bank.
Exceptions, also known as the ‘Rs’, are:

• Rejects prior to interbank settlement for technical reasons such as

formatting error, invalid account numbers, absence of mandate or
non-compliance with the mandate.
• Refusals initiated by the debtor before settlement, either by challeng-
ing an individual invoice or by withdrawing the mandate; if received
after settlement, the refusal will trigger a refund.
• Returns initiated post-settlement by the debtor’s bank if the collec-
tion could not take place on account of insufficient funds, incorrect
account number, account closed or death of the debtor.
 Payment Instruments 31

• Reversals if the Creditor withdraws the collection, requiring a refund

if post-settlement.
• Refunds if the debtor requests reimbursement.

The scheme rules will normally define the following times and deadlines:

• The notification period by the creditor to the debtor, normally an

invoice or bill stating the date at which the amount due will be col-
lected from his/her account, expressed in calendar days before the due
date.
• The latest reception date of the collection by the debtor’s bank for
a one-off or first of a recurring sequence of payments, expressed in
interbank business days before settlement, to allow verification of the
mandate and account number.
• The latest reception date for subsequent collections in a sequence
of recurring collections, normally shorter than for the first and also
expressed in interbank business days before settlement.
• The latest date for settlement of returns, expressed in interbank
business days after receipt of the collection by the debtor’s bank.
• The deadline for refund requests by the debtor for direct debits covered
by a mandate, which can be different if no mandate has been agreed
by the debtor (unauthorized collection).
• The latest date for settlement of a refund.

The mandate must be signed (by hand or electronically) by the debtor

and a file of mandates is held by the ACH and both banks. It will con-
tain the names, addresses and account identifiers of the debtor and the
creditor as well as the payment reference such as a customer reference
identifier. Two possible mandate flows exist (see Figure 2.3):

• Creditor mandate flow, whereby the creditor – for instance an energy

utility – will obtain the customer’s signature on the mandate and send
it to the creditor’s bank for onward transmission to the ACH and/or
the debtor’s bank; and
• Debtor mandate flow whereby the debtor will sign the mandate and
forward it to his bank for onward transmission to the ACH and/or to
the creditor’s bank who notifies the creditor.

Direct debits have become the favoured payment instrument for

enterprises issuing a large number of invoices/bills, such as financial
institutions for mortgage/loan repayments, utilities or telecommuni-
cations operators, as it enables them to automatically collect variable
32 The Structure and Economics of Payment Systems

Debtor’s Creditor’s Debtor’s Creditor’s

bank bank bank bank

CSM CSM
2 2 4
3 Mandate Signed 3 Mandate
Mandate details mandate Mandate details
details details

1 1
Creditor Creditor
Debtor Signed Debtor Agreement
mandate
Creditor’s mandate flow Debtor’s mandate flow

Figure 2.3 Mandate flows

amounts on a predetermined date, thereby optimizing cash flow and

treasury management. Most mobile telephone companies nowadays
only accept direct debits for subscription customers. Reconciliation is
also automatic as the creditor sets the remittance information. Direct
debits are gradually replacing standing orders as even ‘fixed’ amounts,
such as insurance premiums, will rise with inflation. Many utilities offer
discounts if paying by direct debit to encourage customers to change.
Direct debits also reduce time and costs for debtors who, assuming they
agree the amount, do not need to initiate a payment by credit transfer
or cheque; this also gives them peace of mind that the bill will be paid
and that the electricity or telephone will not be cut off if they miss the
deadline while travelling.
Costs are low as processing is entirely automated once the mandate
has been set up.

6 Cards
Historically, credit cards originated in the US in the 1920s when hotel
chains and oil companies began issuing them to customers. The inven-
tion of the bank credit card is attributed to John Biggins of the Flatbush
National Bank of Brooklyn who invented the ‘Charge-It’ scheme between
the bank’s customers and local merchants in 1946.6
Cards are operated under schemes whereby banks issuing cards to their
customers rely upon the understanding that these cards will be accepted
at merchants acquired (or signed-up) by other participating banks.
 Payment Instruments 33

Revenues: fee per card issued.

Authorization, clearing and settlement charges
to acquirer and issuer.

Scheme operator:
clearing, settlement

Issuing bank Acquiring bank

Interchange fee
Revenues: interchange fee, FX, Revenues: merchant fees
customer fees and interest on credit

€99.5
€100 ⫽ £79.20
⫹ fee

Meal for €100

UK card holder French restaurant

Figure 2.4 Cards

Figure 2.4 illustrates the typical example of a British tourist paying

for a meal in France at a restaurant displaying the logos of the cards it
accepts (for instance Visa and/or MasterCard). The acquiring bank will
credit the restaurant, within an agreed deadline, with the amount of
the bill less a merchant fee, in our case 0.5 per cent which he cannot
charge to the customer. The acquiring bank will obtain refund through
the scheme’s clearing and settlement mechanism from the issuing bank,
which will debit the cardholder by the full amount of the restaurant bill,
converted into his/her currency, plus a commission. An interchange fee
is also paid by the acquiring bank to the issuing bank. The scheme owner
gains revenues from fees per card issued under its brand and charges for
authorizations, clearing and settlement.
The same principles apply when cards are used to withdraw cash
at ATMs (automated teller machine, or cash dispenser) operated by a
different bank than the issuer.

6.1 Card types

The most commonly used card is the debit card which is linked to a bank
account, allowing the holder to withdraw cash at ATMs and pay for goods
34 The Structure and Economics of Payment Systems

and services at retail outlets, petrol stations, restaurants, etc. The amount
of the purchase or withdrawal is debited near-instantly from the holder’s
account. Some debit cards are only intended for withdrawals at ATMs
and therefore called cash cards. A cheque guarantee feature (see sec. 3 this
chapter) is often incorporated with debit cards.
The same applies to deferred debit cards except that the amounts are
accumulated, up to an agreed ceiling or limit, until the monthly date
at which the full amount of the purchases since the last statement is
debited from the holder’s account through a direct debit. Credit cards
offer revolving credit facilities whereby the holder, upon receipt of
his monthly statement, can choose to settle the full amount or pay
only part (subject to a minimum), in which case the issuing institu-
tion will charge interest on the outstanding balance. Deferred debit
cards are sometimes assimilated to credit cards as the holder bene-
fits from a credit facility for a maximum of one month, but they
do not offer true credit that enables the holder to postpone part
payment beyond the monthly settlement date, even if willing to pay
interest.
Prepaid cards (or stored-value cards), either for a fixed amount at pur-
chase and disposable or re-loadable, are an alternative to cash and the
stored amount is reduced by each purchase. They are mostly closed sys-
tems used for public transport (for example the contactless Octopus in
Hong Kong, Oyster in London, Z-pass on US toll roads) and increasingly
for mobile telephones. In Hong Kong, the Octopus card has reduced the
weight of coins handled daily from 60 tons to 1 ton.
Electronic purses are prepaid cards accepted at a wider range of out-
lets or even country wide. They have been extremely successful and are
commonly used at a national level in Belgium and the Netherlands (Pro-
ton and ChipNick respectively) for purchases, public transport, parking
meters, etc., but the Moneo card in France has met with only limited
success as small retailers refuse to pay the merchant fee which would
reduce their slim profit margins.
Corporate purchasing or procurement and Travel and Entertainment (T&E)
cards are issued to enterprises to pay for supplies and services, either to
the company itself and/or to selected employees for business travel and
entertainment.
Affinity cards are linked mainly to charities which collect a fraction
percentage of the purchases.
Finally private cards are issued mainly by retail chains and petrol com-
panies for use at their stores and outlets, with or without revolving credit
facilities.
 Payment Instruments 35

6.2 Card technologies

Technology has moved forward since, up until the 1970s, the merchant
would telephone for authorization for sales above a predetermined floor
amount, the cardholder would sign a 3-part paper voucher (one copy
for him, one copy for the merchant, one copy for the acquirer), printed
by the imprinter with the amount added by hand, which was read opti-
cally when sent to the acquirer, giving him/her a very generous float by
the time the details appeared on his statement. The identity of the card
holder was verified by comparing his signature on the slip with that on
the reverse of the card, a weak security had the card been stolen! Card
details were subsequently registered on a magnetic strip at the back of
the card and swiped through a point-of-sale (POS) terminal which would
dial up an on-line authorization centre and, upon acceptance, print the
slip for signature. This reduced costs and float while enabling ‘on-line
authorization to the issuer’ to verify the available credit and whether the
card had not been lost or stolen, but did not significantly reduce fraud
or solve the problem of identity verification.
The next step was the chip card, which originated in France in the
early 1970s, where an electronic chip is embedded in the card enabling
identity verification by the cardholder entering a secret PIN (personal
identification number) on the POS numeric keyboard. With the excep-
tion of the US, this chip and PIN technology is being adopted worldwide
under the EMV (Europay, MasterCard, Visa) standard and has signifi-
cantly reduced fraud. In addition to increased security (the chip destructs
if tampered with), the programming facilities of the chip enable the card
to be used for a multitude of additional functionalities, such as contact-
less cards which need only be waved close to the POS terminal, loyalty
schemes and mobile payments by interfacing with the SIM card of mobile
telephone handsets. Prepaid cards and electronic purses could not have
been introduced without the chip card which also enables several func-
tions to be combined onto one card, the holder choosing whether each
purchase should be charged to his debit card, credit card or electronic
purse.

6.3 Card schemes

The card business is dominated by the two International Card Schemes
(ICS) or networks, MasterCard and Visa. Visa was originally launched by
Bank of America as the BankAmericard in 1956 while MasterCard was
established as a competitive alternative in 1966 as the Interbank Card
Association (ICA). They initially offered credit cards and subsequently
36 The Structure and Economics of Payment Systems

created schemes for debit cards, offering an interoperability framework

and cobranding with national debit card schemes such as Carte Bancaire
in France or Bancomat in Italy; through these it has become possible for
travellers to draw cash from an ATM virtually worldwide.
By September 2006, financial institutions had issued 1.51 billion Visa
cards, used for annual purchases and cash withdrawals amounting to
$4 trillion.7 In 2007, MasterCard’s customers had issued 916 million
cards and purchases on a local currency basis amounted to $2.3 trillion.8
Some banks are dual-issuers, issuing cards from both schemes. Several
anti-monopoly and anti-collusion lawsuits have been launched against
them in the US and in Europe by merchants and the EC, resulting in
multi-million dollar fines and leading both schemes to abandon the
cooperative bank-owned governance model. MasterCard, headquartered
in Purchase (!) NY, floated in 2006 and Visa achieved the richest initial
public offering in US history, raising $17.9 billion on 18 March 2008 in
the midst of the credit crunch from the sub-prime crisis.9 Visa Europe
remained however independent of the new Visa Inc. and retained its
ownership structure by its European member banks. We should remem-
ber that the networks, which are in effect processors, are not affected by
customer defaults as credit is extended by the issuing banks.
The power of the ICS’s brand has become a sensitive issue, but it is
clear that a merchant accepts payment from an unknown customer, pre-
senting a card issued by an unknown bank but bearing the MasterCard
or Visa logo, solely on the certainty that he will be paid through his
acquirer from the same scheme. Some years ago Citibank, arguing that
their brand was stronger than Visa’s, demanded that the Visa logo be
printed on the back of the card; Visa refused and Citibank temporarily
stopped issuing Visa cards.
Other noted brands are American Express who operates a success-
ful closed credit card scheme, sometimes in partnership with a local
bank or airline, Diners (once owned by Citigroup) and Discover, orig-
inally launched by the Dean Witter retail brokerage firm and floated
in 2007 following the merger with Morgan Stanley. In April 2008 Dis-
cover announced a deal to acquire Diners from Citi. These are mainly
targeting high-net worth individuals and the corporate segment. Their
fees are higher than MasterCard’s and Visa’s which explains why fewer
merchants accept them.
Japan operates the JCC scheme and China has recently launched the
China Union Pay cards: with such a large domestic customer base, they
do not feel the need to join the ICS’s. European banks created Europay
featuring the Eurocard in conjunction with the EC debit card in the 1970s
 Payment Instruments 37

which they sold in 2002 to MasterCard Europe; we will see when talking
about SEPA in Chapter 6 (sec. 7.5) how short-sighted the move was.

6.4 Operational and commercial considerations

Card payment processing costs are low and driven by economies of scale.
All basic functions (issuer and acquirer processing, queries and investi-
gations, replacement of lost/stolen cards) are generally outsourced to
shared service centres – some at national scale – large volume financial
institutions or third-party processors; however, the scheme owners retain
the clearing and settlement functions.
From the customer’s standpoint, cards are cheap and convenient when
used as a payment instrument, much less however as a source of credit.
They also offer the best reconciliation facilities through statements list-
ing full details of all transactions: date, name of merchant, amount
in foreign and home currencies, and lately under regulatory pressure
exchange rate and fees. These statements are generally sent by mail but
also recently electronically, allowing automated reconciliation through
home-finance or corporate accounting packages; for corporate cards,
issuers will even sort the payments by cost centre and/or expense type
(travel, procurement, etc.).
From the merchant’s standpoint, sales volumes rise as consumers are
more prone to impulse buying if payment is deferred. The merchant fee
paid to the acquirer is effectively an insurance premium that he will be
paid, providing it has followed the security and anti-fraud measurers
dictated by the scheme. Retailers also benefit from the reduction in cash
handling and safekeeping costs. Nevertheless, merchants are in constant
dispute and even litigation to reduce the merchant fees.
For banks, cards can be an extremely profitable activity. The word
‘can’ is used intentionally as profitability depends on services offered
and sophisticated customer relationship management (CRM). Profitabil-
ity is negligible on the basic domestic payment functions for debit cards
and deferred debit cards, which explains why there is little competition
in countries where those instruments prevail and why customers there
rarely carry a card not issued by the bank holding their current account.
The high interest charged on revolving credit cards is, on the other hand,
an extremely lucrative source of revenue giving rise to fierce competition
and aggressive marketing, where issuers seek to differentiate themselves
through rewards (principally air miles), the credit limit, the interest rate
and interest moratoriums on balance transfers when customers switch.
In January 2008 a survey showed that an estimated 2.6 million British
consumers (7 per cent of credit card customers) planned to transfer credit
38 The Structure and Economics of Payment Systems

card debt run up over the Christmas shopping season to a new card, tak-
ing advantage of 0 per cent interest rate introductory offers by no less
than 169 cards!10
As mentioned in Chapter 1 (sec. 2.1), several non-banks have entered
this market: retail chains, automobile associations, sports clubs, airlines,
etc. Particularly in the US and the UK, individuals will therefore carry
several cards and draw on all available credit lines for payments leading
to high levels of personal debt, creating a social and economic problem
which has raised the concern of central banks.
Issuers endeavour to segment their customers by offering cards with
low limits (for youths and students) and ‘gold’, ‘platinum’ or ‘diamond’
cards to the more affluent customers, appealing to status-consciousness
and linked to profitable value added services: high credit limits (in some
cases unlimited), insurance, concierge services for priority travel, theatre
or restaurant bookings, etc. Issuers are also mining the accumulated data
on their customers’ expenditure preferences and patterns; the first efforts
were mainly aimed at fraud reduction, attempting to detect card thefts
through unusual behaviour. Major efforts are now devoted to CRM to
fine-tune product and service offerings according to perceived individ-
ual customer preferences, as well as to detect utilization patterns which,
correlated with peer behaviour, might indicate potential delinquency or
propensity to switch to another card.
In summary, profitability for the issuer is mainly driven by for-
eign exchange gains on cross-currency purchases, interest revenues for
revolving credit cards and the targeted marketing of value-added services.

7 Payment channels
Payments can be initiated through a variety of channels. Few, except for
high value transfer requests by retail customers, originate today from a
visit to a branch or a free-format letter. Cheques and credit transfers are
still sent by mail, but most banks now encourage customers to initiate
domestic and international credit transfers up to a certain amount, set
up mandates for direct debits and pay bills for pre-established beneficia-
ries such as utilities or telecommunication operators through proprietary
ATM’s, banking kiosks, telephone banking and internet banking.
In addition, the dramatic increase in internet shopping and procure-
ment has demanded the implementation of secure payments over the
internet.
We will discuss in Chapter 4 (sec. 4) the security measures that have
been introduced to prevent fraud.
 Payment Instruments 39

8 Statistics and comparative trends

Inertia is a major factor in the choice of payment instrument: habits die
hard and any plan to shift customers away from one instrument towards
another (for instance move to direct debits, withdrawal of cheques)
requires a concerted effort between banks, regulators, industry bodies
and consumer associations extending over several years. This inertia
is compounded by the fact that cross-subsidization between instru-
ments and customer segments is rife and that pricing is generally not
transparent, particularly to retail customers.

8.1 Cash
More than any other instrument cash is profoundly anchored in col-
lective habits and mentalities and profound differences can be observed
from country to country. Measured as a percentage of GDP, the value of
payments in cash is relatively stable. This explains why new instruments
aimed at replacing cash, such as electronic purses, are slow to gain accep-
tance, even if they offer obvious benefits such as immediate availability
of funds combined with reduced handling costs and improved security
for retailers who would hold less cash on their premises. Small retailers
such as bakers or newsagents are obviously reluctant to pay merchant fees
for card transactions of any type: credit, debit, prepaid or purse. Cash
payments are more common in Germany than in any other European
country; Germans are used to paying large amounts for substantial pur-
chases such as cars in cash, as opposed to cheque or transfer. It was under
pressure from Germany that a a500 note was issued, the highest denom-
ination in any currency, while it would have perhaps been more logic to
provide a a1 note.
Taking the total value of notes and coin in circulation divided by the
population as an indicator, Japanese will hold $5,541 in contrast to
$2,736 in the US and $2,700 in the euro-zone. We note the similarity
between the US and the EU according to this indicator, while the ratio
between cash and GDP shows a slight difference: 6.2 per cent in the
US and 7.7 per cent for the euro-zone. At the other extreme the British
use relatively little cash ($1,443 per capita, 3.4 per cent of GDP and
4.5% of M1 money supply), remembering that the cheque was invented
by Scottish bankers. Two countries diverge substantially from others,
Switzerland and the US. The amount of cash per inhabitant is $5,007
in Switzerland and 9.4 per cent of GDP. The ratio of cash to M1 money
supply is 59 per cent in the US, in contrast to 17.2 per cent in the euro-
zone and 21.6 per cent in Japan. This can be explained by the role of the
40 The Structure and Economics of Payment Systems

Table 2.1 Cash statistics 2006

Cash value per % of GDP % of M1

head of population money supply

Japan 5,541 16.6 21.6

UK 1,443 3.4 4.5
US 2,735 6.2 59.0
Euro-zone 2,700 7.7 17.2

Sources: BIS, CPSS, Statistics on Payment and Settlement Systems in Selected Countries, March
2008

US dollar and the Swiss franc as refuge currencies. Half of US dollar notes
and coin are held by non-residents. Similarly, at the time of the conver-
sion of European currencies to the euro, the Bundesbank discovered that
close to half the stock of large denomination Deutschmark notes were
held in Eastern Europe. We should also remember that the US dollar has
also often been substituted to the local currency (dollarization) by some
countries to fight hyperinflation, as was the case in Argentina. Table 2.1
summarizes the above statistics for 2006.
From the above statistics we can also derive the turnover of the total
value of payments (M1/GDP) and the velocity of money (or speed of
circulation) which is the inverse (GDP/M1). Historically one observes a
decrease in velocity: money supply increases faster than GDP over cen-
turies in line with monetarization of the economy. The trend reverses in
practically all countries after World War II as more money was deposited
at banks and money supply increased slower than GDP.

8.2 Non-cash instruments

The use of non-cash payment instruments is increasing relative to cash
in all countries producing reliable statistic, but usage of scriptural instru-
ments varies from country to country. The volume of credit transfers
remains relatively stable, direct debits are growing slowly but steadily, but
the major trend is the decline of cheques, replaced by cards for face-to-
face retail transactions and credit transfers or direct debits as well as cards
for remote payments. E-payments are still marginal, except in Singapore
where they represent over 84 per cent of the total number of transac-
tions. This is the result of a deliberate policy to create a cashless society.
Some statistics include e-payments, but there is not always agreement
on what the term includes. Other countries with significant proportions
of e-payments include Belgium (4.9 per cent), the Netherlands (3.9) and
 Payment Instruments 41

Table 2.2 Use of non-cash payment instruments in the EU and selected countries
in per cent (2006), excluding e-payments

Credit transfers Direct debits Cards Cheques

Austria 47.5 35.7 15.2 0.3

Belgium 42.5 11.7 43.0 0.7
Bulgaria 68.2 1.6 30.2 0.0
Cyprus 14.8 15.9 32.3 37.0
Czech Republic (2004) 52.9 34.8 10.9 0.0
Denmark 21.6 14.2 62.6 1.6
Estonia 39.7 7.1 53.1 0.0
Finland 42.5 5.1 52.3 0.0
France 17.5 18.3 37.6 25.6
Germany 42.2 42.8 14.2 0.6
Greece 20.0 11.2 49.0 19.0
Hungary 76.7 9.3 13.8 0.0
Ireland 27.6 18.0 33.8 20.6
Italy 29.6 13.3 34.3 12.6
Japan* 33.1 0.0 62.5 10.7
Latvia 63.7 2.2 34.1 0.0
Lithuania 52.1 3.9 43.0 0.0
Luxemburg 48.3 10.1 38.5 0.3
Malta 17.2 3.1 27.0 52.8
Netherlands 32.7 27.2 36.3 0.0
Poland 71.3 1.1 27.5 0.0
Portugal 10.1 11.3 63.6 15.0
Romania 75.7 10.6 9.5 4.0
Singapore1 * 1.0 2.5 6.5 4.6
Slovakia 66.8 16.1 17.0 0.0
Slovenia 54.9 12.6 32.2 0.3
Spain 14.5 44.7 35.7 3.5
Sweden 29.2 10.0 60.7 0.0
UK 21.2 19.8 46.6 12.3
US* 6.6 9.2 51.6 32.6

Note: 1 E-payments represent 84%

Sources: ECB, Payment Statistics, November 2007; * BIS, CPSS, Statistics on Payment and
Settlement Systems in Selected Countries, March 2008

Switzerland (1.7). Table 2.2 shows the distribution of non-cash payments

across the various instruments in the EU and selected economies.
We can observe the difference between countries where electronic
instruments dominate as opposed to cheque countries such as the US
(32.6 per cent), France (25.6), Italy (12.6) and the UK (12.3). Credit trans-
fers are highly used in Germany, Benelux and Nordic countries. They are
42 The Structure and Economics of Payment Systems

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0%
96 06 96 06 96 05 96 06 96 06 96 06 06
France Germany Japan Sweden Switzerland UK US

Cards Cheques Direct debits Credit transfers

Figure 2.5 Trends in payment instruments in various countries

also heavily used in Eastern Europe, possibly as a heritage of the Soviet

economy which imposed cash for individuals and credit transfers for
enterprises. If a person in some countries wished to cash a credit trans-
fer, he/she had to go to the issuing branch. Direct debits are a sign of
maturity and a logical follow-on from a high usage of credit transfers.
We can observe their strong growth in countries having undergone a
change in political and economic regime such as Spain (44.7 per cent)
and the Czech Republic (34.8 per cent). The majority of Eastern European
and Baltic countries wisely chose to leapfrog cheques.
Figure 2.5 shows trends in selected countries over 10 years.
The decline of cheques is immediately visible and some countries
have either eliminated them (Sweden, The Netherlands) or reduced
their number to insignificance (Belgium, Germany, Japan, Switzerland).
Cards show the strongest growth, particularly debit cards as they replace
cheques. Sweden and Switzerland show a strong reduction of credit
transfers as cards show strongest growth whilst direct debits remain
stable.
3
Cross-Currency Payments and SWIFT

We have so far been talking about payments involving one currency, be

it US dollars, euro or Japanese yen. International trade and mobile indi-
viduals increasingly demand payments to be effected to settle debts in a
different currency than that in which the initiator holds his account, for
instance a Japanese manufacturer invoicing a US importer in yen. These
used to be called ‘international payments’ or ‘cross-border’ payments,
but since the advent of the euro which is now the legal currency in 15 of
the EU countries, it is more correct to distinguish between cross-currency
payments for our example above and cross-border payments when credi-
tor and debtor are located in different countries but the payment is in a
common currency – for instance a euro payment between euro accounts
in the Netherlands and Spain.
Generally speaking, ‘currencies do not travel’: the settlement of pay-
ments in a given currency takes place at the central bank which issues
it. A presence in the country of the currency is therefore required. Credit
cards which can be used outside their country of issue are the payment
instrument most used by individuals when travelling or ordering goods
from abroad (see ch. 2 sec. 6), but credit cards are a relatively new instru-
ment and not suitable for commerce or financial markets. Cheques can
be presented abroad, but the beneficiary would have to wait a long time
before his account is credited as the cheque has to be recovered: phys-
ically (before the advent of electronic imaging and transmission) sent
back across the oceans to the drawer’s bank to verify the signature and
availability of funds.

1 Correspondent banking
To satisfy the demand for international payments, banks developed the
system of correspondent banking, opening accounts in the local currency

43
44 The Structure and Economics of Payment Systems

EUR GBP
Belgian bank British bank

British bank’s Nostro
€13,550 Belgian bank’s Nostro  £10,000

 €13,550
charges
£10,000

Meat
invoice

£10,000  €13,550
Belgian supermarket Scottish beef farmer

Figure 3.1 Correspondent banking – direct relationship

with each other, called nostro accounts (from the Italian ‘nostro’ meaning
‘our’).
Figure 3.1 illustrates the case of a Belgian supermarket chain order-
ing Angus beef from a Scottish cattle farmer, who invoices it for GBP
10,000. The debtor’s Belgian bank will convert the £10,000 into euros at
the prevailing exchange rate (say a13,550), credit his British correspon-
dent’s nostro account in euros and debit the supermarket’s account by
the same amount plus charges. He will instruct the British bank to credit
the farmer’s account with GBP 10,000, which the British bank will first
debit from the Belgian bank’s nostro account.
This case is relatively simple in the sense that the Belgian bank’s cor-
respondent, also referred to as its sterling clearer, happens to also be the
Scottish farmer’s bank. This is not always the case as banks normally only
entertain correspondent relationships with only two or three correspon-
dents per currency. We can well imagine that the farmer will hold his
account with a bank in Edinburgh which will not be the Belgian bank’s
clearer in the City.
In this case (see Figure 3.2), the Belgian bank will:

• Instruct the Scottish bank to credit the farmer, indicating that cover
will come from his named sterling correspondent;
• Instruct his correspondent to credit the beneficiary’s bank with
the £10,000, which will be effected through the UK clearing and
settlement system.
 Cross-Currency Payments and SWIFT 45

Correspondent
Belgian bank
bank

Belgian bank’s Nostro  £10,000

British bank’s Nostro
€13,550 UK domestic

payment systems
 €13,550
charges
Beneficiary’s
bank

£10,000

Meat
invoice

GBP 10,000 (€13,550)

Belgian supermarket Scottish beef farmer

Figure 3.2 Correspondent banking – no direct relationship

An even more complex case arises when the payment must be effected
in a third currency other than the debtor’s and the creditor’s. Like most
commodities, oil is traded in US dollars, and Figure 3.3 overleaf illustrates
the case of a French chain of petrol stations ordering oil from Saudi
Arabia.
Payments in US dollars need to be settled at the Federal Reserve Bank
in the US, so the ordering customer’s French bank will credit the dollars
via its US dollar correspondent, the US domestic clearing system and
the Saudi bank’s US correspondent. Printed and on-line directories indi-
cate the names of each bank’s correspondents or clearers in the major
currencies.
As for any account, correspondents issue statements for the nostri
accounts they hold. The ordering banks will reconcile these statements
to ensure that all payments they instruct have been correctly effected and
that no payments have been debited by error. The same applies for all
intermediary banks along the chain. If we consider that a major clearer
will today transact between 50 and 100,000 international payments daily
it is obvious that this reconciliation cannot be effected manually.
These procedures reflect a credit transfer. Direct debits are more diffi-
cult to implement cross-border on account of the different legal regimes,
schemes and consumer protection rules prevailing in each country. We
will see in chapter 6, (sec. 6.2) how cross-border euro direct debits will
be available from 2009 within the SEPA framework.
46 The Structure and Economics of Payment Systems

USD
clearing and
settlement
US

French bank’s USD clearer Saudi bank’s USD clearer

Saudi
France
Arabia

French bank Saudi bank

French petrol USD 1,000,000 Saudi oil

company producer

Figure 3.3 Correspondent banks – payments for goods quoted in a third currency,
e.g. commodities

2 The SWIFT network

Up until the 1970s, international (as all cross-currency payments then
were) payments were transacted between banks by telegram and telex
(hence the name wires sometimes given in the US to international pay-
ments), secured by a system of manually calculated sequential test keys.
While standards had been developed by each country for domestic pay-
ments, international payments were the last to be automated on account
of the differences in language, formulation and practices.
To overcome this problem, Citibank in the early 1970s (then The First
National City Bank of New York) started to develop the Marti (Machine
Readable Telex Input) system, based on a standardized structure for telex
messages to effect payments. Out of fear of a major competitive threat, a
group of European banks launched the MSP (Message Switching Project)
with the view to develop a standardized automated electronic commu-
nication system for international payments. To achieve critical mass,
the project was rapidly extended to the major 69 banks across Western
Europe and North America (including Citibank which abandoned Marti).
The Society for Worldwide Interbank Financial Telecommunication was
incorporated as a non-profit bank-owned cooperative in Belgium in 1973
 Cross-Currency Payments and SWIFT 47

and the SWIFT1 network cut over to live operation in May 1977, after
arduous negotiations with the European postal authorities who then held
a monopoly on telecommunications and could foresee the disappearance
of a lucrative market.
Although the implementation of a secure international private net-
work was in those days a technical prowess, SWIFT’s main achievement
lies in the development of internationally accepted standards for finan-
cial transactions. Messaging standards were first developed for customer
payments, interbank payments and nostro account statements. Stan-
dardized bank addresses were also published known as the BIC codes
(Bank Identifier Code), composed of a unique 4 character bank code,
a 2 character country code, a 2 character location code and an optional
3 character branch code: BBBB CC LL (bbb). The SWIFT system would
validate conformity with the standards and reject any message submit-
ted with errors, ensuring therefore that the sender need only capture
the data once and that messages delivered could be processed automati-
cally by the recipient. With strict guidelines for referencing, this enabled
full end-to-end automation of the processes and flows described in the
previous section, including automated nostro reconciliation. Messages
are encrypted and sophisticated authentication algorithms guarantee
origination, integrity and non-repudiation. Subject to compliance with
defined operating rules and procedures SWIFT also accepts some liabil-
ity for the messages it processes. From early on SWIFT offered a range
of interfacing terminals and software to the network, originally to pro-
vide connectivity to its users confronted with the lack of solutions from
the marketplace. The network subsequently expanded geographically to
include all major financial centres and, at the end of 2007, connected
nearly 8,300 financial institutions in 208 countries.
SWIFT extended in parallel the standardization of messages to cover
virtually all financial transactions: payments and cash management,
treasury (foreign exchange, money markets) and derivatives, securities
and trade (collections and documentary credits). The major break-
through occurred in 1987 after the banks who owned SWIFT accepted,
after protracted hesitations as they feared dilution of their custody busi-
ness, that securities broker dealers and fund managers could connect to
the network to enable the automation of cross-border securities clearing
and settlement. In 2006 the SWIFT membership also approved the par-
ticipation of major corporations to communicate with the bank which
sponsored their participation. Figure 3.4 shows the evolution of the dis-
tribution of SWIFT traffic between markets from 1996 until the first two
months of 2008.
48 The Structure and Economics of Payment Systems

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0%
Average 1996 Jan–Feb 2008
Trade Treasury Securities Payments

Figure 3.4 Evolution of SWIFT’s traffic distribution

Sources: SWIFT 1996 annual report; www.suift.com

We can see how securities messages have grown to almost equal pay-
ments in volume and have become the strongest contributor to growth.
In 1986 SWIFT entered value-added services by developing and oper-
ating, on an outsourced basis, a netting system for the Ecu (the basket
currency precursor to the euro), which subsequently evolved as the
EBA’s EURO1 netting system (see ch. 6 sec. 5). Several others followed
such as an automated matching service for confirmation of foreign
exchange trades (ACCORD) and solutions for cash reporting, corporate
actions, exceptions and investigations, etc. In addition, SWIFT maintains
directories of the BIC codes, correspondents and settlement agents.
SWIFT also now transmits files without format validation; the current
SWIFTNet platform is IP-based and messaging standards are gradually
migrating towards the XML syntax. SWIFT maintains its leading role in
standards development.
SWIFT itself is not a payment system, but serves as transport net-
work for virtually all major payment and securities market infrastructures
and is arguably today the leading global provider of financial messag-
ing and processing services. During 2007, SWIFT processed an average
daily volume approaching 14 million messages, a five-fold increase from
 Cross-Currency Payments and SWIFT 49

the 2.7 million messages in 1996. Aided by strong volume growth,

SWIFT has pursued a systematic policy of tariff reduction to fend off
competition from other networking service providers and the internet,
while constantly striving towards maximizing operational resilience.
Finally, SWIFT organizes the annual SIBOS conference and exhibition,
the major venue for debate, networking, self-congratulation and lavish
entertainment for the payments and securities processing fraternity.

3 The correspondent banking and clearing business

Correspondent banking and clearing has become increasingly concen-
trated and competitive since the introduction of the euro, as nostro
accounts are no longer required in 15 legacy currencies. The trend to
reduce correspondents accelerated and it is rare that banks will main-
tain relationships with more than two to three clearers per currency.
Global banks aggressively market their wide local presence and access
to payment systems in several currencies. Nostalgia abounds on ‘the
good old days of correspondent banking’ based on balances, reciprocity
(each bank giving the other approximately equal volume of business
in their respective currencies), exotic trips and personal relationships
cultivated over liquid lunches and days on the golf course. Remuner-
ation is nowadays negotiated around interest on balances, credit lines
and above all transaction fees based on STP: the ability for a clearer to
receive the payment over SWIFT and process it straight through to final
reconciliation without any human intervention. Exceptions and inves-
tigations caused by non-compliance with standards or agreed practices
are penalized heavily, up to ten times the fee for an STP payment.

4 Remittances
This is the term given to low-value payments transferred mainly by
migrant workers to their families back home for amounts ranging
between $100 and $500. Volumes and values are difficult to estimate as
a large number are effected by ‘informal’ channels such as hawala which
are estimated to add around 50 per cent to some 175 million officially
recorded remittances which exceeded $232 billion in 2005; countries
contributing the largest value are the US and Saudi Arabia.2 In 20 of the
largest recipient countries, these remittances amount for over 10 per cent
of GDP.3
Cross-currency transfer services from banks, based on the correspon-
dent banking model, are expensive and also assume that payers and
50 The Structure and Economics of Payment Systems

beneficiaries, often in remote areas, hold bank accounts which is not

always the case.
These services were originally offered by non-banks: Western Union –
leveraging their telecommunication infrastructure – and MoneyGram.
Funds can be sent and received from accredited service points which
accept payment and disburse the funds in cash during longer opening
hours than bank branches. Fees are low, but the total cost should be
considered including the exchange rate and any fees from the disburse-
ment agent. Substantial revenues are derived from ‘emergency’ services:
transferring money urgently to travelling children stranded penniless
following the theft of their wallet on Machu Pichu. Commercial banks
have woken up to this opportunity and several now offer low-priced
services over ‘corridors’ between countries taking in large numbers of
migrant workers from specific areas for historical or cultural reasons: US
to Mexico, France to Africa, Gulf to Middle East and Asia, Spain and
Portugal to South America, UK towards India, Pakistan and Africa, etc.
The hawala system is typical of how the informal channels work.
A person in the UK wishing to remit funds to Pakistan will approach
a ‘hawaladar’ indicating the amount he wishes the beneficiary to receive
in rupees. The hawaladar will calculate the equivalent to be paid in ster-
ling and add a commission. He will then contact a hawaladar close to
the beneficiary who will pay him/her. How the hawaladars settle between
them is unclear: several intermediaries might be involved and settlement
could await an offsetting transfer of funds in the reverse direction, for
instance paying for goods imported in the country of the first benefi-
ciary. The system is based on trust and a code of conduct and seldom
fails. It is also cash based and therefore offers anonymity.
This raises of course the issue of compliance with Anti Money Launder-
ing (AML) and Counter Terrorist Financing (CTF) legislation, customer
identity and transfers to black-listed individuals. The major non bank-
ing systems listed above certainly comply with these measures and are
licensed. In the UK, HM Revenue and Customs maintains a register of
these operators and conducts regular site visits.4 The BIS published the
following principles in 2007 shown in Box 3.1:

Box 3.1 The General Principles and related roles5

The General Principles are aimed at the public policy objectives of
achieving safe and efficient international remittance services. To this
end, the markets for the services should be contestable, transparent,
accessible and sound.
 Cross-Currency Payments and SWIFT 51

Transparency and consumer protection

General Principle 1. The market for remittance services should be
transparent and have adequate consumer protection.
Payment system infrastructure
General Principle 2. Improvements to payment system infrastructure
that have the potential to increase the efficiency of remittance ser-
vices should be encouraged.
Legal and regulatory environment
General Principle 3. Remittance services should be supported by a
sound, predictable, non-discriminatory and proportionate legal and
regulatory framework in relevant jurisdictions.
Market structure and competition
General Principle 4. Competitive market conditions, including appro-
priate access to domestic payment infrastructures, should be fostered
in the remittance industry.
Governance and risk management
General Principle 5. Remittance services should be supported by
appropriate governance and risk management practices.
Roles of remittance service providers and public authorities
A. Role of remittance service providers. Remittance service providers
should participate actively in the implementation of the General
Principles.
B. Role of public authorities. Public authorities should evaluate what
action to take to achieve the public policy objectives through
implementation of the General Principles.

Fundamental to success is a distribution network covering remote

areas. This is probably the major obstacle facing banks. Even the major
global banks generally only maintain branches in major cities. Agree-
ments with local banks operating a large branch network are therefore
essential, but few will match the capillarity of the official money transfer
systems or even less that of the informal networks.
It should be noted that the demand for remittances also exists domes-
tically in countries where workers in the major cities wish to send money
to their unbanked families in less developed rural areas. An innovative
system was developed by a Turkish bank based on mobile telephones.
After the amount has been debited from the sender’s account, the bene-
ficiary will receive an SMS message indicating the amount and a one-off
PIN with which the cash can be drawn at any one of the bank’s ATMs.
4
Risks in Payment Systems, Oversight
and Security

Intermediating and evaluating risk is the essence of banking and the

basis for remunerating financial services. The Florentine banker Cosimo
de Medici (1389–1464) would instruct his foreign representatives not to
lend to princes who never repaid, but to limit themselves to financing
reputable and competent merchants, relying upon reputation and the
judgement of his staff as no rating agencies were active then.
The second half of 2007 saw the subprime mortgage crisis erupt on the
front pages of the financial and popular press as top-10 institutions wrote
down billions of dollars worth of debt, forcing them to seek assistance
from Gulf and Asian sovereign wealth funds to restore their depleted
capital ratios. In spite of the much publicized liquidity shortage, all pay-
ment systems had settled normally in the major financial centres up until
the date of writing these lines.
On 11 September 2001 the World Trade Centre and the Wall Street
financial district in New York were devastated and thousands lost their
lives when terrorists flew two planes into the Twin Towers. The cash and
securities settlement systems, some of which were located at the foot
of the Twin Towers, were rapidly restored for some users from back-up
sites. The Federal Reserve, the Bank of England, the European Central
Bank and most central banks announced that they would meet all liq-
uidity demands. The ECB injected liquidity the next day through ‘quick
tenders’. On the thirteenth, the Fed and the ECB published details of a
50 billion dollar swap agreement reached the previous day, to counter
the risk of a dollar liquidity shortage. Overnight overdrafts granted by
the Fed soared from a daily average of nine million dollars throughout
August to four billion on 12 September. Payment systems extended their
opening hours. On 12 September the ECB put 69 billion euros on the
market and 40 the following day. Half of the swap between the Fed and

52
 Risks, Oversight and Security 53

the ECB was used before refund. The next day everything returned to
normal and payment operations were functioning. A banking and stock
market crash had been avoided.
On 7 July 2005 terrorists again struck in London, albeit on a smaller
scale, but no disruption was reported to the payment systems: the lessons
had been learned.
These examples demonstrate both the vulnerability of payment sys-
tems to financial and external events, as well as the resilience which has
been built up over the years in the light of experience.

1 Analysis of risks in payment systems

The exposure to risk and its management within payment systems are
subject to several factors:

The amounts and volumes involved: we have seen how RTGS systems
account generally for over 90% of the monetary value, settling in a few
days the equivalent of their country’s annual GDP, while ACHs are con-
stantly seeking to increase volumes to achieve economies of scale. In
other words, the more successful a payment system becomes, the greater
the concentration of risk.

Technical and financial innovation: payment and securities clearing and

settlement systems must maintain technological leadership to handle
increasing volumes and new instruments while reducing execution times
and fees as well as complying with demanding risk limitation algorithms.
On the other hand, financial institutions are constantly developing more
complex products, assisted by (supposedly!) increasingly sophisticated
risk models. A key measure is the value at risk (VAR), an estimate of the
maximum losses a product or portfolio might incur over a given period
with a specified confidence level (for example: this asset has a 95% proba-
bility of not losing more than 5% of its value over the next 30 days). The
subprime crisis mentioned above illustrated painfully how the repack-
aging and securitization of low quality debt gave birth to instruments
which, instead of reducing and/or spreading risk, gave rise to greater
financial losses over a longer period of time than were thought possible.

Diversity of expertise and skills: payments stretch across the entire bank-
ing organization, functionally and geographically (retail and wholesale
banking, securities and FX trading, etc. across all branches and sub-
sidiaries) calling for cooperation between a multitude of disciplines: IT,
54 The Structure and Economics of Payment Systems

telecommunications, operations, legal, audit, marketing and finally risk

management. Few specialists grasp all issues and their interrelationship,
and interfaces between the various systems are particularly sensitive.

Transparency: payment systems must be auditable and individual pay-

ments must be traceable end-to-end in case of disputes or investi-
gations. Payment services rely on cooperative systems (for instance
ACHs) as well as value-added services, processes and technology through
which providers seek to gain competitive advantage. This ‘cooperation
vs. competition’ dichotomy does not always lead to full end-to-end
transparency.

Achievingtherightbalancebetweencostsandsecurity: maximum security

and minimum risk are achieved at a cost in terms of investments and
operating costs, as well as the opportunity cost of securities immobilized
for collateral. Firm criteria are often developed and imposed by central
banks and supervisory authorities after an incident as we will see in the
following paragraphs.
Effecting payments and managing the related risks are among the core
competencies of commercial banks. Central banks, on the other hand,
are at the heart of payment systems as final settlement agent, liquid-
ity providers and, ultimately, lenders of last resort. They have therefore
naturally assumed responsibility for the oversight of payment systems
and cooperate on this topic within the Committee on Payment and
Settlement Systems (CPSS) of the BIS.
The CPSS has identified that ‘A range of risks can arise in payment
systems, taking the following forms in that context:

• credit risk: the risk that a party within the system will be unable to
fully meet its financial obligations within the system either when due
or at any time in the future;
• liquidity risk: the risk that a party within the system will have insuf-
ficient funds to meet financial obligations within the system as and
when expected, although it might be able to do so at some time in the
future;
• legal risk: the risk that a poor legal framework or legal uncertainties
will cause or exacerbate credit or liquidity risks;
• operational risk: the risk that operational factors such as technical mal-
functions or operational mistakes will cause or exacerbate credit or
liquidity risks; and
• systemic risk: the risk that the inability of one of the participants to
meet its obligations, or a disruption in the system itself, could result
 Risks, Oversight and Security 55

in the inability of other system participants or of financial institutions

in other parts of the financial system to meet their obligations as they
become due. Such a failure could cause widespread liquidity or credit
problems and, as a result, threaten the stability of the system or of
financial markets.’1

2 Risk management
2.1 Financial risks
Financial risk arises when there is a delay between acceptance of the
payment by the system and final settlement. This can be caused by a
temporary failure (liquidity risk) or, more seriously, definite inability
by a participant to meet its obligations, for instance in case of default,
suspension or bankruptcy (credit risk).
These intraday settlement risks are significant in Deferred Net Settle-
ment Systems (DNS) when final settlement occurs at designated times,
mostly at the end of the operating day. In addition to the imposition of
minimum capital and credit rating requirements as part of the member-
ship criteria, this has led to the introduction of limits on the maximum
level of risk that a participant can create. These can be bilateral limits
imposed by each participant on the other direct members, an overall
multilateral net debit limit imposed by the system (the maximum dif-
ference at any point in time between the sum of the values of payments
received minus the payments sent by a participant), or a combination
of both.
Measures must therefore be taken to ensure that a DNS system can
settle daily even in case of default of one or more of its members. The
ultimate remedy would be unwinding, which implies removing some – up
to the point where the defaulting participant’s multilateral net debit limit
can be met – or all payments entered by the defaulting participant since
the last settlement and recalculating the net positions. This is, however, a
measure of last resort used extremely rarely as removal of these payments
might leave the surviving members with insufficient funds to meet their
own obligations. The regulators have therefore imposed that participants
bear the responsibility for covering eventual losses; two arrangements
prevail, either individually or in combination:

• Defaulter pays: the defaulting participant must secure collateral to

cover at least the multilateral net debit limit imposed on him;
• Survivors pay: a loss sharing agreement which stipulates how the sur-
viving participants will share the loss, generally in proportion to the
56 The Structure and Economics of Payment Systems

bilateral limits they have placed on the defaulter, thereby reflecting

the risk they were assuming towards it. These commitments will also
be secured by collateral.

In RTGS systems credit risk is considered to have been eliminated as final

settlement takes place gross for each payment in real-time, particularly
if Y-copy routing is adopted (see ch. 1, sec. 3.4) when the receiving bank
is only notified after settlement, eliminating the possibility of it credit-
ing the beneficiary’s account prematurely. Liquidity risk does however
remain, as payments will be queued if insufficient funds are available at
the settlement account and ultimately returned if not settled by day end.
This risk is however reduced by the provision of intraday liquidity from
the money market or collateralized credit lines from the central bank, as
well as the implementation of increasingly sophisticated queuing algo-
rithms and liquidity saving features (see ch. 1 sec. 3.3). These modern
systems, which combine real-time gross settlement of urgent payments
while lower priority payments are deferred awaiting netting or offsetting,
are referred to as hybrid systems. Early RTGS systems suffered liquidity
problems as participants tended to delay entering payments (particularly
high value ones), waiting for received payments to supply their liquid-
ity: obviously nothing much happened and the system gridlocked as
everybody waited for everybody else! Pricing incentives were introduced,
whereby late payments were subject to a much higher tariff, but most
RTGS systems now impose minimum percentages of the total daily value
that must be entered by specific times throughout the operating day.
All systems rely therefore on a collateral pool owned by its members
to guarantee obligations and/or credit lines. The assets constituting
this pool must be extremely liquid, such as government securities (for
instance Gilts, T-bills, and Bons du Trésor), cash or other instruments
used by central banks for their open market operations. They can be
immobilized or, to reduce opportunity costs, pledged or made avail-
able through repurchase agreements (repos) covering the operating day.
A haircut (generally around 10 per cent) is normally imposed to guard
against fluctuations in market value. Direct computer links are also
implemented between RTGS payment systems and the securities depos-
itories to ensure rapid availability of the collateral should a participant
wish to increase liquidity.
All payment systems must therefore include comprehensive informa-
tion and control facilities enabling participants, the operator and/or the
central bank to monitor balances, limits and collateral valuations in real
time and take the necessary action.
 Risks, Oversight and Security 57

2.2 Legal risks

Legal uncertainty can arise out of:

• the contractual agreements between participants, system owner, sys-

tem operator(s) and eventual subcontractors;
• the legal framework establishing finality of settlement;
• insolvency and bankruptcy legislation; and
• the ability to use collateral posted.

These issues should not be underestimated. In particular for systems

operating crossborder, legal fees to ensure enforceability in all rele-
vant jurisdictions have amounted to a substantial proportion of the
development investment.
A payment system relies upon a series of contractual agreements:

• The statutes, defining the incorporation, shareholding, jurisdiction,

objectives, governance and eligibility criteria. Payment systems can be
private (normally in majority owned by its participants as sharehold-
ers), public (owned by the central bank) or mixed when shareholding
includes the central bank.
• Management and service level agreements if the operation is subcon-
tracted to one or more service providers. Even if the owner(s)
operate the system themselves, subcontractors are always involved
for telecommunications, energy and cooling, IT services, mainte-
nance, etc.
• The operating manual defining the standards, procedures and processes
which must be followed by the operator(s) and the participants, to
include performance criteria, security and business continuity.
• The settlement convention between the clearing system and the set-
tlement entity, defining in particular the cut-off times and the rules
applying to the settlement accounts, credit lines and collateral.
• Agreements between participants regarding netting, bilateral credit lines,
loss recovery procedures and provision of liquidity, in particular even-
tual automated lending/borrowing between long and short members
to facilitate settlement.

We will see the importance of service level agreements and performance

criteria in the section 2.3 covering operational risk.
The legislation surrounding settlement finality must cover payments
settled gross and payments which are netted in a DNS system. The
implementation of RTGS systems required the abolition of the previously
58 The Structure and Economics of Payment Systems

widespread ‘zero hour rule’ which decreed that in a case of bankruptcy,

all transactions by the participant from the start of the day (hour zero)
were void, thereby reinstating credit risk after what was considered to
be final settlement at the central bank. In addition, legal uncertainty
existed in the case of payments accepted by a DNS and netted, which
might have to be unwound.
Finally, a reliable legal framework, known as the laws of secured interests,
must surround collateral to ensure that assets pledged, or subject to a
repurchase agreement, can be effectively and rapidly realized in case of
bankruptcy.
It should however be stated that sound legal frameworks are now in
place for all major currencies, for example the EU Settlement Finality
Directive2 and the Article 4A of the US Uniform Commercial Code.

2.3 Operational risk

Managing operational risk is intimately linked to security and business
continuity. Reducing risk in these areas demands the implementation
of coordinated and harmonized end-to-end procedures as, to quote the
cliché, ‘a chain is only as strong as its weakest link’. Security and avail-
ability must therefore extend to all components (staff, procedures, IT
hardware and software, telecommunications, power and cooling supply,
etc.) not only at the central system, but also at subcontractors and each
participant: most systems insist on technical and process qualification
tests before a participant can operate live. This section will concen-
trate on operational resilience and business continuity; security will be
discussed in section 4 of this chapter.
Design specifications should ensure no ‘single point of failure’: several
institutions discovered, at their cost after a major outage, that the same
power grid served main and back-up sites or that telecommunication
links, in spite of entering a building at different points, rejoined in the
same cable at the end of the street! Service level agreements usually spec-
ify execution time under peak volume requirements, as well as response
times to information requests such as authorizations for card payments
or calls for collateral. Regarding business continuity, criteria include the
minimum availability during operating hours (normally over 99.9%),
allowable down time for maintenance and system updates, as well as the
maximum delay to resume service in case of failure.
The 9/11 terrorist attack highlighted dramatically the need to cope
with multiple failures and showed that it was unrealistic to rely upon
personnel moving to a back-up operating site following a major regional
 Risks, Oversight and Security 59

disaster. It also prompted an industry wide reappraisal of business conti-

nuity procedures not only for payments, but taking a holistic view across
the entire financial system including exchanges, securities settlement
systems and other critical market infrastructures, covering operational
issues and provision of liquidity to avert a financial crash. The US reg-
ulators (Federal Reserve System, SEC and OCC) and the ECB published
stringent new guidelines to guarantee business continuity. Although all
critical infrastructures operate primary and secondary sites, both stress
the requirement for ‘out of region’ fully staffed back-up centres for the
central system and critical participants (institutions which contribute
substantial volumes or value). The ECB guidelines apply to Systemi-
cally Important Payment Systems (SIPS) defined as ‘A payment system is
systemically important if a disruption within that system could trigger
or transmit further disruptions amongst participants or systemic dis-
ruptions in the financial area more widely.’3 They recommend ‘that
SIPS should aim to recover and resume critical functions or services
(including critical services outsourced to third-party providers) no later
than two hours after the occurrence of a disruption.’ Regarding critical
participants, the ECB also recommends that operating a secondary site
should be part of the requirements to join the system and states: ‘At a
minimum, relevant participants should be able to close one business day
and reopen the following day on the secondary site’.
In addition to staffing issues, the operation of distant sites poses tech-
nical problems with respect to data logging and integrity as ‘synchronous
disk mirroring’, to ensure that identical data is available in ‘hot standby’
mode at both sites, has physical limitations with respect to distance. Sec-
ondary sites should also be sized to absorb higher volumes as experience
shows that flows exceed the daily average following a serious disruption.
Regulators also demand the implementation of contingency arrange-
ments to ensure that, at a minimum, critical payments (for instance
settlement of other payment systems or related to monetary policy)
can be handled even in case of catastrophic unavailability of primary
and back-up sites: ‘The provision of a ‘minimum service level of criti-
cal functions’ could be achieved, for example, through a combination
of predetermined business authentication procedures based on man-
ual, paper-based processing, authenticated facsimile messages, or a basic
PC-based system using physical media for data transfer.’4
Finally, the regulators stress the need to thoroughly document and
regularly test and rehearse contingency plans under a variety of scenar-
ios. Industry-wide contingency tests are conducted by the supervisory
authorities at random intervals.
60 The Structure and Economics of Payment Systems

2.4 Systemic risk

Systemic risk is the central bankers’ terminology for the ‘domino effect’,
the risk that failure of one or more participants to settle might spread
to other institutions and degenerate into a major financial crisis. The
CPSS therefore published in 2001 the Core Principles for Systemically
Important Payments Systems, as defined in the previous paragraph. The
core principles are detailed in Box 4.1.

Box 4.1 Core principles for systemically important

payment systems5
I. The system should have a well-founded legal basis under all
relevant jurisdictions.
II. The system’s rules and procedures should enable participants to
have a clear understanding of the system’s impact on each of
the financial risks they incur through participation in it.
III. The system should have clearly defined procedures for the man-
agement of credit risks and liquidity risks, which specify the
responsibilities of the system operator and the participants and
which provide appropriate incentives to manage and contain
those risks.
IV. The system should provide prompt final settlement on the day
of value, preferably during the day and at minimum at the end
of the day.∗
V. A system in which multilateral netting takes place should, at
a minimum, be capable of ensuring the timely completion of
daily settlements in the event of an inability to settle by the
participant with the largest single settlement obligation.∗
VI. Assets used for settlement should preferably be a claim on the
central bank; where other assets are used, they should carry
little or no credit risk and little or no liquidity risk.
VII. The system should ensure a high degree of security and oper-
ational reliability and should have contingency arrangements
for timely completion of daily processing.
VIII. The system should provide a means of making payments which
is practical for its users and efficient for the economy.
IX. The system should have objective and publicly disclosed criteria
for participation, which permit fair and open access.
 Risks, Oversight and Security 61

X. The system’s governance arrangements should be effective,

accountable and transparent.
∗
Systems should seek to exceed the minima in these two core
principles.

The key issues for Principles I – VII have been discussed in the previous
section. The asterisk∗ for Principles IV and V, indicating that ‘Systems
should seek to exceed the minima in these two core principles’, refer to
the recommendations of the CPSS that countries should at least imple-
ment an RTGS system for high-value payments and that multilateral
netting payment systems should be able to withstand the inability to
settle by more than one participant.
Principle VIII, which might initially appear as ‘motherhood and apple
pie’, is a reminder that cost efficiency should include all costs: fees, liq-
uidity, joining investment and operating costs. Operators should also
consider that institutions of different size might wish, or be compelled,
to join and that economic interfacing solutions must be made available
to low-volume participants.
Principle IX clarifies that access criteria must be transparent and
objective, for instance minimum capital requirements or market share.
The definition of ‘systemically important’ might appear subjective; the
ECB6 considers that all RTGS systems, high value systems and retail pay-
ment systems for which there is no national alternative, for instance
ACHs handling credit transfers and direct debits irrespective of the value,
must be defined as SIPS.

2.5 Settlement risk in financial markets

All trading operations include two steps: the actual trading when the
price is struck, and the settlement which consists of delivering the cur-
rencies or securities brought or sold. It is important that this takes place
simultaneously to avoid a settlement fail, for instance the cash for secu-
rities being paid but the securities not being delivered. All securities
are today dematerialized: shares are no longer physically printed and
handed over, but ownership exists as accounting records in a Central-
ized Securities Depository (CSD) which are modified when the trade
is settled. To eliminate settlement risk, regulators have mandated the
implementation of Delivery versus Payment (DVP). This is normally
achieved by interlinking the CSD with the relevant RTGS system which
62 The Structure and Economics of Payment Systems

sends confirmation that the payment has been irrevocably settled fol-
lowing which the change of ownership will be registered. This concept
will be further enlarged in Chapter 11, while Chapter 10 will look at the
elimination of settlement risk in foreign exchange trading.

3 Regulatory oversight
Until the 1990s, central banks were solely responsible for supervising
commercial banks, while separate bodies, such as the SEC in the US,
supervised securities trading. The emergence of global and universal
banks combining retail, commercial and particularly investment bank-
ing and securities services prompted the devolution of supervision to
independent bodies such as the Financial Services Authority (FSA) in
the UK, the Bundesaufsichtsamt für Finanziellen Instituten (BaFin) in
Germany. Arguing that they issue money which is the fundamental
means of exchange, that it is essential to maintain stability of the
financial system and that safe and efficient systems are vital for the imple-
mentation of monetary policy, the central banks maintained oversight
of the payment and settlement systems: ‘Oversight of payment systems
is a public policy activity focused on the efficiency and safety of systems,
as opposed to the efficiency and safety of individual participants in such
systems. . . In many countries, the central bank’s oversight role is consid-
ered an integral element of its function in ensuring financial stability’,7
clearly separating oversight over the payment system from supervision
of its participants.
Oversight is defined as: ‘Oversight of payment and settlement systems
is a central bank function whereby the objectives of safety and efficiency
are promoted by monitoring existing and planned systems, assessing
them against these objectives and, where necessary, inducing change’.8
As mentioned previously, central banks cooperate on these issues and
define guidelines within the CPSS at the BIS which are defined in Box 4.2.
Principle E is of particular importance as payment and securities
systems increasingly offer their services internationally and the above
publication further specifies that ‘there should be a presumption that the
central bank where the system is located will have this primary respon-
sibility’. Under this principle, oversight of SWIFT for example rests with
the National Bank of Belgium where SWIFT is headquartered.
Banks also cooperate in national bodies to define payment services and
monitor performance of the various systems, such as APCA (Australian
Payments and Clearing Association) in Australia and APACS (Association
for Payment and Clearing Services) in the UK. We will see in Chapter 6
 Risks, Oversight and Security 63

Box 4.2 General Oversight Principles9

A. Transparency: Central banks should set out publicly their over-
sight policies, including the policy requirements or standards for
systems and the criteria for determining which systems these
apply to.
B. International standards: Central banks should adopt, where rel-
evant, internationally recognized standards for payment and
settlement systems.
C. Effective powers and capacity: Central banks should have the
powers and capacity to carry out their oversight responsibilities
effectively.
D. Consistency: Oversight standards should be applied consistently
to comparable payment and settlement systems, including systems
operated by the central bank.
E. Cooperation with other authorities: Central banks, in promot-
ing the safety and efficiency of payment and settlement systems,
should cooperate with other relevant central banks and authorities.

(sec. 1) how banks also established the European Payments Council (EPC)
to address the Single Euro Payments Area (SEPA). These organizations
are often blamed for preserving the interests of the banks and ignoring
the demands from their customers. A novel approach was taken in the
UK in 2007 through the creation of the Payments Council, whose Board
includes external directors, alongside the Bank of England as an observer
(see ch. 6 sec. 8).

4 Fraud prevention, security and anti money laundering

As payment services develop, security measures must take account of the
wider range of payment means and channels that emerge, bringing with
them new opportunities but also introducing new risks.
A comprehensive study of the security procedures and technology
across all instruments and channels, starting with measures to prevent
counterfeiting of banknotes, would be beyond the scope of this book. We
will attempt however to provide an overview of the principal risks and
fraud prevention measures surrounding electronic payments and cards.
 64 The Structure and Economics of Payment Systems

Security and fraud prevention centre principally around four issues:

• Authentication: is the originator of the payment or the payment instruc-

tion the account owner or a person authorized to effect payments from
that account, credit card or payment instrument? Is the beneficiary or
receiver of the payment or information the intended recipient? At both
ends of the communication link we need assurance that the ‘real’ and
‘asserted’ identities are correct.
• Confidentiality: has any outside third party gained unauthorized access
to the payment information, or to any data which would allow it to sub-
sequently fraudulently access the account or modify any instruction?
• Integrity: has any unauthoriszed third party fraudulently changed
any of the information: amount, beneficiary, account holding bank,
account number(s), etc.?
• Non-repudiation: could the originator, or any authorized party, deny
having taken any action he/she has actually undertaken?

Major concerns lie with identity theft, or phishing, whereby fraudsters

attempt to acquire passwords, usernames, dates of birth, PINs, etc. by
posing as a trustworthy counterparty. Methods range from the sophis-
ticated, whereby criminals insert a dummy banking or on-line retailer
website to capture payment information, to simplistic requests, by
e-mail or SMS, purporting to come from trusted entities such as banks or
government agencies requesting details such as passwords or PINs. Other
scams have been based on capturing card details and PINs by installing
fraudulent readers and/or cameras at ATMs. Most of us will have received
e-mails from Nigerians suffering from a terminal illness requesting our
account details to transfer their wealth to someone they can trust to use
it for the greater benefit of mankind! These simplistic scams, including
miraculous lottery wins, often deliberately target vulnerable segments of
the population such as the elderly. Most banks state on their websites,
or in the opening messages from call centres, that they will never ask for
information such as card PINs.
In the retail space growing volumes and the proliferation of new inter-
net payment methods linked to e-mail or pre-payment schemes, some
of which shift the security issues away from the financial institutions to
new players, can alter the underlying security for the user.
Although banks enhance their security continuously, organized crime
is constantly adapting and developing new ways to collect and use iden-
tity information, bank account and credit card details. This can include
collusion with staff within the financial institution or it’s customers,
 Risks, Oversight and Security 65

hacking of databases, recuperating confidential information from waste

bins or stealing data in transit e.g. CD-ROMs, tape back ups and lap-
tops. Banks and merchants need to constantly review their technical and
organizational security procedures to maintain and enhance the level of
security they achieve.
In the business to business space, payments are becoming more and
more embedded in the exchange of data that encompasses the complete
trade chain; this has brought new routes through which payments are
initiated and therefore the need to create trust schemes that can handle
these.
In addition to these security and fraud prevention issues, an increas-
ing regulatory burden is imposed on payment service providers in terms
of customer identity verification or Know your Customer (KYC), Anti
Money Laundering (AML) and Counter Terrorism Finance (CTF). These
issues are overarching, irrespective of the payment instrument and/or
channel used.

4.1 Internet banking

The implementation of internet banking websites requires varying levels
of security depending on the value of the information and transactions
that can be undertaken. Many different methods of authentication are
available in the marketplace now, for example:

• username and password;

• partial PINs and passwords;
• one-time passwords based on using a ‘grid’ or ‘matrix’ of stored values
on a plastic card;
• one-time passwords generated by a time synchronized device (such as
‘key fob’ type devices;
• one-time passwords generated from a pseudo-random sequence
(including EMV-CAP cards and readers);
• a manual (disconnected) device in conjunction with an on-line
challenge/response protocol;
• an automatic device (for instance connected via a USB port) with a
similar challenge/response protocol; and
• a device which is capable of storing keys and performing asymmetric
cryptography, such as a smart-card or USB token, and which there-
fore can be used as client in a Public Key Infrastructure (PKI: see next
section).
66 The Structure and Economics of Payment Systems

4.2 Wholesale banking and trust schemes

Frequently, higher levels of security and fraud prevention measures
are implemented on account of the values and volumes transacted,
particularly for direct computer-to-computer links. The objective is to
combine authentication (at both ends), confidentiality, integrity and
non-repudiation. Much of this is achieved through Public Key Crypto-
graphy which is based on the use of a private key and a public key that
are inextricably linked. The private key can be used to sign a message
and the public key, available to all counterparties, can then be used
to check the signature. The mathematics ensure that the private key
cannot be used check the signature, or the public key to create the signa-
ture. These keys can also be used to encrypt data using either the private
or public key that can only be decrypted using the public or private
key respectively. Through the use of various protocols this technology
also provides non-repudiation. The Internet Banking Application itself
should also ensure that the user with the correct level of authority is initi-
ating the transaction and, where appropriate, enforce dual and multiple
signatures.
Throughout the development of payment services, individual pay-
ment messages and files of payments initiated by the bank customer
have always been sent to the Bank itself or the Automated Clearing
House (ACH) directly. In trading terms this has meant that the payment
message has been created, signed and sent to the Financial Institution
by the buyer/payer. Except in limited EDI message types, the pay-
ment has often been divorced from the underlying information or trade
cycle which requires extensive communication between the trading
enterprises.
As discussed previously, banks are seeking to develop services which
enhance their position in the trade cycle and compensate for loss of
revenues on the commoditized basic payment services. These required
the development of a new trust model to allow messages and eventually
payments initiated by one bank’s customers to be checked and trusted
by another bank and its customers. This has been achieved by linking
the development of individual Public Key Infrastructures (PKI) at banks
across the world to a single Trust scheme such as IdenTrust. IdenTrust
has developed a legal framework alongside a set of policies, procedures
and technical standards for authentication and to issue trusted identi-
ties worldwide. Reverting to the classical four-corner model a customer
who has been issued a Digital Certificate, confirming that their pub-
lic key has been issued to it by a third party whom all the recipients
 Risks, Oversight and Security 67

can trust such as an IdenTrust member bank, can conduct trusted

business with a trusted customer of another IdenTrust member bank.
Schemes such as IdenTrust provide a global framework for the provi-
sion of certificate authority services, enabling financial institutions to
extend their full range of services onto the internet and become trusted
third parties for e-commerce transactions. Through this they can develop
new lines of business as the internet becomes a preferred transaction
medium and their customers will have the ability to leverage a single
bilateral relationship with a financial institution for all e-commerce deal-
ings. At the same time this offers businesses and financial institutions
a way to add value by proactively managing the risk associated with
e-commerce.

4.3 Cards
With the exception of the US, chip card technology combined with a PIN
under the EMV (Europay, MasterCard, Visa) scheme is gaining worldwide
acceptance and fraud is significantly reduced in comparison with signa-
ture and magnetic stripe. In any case, signature or PIN verification can
only take place when the card holder is physically present in the retail
outlet at the moment of purchase.
The expansion of telephone ordering and internet shopping has also
required the development of security measures to enable Card Not
Present (CNP) transactions. The most common is the CVV2 (Card Veri-
fication Value 2) code, generally referred to as the ‘security code’, which
is a three or four digit number most often placed on the reverse of the
card. It does not however offer protection should the card be stolen. Visa
and MasterCard have therefore developed additional security procedures
such as Verified-by-Visa and SecureCode for MasterCard to be used on
the internet.
In addition to ensuring the authenticity of the card holder, card
providers have sophisticated transaction monitoring systems that ensure
they detect unusual activity or behaviour patterns on an account and can
take the appropriate actions to stop fraud as early as possible.
The Payments Card Industry Standards Council, regrouping the major
brands (Amex, Discover, JCB International, MasterCard, Visa), have also
developed comprehensive requirements for a Data Security Standard (PCI
DSS) covering security management, procedures, policies, network archi-
tecture, software design and backed by comprehensive self-assessment
questionnaires. All issuers, merchants and acquirers have to prove they
abide by these rules and undergo specific assessments to prove it.
68 The Structure and Economics of Payment Systems

4.4 Anti money laundering (AML) and counter-terrorism

finance (CTF)
Since 9/11, legislation and regulation to counter AML and CTF have
multiplied and placed an increasing burden on the banks.
The Financial Action Task Force on Money Laundering (FATF) was
established by the G-7 Summit as early as 1989 and, up until the time
of publication, was endorsed by 34 countries and had published 40 Rec-
ommendations against money laundering supplemented by nine Special
Recommendations against terrorism finance. The principal focus is on
Customer Due Diligence (CDD): not maintaining anonymous accounts;
verification of identity of retail customers; and business purpose and
identity of beneficial owners of corporate and correspondent banking
relationships. Accurate and meaningful originator information must be
included on all funds transfers and preserved throughout the entire pay-
ment chain. Records must be maintained for at least five years. Financial
institutions should report any suspicious activity, customer or financial
institution.
The USA Patriot Act (Uniting and Strengthening America by Providing
Appropriate Tools Required to Intercept and Obstruct Terrorism) of 2001,
among other provisions, strengthens most of the FATF recommendations
by incorporating them into law.
Ordering customers and beneficiaries must be checked against stop lists
of embargoed persons, entities or countries; each country maintains its
own, the most commonly used being issued by the US Office of Foreign
Asset Control (OFAC). These measures have forced many banks to invest
in substantial developments to ensure compliance, adequate audit trails
and investigation tools.
5
The Role of Payment Systems
in the Economy

Except under a barter economy, an exchange of money is necessary to

settle any purchase of goods or services, whether in fiduciary (cash)
or in scriptural money. In the latter case, the bank can choose to set-
tle via a payment system or directly with the creditor’s bank. This
practice, where banks held correspondent accounts with each other,
has largely disappeared for domestic payments within the same cur-
rency, but was widespread until technology enabled the introduction
of electronic payment systems during the second half of the twenti-
eth century. The correspondent account system would have in any
case become unmanageable because of the necessity to reconcile all the
accounts and the opportunity cost of maintaining idle balances scat-
tered around the banking system. Payment systems have enabled banks
to centralize the settlement of liabilities and reduced demands on liquid-
ity and processing costs through economies of scale and the networking
effect.

1 The economic role of payment systems

The role of payment systems is to ensure the convertibility of liabili-
ties on commercial banks, otherwise known as commercial bank money,
embodied by the balances (or credit lines) customers hold on their bank
accounts. The commercial bank assumes a claim acceptable over the mar-
ket on behalf of its client, followed by the central bank which substitutes
it for a claim in central bank money acceptable by all banks. This role
was, until the second half of the twentieth century, assumed in London
by the Accepting Houses which, subject to a fee, rendered ‘acceptable’
by guaranteeing it a bill of exchange between two customers drawn on
a foreign bank. In a modern payment system, the commercial banks use

69
70 The Structure and Economics of Payment Systems

their settlement account with the central bank to guarantee finality. The
payment system substitutes all interbank settlements by one settlement
in central bank money.
We noted in Chapter 1 (sec. 2.5) that an efficient money market
is indispensable to the smooth functioning of payment systems, as it
enables banks to fund their end-of-day settlement positions in DNS sys-
tems and to access intraday liquidity in RTGS systems. The centralization
of settlement and clearing operations contributes to the pricing of money
on the domestic market (interest rate) and of currencies on the interna-
tional markets (exchange rate). Even in a DNS system the bank’s treasurer
will follow the position throughout the day and, if required, start fund-
ing an anticipated short position early on to avoid higher interest rates
at closure if the market dries up or the spread between the lending and
borrowing rates has widened.
The main impact of a payment system is to unify the market. This was
observed at the launch of the euro and the TARGET high-value RTGS
system. Within a few months interest rates converged across the euro-
zone, except for minor differences due to country credit ratings. The
ECB was able to implement its monetary policy effectively through an
efficient payment system.
Much was written during the subprime crisis about the ‘interdepen-
dencies’ of risk between institutions and instruments across geographies.
The central banks responded by coordinated interventions which they
were able to implement, and measure the impact, through the rele-
vant payment systems and securities markets whose settlement systems
interconnect. Will the major high-value payment settlement systems
interconnect in the same way as we progress towards harmonization of
standards and procedures? Would that provide a more effective tool for
central bankers coordinating their interventions to avoid systemic risk,
or would the risk of single point of failure arise?

2 Payment systems and money velocity

Money velocity can be influenced by the performance of payment sys-
tems in terms of execution speed, cost and security. Money supply is
equal to the value of payments multiplied by money velocity. For a
given value of transactions, a highly efficient payment system will reduce
the speed of execution and increase the turnover of money supply.
Conversely, for a given value of transactions, the demand for liquid-
ity will diminish as money velocity increases, payment systems become
 The Role of Payment Systems 71

more efficient and execution time reduces. Theoretically, money sup-

ply should diminish for a given velocity, but increased demand from a
growing number of transactions is more than compensating the effect
of a higher velocity from technological advance: the growth of money
supply is slower than the growth of the GDP.
Central banks will use the information provided by payment systems
to determine their monetary objectives and to measure the delay in reac-
tion to their interventions. As settlement institutions, they have visibility
on all operations, not only payments but also government debt. The
monetary channel measures the diffusion of interest rates on the mar-
kets by arbitraging between the different maturities on the yield curve
or between different financial instruments. The balance sheet channel
reflects the impact of monetary policy on the value of the collateral
and consequently collateralized credit lines. The payment systems are
therefore one of the channels through with central banks transmit mon-
etary policy and can measure the impact of their interventions. As banks
require liquidity during the final phases of the settlement cycle, the
availability of central bank money is critical. They rely on the central
bank for day-to-day supervision of the markets and as lender of last
resort.
In theory, money velocity depends short term on interest rates and
long term on economic growth as the number and the value of transac-
tions increase. Historically, money velocity slowed until the 1950s and
accelerated thereafter. This deceleration can be explained by the mon-
etarization of the economy. That phase ended after World War II and
was followed by economic growth and the widespread adoption of bank
accounts: money velocity accelerated to stabilize since the turn of this
century.
Looking at Fisher’s equation MV = PT (Money supply × Velocity =
Price × number of Transactions), an increase in performance of payment
systems, given a constant money supply, should reduce the number of
payments (or limit their growth) required to effect the transactions. The
leverage of monetary policy should be reinforced by shrinkage of the
money supply or its lower growth than GDP. As velocity increases,
the number of transactions diminishes. At the extreme hypothetical
case, an infinite velocity (nil execution time) would result in a number
of transactions declining to zero: we would revert to a barter econ-
omy with networks of supercomputers swapping goods and services
in real time; money would have been substituted by the payment
system and the central banks would have lost control of the money
supply!
72 The Structure and Economics of Payment Systems

3 Network economics
As with all networks, payment systems will benefit from economies
of scale and value-added services which increase its attractiveness and
profitability. Under a regime of perfect competition, prices will con-
verge towards the marginal cost. As the profitability of the core generic
payment service will diminish with competition, competitive differen-
tiation will be dictated by segmented value services through which the
service providers will attempt to increase their profits. Ultimately, the
core payment service becomes a public utility while value-added services
are market-driven.
Networks are driven by economies of scale; costs are largely fixed so
additional transactions reduce the unit processing cost. The value of a
network depends therefore on the development investments, the ser-
vices offered and the volume of transactions processed. It is also linked
to the square of the number of participants according to Metcalfe’s Law.
These factors can however exercise a negative influence, for instance
when a network reaches saturation, obsolescence or even becomes a
de facto monopoly leading to complacency and reluctance to invest in
new developments.
We have seen in Chapter 1 (sec. 2.2) how payment system owners can
vary the various pricing components (joining fee, annual charge, trans-
action fee) according to their objectives at each phase in the life-cycle
of the system. Once a network has reached critical mass, the feedback
effect kicks in: unit costs diminish with growing volumes; new adher-
ents seek to join; and investments can be devoted to the development
of new products and services. Telecommunication operators for instance
decided to invest in mobile telephony in the full knowledge that it would
cannibalize their revenues from fixed lines calls and sought alternative
revenues from mobile and other services such as broadband and value-
added telecommunication services. Similarly, SWIFT launched its file
transfer services knowing that the revenues from the individual message
charge would suffer.
Some systems, such as SWIFT, become immune to new entrants com-
peting on their core transmission services owing to the sheer numbers of
participants and geographical coverage: the barriers to entry are too high.
Volume in payment systems is however highly concentrated among a
small number of large institutions: in the UK for example, the five largest
participants represented 80 per cent of the number of payments over the
CHAPS RTGS, 76 per cent of cheque clearing and 76 per cent of the BACS
ACH volumes in 20061 . With the exception of RTGS systems, payment
 The Role of Payment Systems 73

networks are therefore constantly under the threat of a couple of large

users deciding to exchange payments bilaterally, settle the net amount
over the relevant RTGS and retain participation to transact with the
rump of small volume users or remote geographical locations. Each bank
merger generates more ‘on us’ payments which are removed from pay-
ment systems. Payment systems and networks are therefore constantly
seeking to develop value-added services to lock-in customers.
The trading and settlement of European securities is currently domi-
nated by three systems, interconnected to other major financial markets.
New entrants are likely to concentrate on niche markets and trading
platforms have emerged, such as Turquoise. In spite of the backing of
major participants, few survive and most eventually retreat to special-
ized instruments, often after the incumbents have reduced fees. Mission
accomplished some will say, raising the question of whether they were
mainly launched as a scarecrow. Another example is the ill-fated attempt
by Eurex, the derivatives exchange, owned jointly by Deutsche Börse
and the Swiss Exchange, to set up a trading and settlement platform in
Chicago which closed after two years. One consequence was the merger
between the two local incumbents: the CBOT and the CME.
By their very nature networks become a monopoly used by an
oligopoly of participants. Rising volumes, declining unit costs, prod-
uct and technological innovation contribute to this concentration. The
barriers to entry to compete on the core service becoming too high,
new entrants are constantly seeking value-added services attacking the
Achilles heel of the incumbent to establish new networks. They will
establish a temporary monopoly before being in turn challenged by a
new entrant. Ultimately, as the window of competitive advantage cre-
ated by new services shortens, the networks regroup and consolidate as
we will see in the subsequent chapters.