SlideShare a Scribd company logo
1 of 78
Download to read offline
Interchange fees, a balancing act
An empirical examination of a possible
regulatory benchmark
Michal Kalina1
Duisenberg School of Finance
April, 2015
Thesis, LLM Finance & Law
Supervisor: prof. dr. Maarten Pieter Schinkel
1 Author can be contacted through: michal@savnar.nl
2
Interchange fees, a balancing act
An empirical examination of a possible regulatory benchmark
Michal Kalina1 (20120064)
Duisenberg School of Finance, LLM Finance & Law Programme
Master thesis, 20 April 2015
Abstract
In this paper I show the notional development of the level of the interchange fee for
debit card payments in the Netherlands, based on an alternative method to the tourist-
test. The alternative has a few a priori attractive features, such as a closer resemblance
to a debit card favouring market, whereas most of the literature does not distinguish
between debit and credit cards or focusses on the latter. Countries in Europe with high
card usage per capita and low interchange fee levels, are typically debit card markets.
However, using cost data for 2002 and 2009 I show that the method is likely to exhibit
market disturbing effects by reversing the market structure through a negative and
much larger interchange fee than in reality. In the long run, calibrated to actual
institutional settings, the method would create the wrong incentives, stimulating the
use of the more expensive payment instrument and discouraging the more efficient. The
results indicate that application of this alternative method would be ill-advised for
countries such as the Netherlands, with rising costs for cash and declining costs for
debit card; thus, regulators should be looking for other benchmarks.
1
I would like to express my gratitude to a few people without whose support or doings, this thesis would not have
been accomplished: Joe McCahery for directing a superb programme in Finance & Law; Maarten Pieter Schinkel
for sharing his insights on the subject and support to set the scope; Søren Korsgaard for elaborating on his model,
enabling me to properly calibrate it; Nicole Jonker for showing me the adjustments in cost item classifications in
the 2009 data set compared to the 2002; Frank Leerssen at Rabobank for his support and flexibility to combine
my work and study; and foremost my wife, Caroline Spoor-Kalina, for the best support I could wish for throughout
the course, this thesis and beyond.
The views expressed in this study are my own, as are any remaining errors.
3
Table of Contents
1 INTRODUCTION..............................................................................................................................................8
1.1 Background and motivation.............................................................................................................8
1.2 Research question .............................................................................................................................11
1.3 Academic contribution.....................................................................................................................12
1.4 Thesis structure..................................................................................................................................13
2 PAYMENTS MARKETS, INTERCHANGE FEES AND REGULATION ..........................................14
2.1 Basics and terminology ...................................................................................................................14
2.2 Social costs of payments..................................................................................................................15
2.3 Challenges in payment markets...................................................................................................15
2.4 The concept of interchange fees ..................................................................................................17
2.5 Justifications for interchange fees...............................................................................................18
2.6 Some landmarks in interchange fee litigation........................................................................20
2.7 A note on SEPA and PSD..................................................................................................................23
3 REVIEW OF LITERATURE........................................................................................................................26
3.1 Preliminary...........................................................................................................................................26
3.2 Models of interchange fees and policy implications............................................................27
4 ANALYTICAL FRAMEWORK ...................................................................................................................37
4.1 Interchange Fee model ....................................................................................................................37
4.2 Costs model ..........................................................................................................................................41
5 DATA ................................................................................................................................................................45
5.1 Data collection 2002.........................................................................................................................45
5.2 Data collection 2009.........................................................................................................................46
5.3 Data..........................................................................................................................................................47
6 ESTIMATION.................................................................................................................................................49
6.1 Payments developments in the Netherlands 2002 - 2009................................................49
6.2 Calibration ............................................................................................................................................53
4
6.3 Obtained results .................................................................................................................................56
6.4 Robustness............................................................................................................................................57
7 CONCLUSIONS..............................................................................................................................................60
7.1 Discussion of the results .................................................................................................................60
7.2 Limitations of the study...................................................................................................................61
7.3 Afterword..............................................................................................................................................61
8 REFERENCES ................................................................................................................................................62
8.1 Bibliography.........................................................................................................................................62
8.2 (Pre-) Legislation, Cases and Miscellaneous Links...............................................................66
9 APPENDICES .................................................................................................................................................69
5
LIST OF ACRONYMS AND ABBREVIATIONS
ABC Activity Based Costing
ACH Automated Clearing House, see CSM.
ACM Autoriteit Consument & Markt. Founded in April 2013, merger of Dutch
Telecom regulator (OPTA), Dutch competition authority (NMa) and Dutch
Consumer Rights supervisor.
AFM Autoriteit Financiële Markten. Dutch supervisor of financial markets
conduct.
AML Anti-Money Laundering
ATM Automated (or automatic) Teller Machine.
BBAN Basic Bank Account Number.
BGC BankGiroCentrale. Former interbank payments processor (CSM) in the
Netherlands.
CSM Clearing and Settlement Mechanism. In the context of this paper, CSM can
be deemed synonym for ACH. More info: e.g. Kokkola (2010).
CUP China Union Pay, a four-party scheme card brand based in Shanghai.
DNB De Nederlandse Bank, Dutch central bank.
EC European Commission.
ECB European Central Bank.
ECJ European Court of Justice.
EEA European Economic Area: all Member States of the EU plus Iceland,
Norway and Liechtenstein.
EIM Economisch Instituut voor het Midden- en Kleinbedrijf (~Economic
Institution for the SME sector). Now part of Panteia.
EP European Parliament.
EPC European Payments Council. More info:
http://www.europeanpaymentscouncil.eu/
FD Financieel Dagblad, Dutch leading daily financial newspaper.
HACR Honour All Cards Rule. A business rule set by a scheme prohibiting the
merchant who accepts a particular card brand to distinguish between
issuers (general implementation of the rule in Europe, differs from the US).
IBAN International Bank Account Number.
6
IF Interchange Fee.
JCB A three-party scheme credit card company based in Tokyo.
MIF Multilateral Interchange Fee.
MinFin Either the Dutch Minister of Finance or the Ministry of Finance.
MOB Maatschappelijk Overleg Betalingsverkeer (~Societal Consultation on
Payments), by order of MinFin in 2002. Chaired by DNB. More info:
http://www.dnb.nl/binaries/Oprichting%20MOB_tcm46-274623.pdf
NBC Nationale BetalingsCircuit. See Box 3.
NFC Near Field Communication or Near Field Communication technology.
NMa Nederlandse Mededingingsautoriteit. Dutch competition authority. Now
part of ACM.
NVB Nederlandse Vereniging van Banken (Dutch Banking Association)
NYSE New York Stock Exchange
OFT Office of Fair Trading. UK’s consumer protection and competition
authority.
OJEC Official Journal of the European Communities.
PIN Personal Identification Number. Also the brand and scheme of the former
Dutch domestic debit card. Became part of Dutch vocabulary: to pin is to
make a debit card payment or ATM cash withdrawal.
POS Point Of Sale.
PSD Payment Service Directive (Directive 2007/64/EC). See References.
PSP Payment Service Provider, term introduced in PSD
RTGS Real Time Gross Settlement (system)
SCT SEPA Credit Transfer (scheme)
SDD SEPA Direct Debit (scheme)
SEPA Single Euro Payments Area. The jurisdictional scope of the SEPA Schemes
currently consists of the 28 EU Member States plus Iceland, Norway,
Liechtenstein, Switzerland, Monaco and San Marino. Note that this is a
larger area than the Euro-countries. More info:
http://www.europeanpaymentscouncil.eu/index.cfm/knowledge-
bank/epc-documents/epc-list-of-sepa-scheme-countries/
SME Small and Medium-sized Enterprise
SO Statement of Objections
7
St.BEB Stichting Bevorderen Efficient Betalen (Foundation to Promote Efficient
Payments). Founded in Nov. 2005 to manage EUR 10 million donated by
the banks as agreed in the Convenant Betalingsverkeer. More info:
http://www.efficientbetalen.nl/
SWIFT Society for Worldwide Interbank Financial Telecommunication. A member-
owned cooperative, predominantly owned by banks.
TFEU Treaty on the Functioning of the European Union (2012).
TPP Third Party PSPs, term introduced in PSD2
8
1 INTRODUCTION
“The most important financial innovation that I have seen the past 20 years is the automatic teller machine,
that really helps people and prevents visits to the bank and it is a real convenience. How many other
innovations can you tell me of that have been as important to the individual as the automatic teller machine,
which is more of a mechanical innovation than a financial one?”
Paul Volcker, former FED Chairman1
1.1 Background and motivation2
The history of the payments industry is the history of banks and money. And it is full of
“mechanical innovations”, some of which successful like the ATM, while many others were
not or only briefly so. For example, after almost 20 years the Dutch banks have
decommissioned the Chipknip, a domestic e-wallet type payment product3, per the beginning
of this year. An empirical cost study in 2002 showed that a payment made by Chipknip had
already then by far the lowest variable social costs compared to cash, debit card or credit
card4. For the longest time most consumers had one in their wallet and the system had
worked without an interchange fee, yet as a payment product it never became a real success5.
One may wonder why. Clearly, the absence of an interchange fee is no guarantee for a
successful payment instrument. Nor is the presence of one. This paper is however not a post-
mortem on the product.
The worldwide success of payment cards is undisputed and it provides a fruitful ground to
academics for study. And for regulators to intervene, so it seems. For long, offering payment
services has been the exclusive domain of banks and their legal monopoly on providing
current accounts or creating money remains until today. Being mostly private institutions
entrusted with the facilitation of a public good6, banks are obligated to comply with a sizeable
and increasing set of rules and regulations. On the one hand, this may ensure the security and
1
Speech at the occasion of addressing ideas for reforming financial services at Wall Street Journal Future of
Finance Initiative in the UK, December 2009. [1], 21 August 2014.
2
Several terms and concepts introduced in this paragraph will be properly explained later on in this paper.
3
e-wallets currently on the market are more advanced and can typically contain several virtual payment ‘cards’.
The Chipknip was designed as an alternative to coins and low-denominated banknotes.
4
3 eurocents, compared to 11, 19 and 80 eurocents for a cash, a debit card and a credit card transaction
respectively. Source: (DNB) Brits & Winder (2005), table 4.3, p.26.
5
The officially first Chipknip transaction was done by Wim Duisenberg, then president of DNB on October 26,
1995. It reached its peak in 2010 with 178 million transactions that year; by comparison, there were 2,154 million
debit card transactions (source: Currence Annual Report 2011, p.8). The decommissioning was announced by the
banks and the brand owner Currence in March 2013, to take effect as of 1 January 2015.
6
See also e.g. Freixas & Rochet (2008) par. 1.1.
9
safety of the entire payment system. For example, banks are expected to take appropriate
measures to fight illegal activities such as fraud, money laundering, terrorism financing or tax
evasion, or prevent payments to entities (e.g. persons, institutions and countries) that have
been sanctioned1. On the other hand the regulatory burden amounts to a significant cost level
and since the efficiency of the payment system is also of fundamental concern to society, it
constitutes another justification for public intervention. A frequently taken route by
governments to increase efficiency, is to create or let market participants create (regulated)
monopolies that can then capture the economies of scale that so typically characterise
payments processing volumes2. Another way to increase efficiency is to increase competition,
or so the contemporary paradigm dictates. In the payments industry however this can lead
to different or even opposite results3. To illustrate this point from theory: Matutes and Padilla
(1994)4 model a case of three competing banks who choose membership to some ATM
network and find that if equilibrium exists, it will not be efficient as there will be either three
incompatible ATM networks (so the cardholder can only withdraw cash at ATMs from his
own bank) or two banks sharing their ATM network and leave the third bank out. To illustrate
the point from practise: it is generally posited that the US banking sector and the US payments
industry is competitive, and more competitive than for example (most countries in) Europe.
Yet the interchange fees for credit cards are significantly higher in the US (around 50%) than
in Europe, the UK or Australia and that was even before regulatory action brought it further
down in the latter three regions5. One explanation offered for this phenomenon is known as
reverse competition and can be summarized as follows: card schemes compete with each
1
Economic and political sanctions on territories and/or specific persons therein; current examples include Russia,
the Crimean part of the Ukraine, Iran, Cuba (although these are now in the process of being (partially) lifted),
Zimbabwe, North Korea, and others. There are multiple sanction lists, most importantly the US OFAC list, the
EU-sanctions list and lists maintained by designated agencies (e.g. Interpol). Many banks have recently found
themselves facing criminal charges and huge penalties for not taking appropriate measures and/or failing to report
suspicious transactions. For example: JP Morgan Chase $1.7 billion penalty regarding Madoff’s Ponzi scheme;
HSBC for facilitating transactions to and from Cuba, Libya, Iran (and others) and for drug cartels in Mexico and
Colombia ($1.9 billion), similarly BNP Paribas ($8.9 billion); UBS ($780 million) and Credit Suisse ($2.6 billion)
for facilitating tax evasion; Commerzbank for “unsafe and unsound practises” violating US sanctions and AML
laws ($1.7 billion). [2], [3], [4], [5], 12 March 2015.
2
Beijnen and Bolt (2009), in Bolt (2013) p.75, estimate economies of scale for card payments about 0.25–0.30,
based on cost data of 8 European processors spanning 15 years. This means that a 100% increase in payment cards
volume leads to just a 25–30% increase in total costs.
3
Opposite to conventional economic wisdom. Economic history seems apt in bringing forward counter-intuitive
results. For example Gresham’s Law (“bad money drives out good money”) or Akerlof’s lemon (“poor quality
second-hand cars drive out good quality second-hand cars”). Some scholars regard government intervention as a
cause of counter-intuitive results, e.g. Fisher Black on systemic risk: “… It means that they [the government] want
you to pay more taxes for more regulations, which are likely to create systemic risk by interfering with private
contracting…” (in Danielsson, 2013, p.35).
4
Taken from Freixas & Rochet (2008) p.87.
5
See Hayashi (2004), p.3, chart 1, also to be found in Weiner & Wright (2005), p.299, chart 2.
10
other by offering ever higher interchange fees to banks that issue their cards. These IFs
constitute revenues to these bank and this results in higher fees for card payments for
merchants who pass these costs on to consumers, either directly or indirectly1. The European
Commission (EC) uses this argument in their assessment of the economic effects of
interchange fees2. A second effect of reverse competition is that the increased interchange fee
(IF) level then may serve as a protection for incumbents from market entry of new card
schemes as these will at least have to match the existing IF level to get a foothold.
Similarly, the increasing regulatory burden can also work as a barrier to market entry as a
large part thereof are sunk costs. In this sense it may perhaps seem ironic that the EC has
recently proposed more regulation to increase competition and foster innovation on the
payments market. I’m referring in particular to the ‘Payments Legislative Package’3,
consisting of an Interchange Fee Regulation (IFR) and a revised Payments Services Directive
(PSD). Years of legal combat and emotional debate have preceded the IFR4, in courts and in
political and academic circles5. The IFR has largely been voted for on 10 March 20156 in the
Plenary of the European Parliament7. Once official, a maximum IF will apply for debit cards
of 0.2% and for credit cards of 0.3% for cross-border consumer transactions as of 2 months
after the IFR enters into force (art. 3) and equally for all consumer transactions as of 2 years
1
By applying a surcharge on the card payment or by slightly increasing overall prices of the goods sold.
2
See for example SDW(2013) 288 final, the EC’s Impact Assessment accompanying the proposals for PSD2 and
IFR (see footnote 3 below). Europe Economics, a London based consultancy firm, in a report commissioned by
MasterCard finds this assessment to be “plainly wrong” (Europe Economics, 2014, p.11).
3
Presented by the EC on 24 July 2013. Core pieces are COM(2013) 547 final, hereafter PSD2 (Payment Services
Directive 2); and COM(2013) 550 final, hereafter IFR (Interchange Fee Regulation). The EC currently expects
adoption in May 2015.
4
Parts of the PSD2 are also still under heavy debate, see §2.7.
5
As another example of possible adverse outcomes: several banks have voiced that the EC’s choice to let the
cardholder choose the payment card brand at the moment of the transaction when more than one brand is accepted
at the POS (IFR art. 8(18)), will lead to higher costs for the users and strengthen the MasterCard/VISA duopoly
in Europe. The argument is that the cardholder will choose the brand that’s on top of mind which is most likely
one of the two international brands and not some cheaper yet less known local brand; thus, this will start an ever
intensifying advertising campaign, a race ultimately to be won by the two deep-pocketed companies. See for
example [12], 9 April 2015. The hypothesis bears similarity to the reverse competition argument, formal economic
literature however points in the opposite direction, see e.g. Rochet & Tirole (2002). For more on consumer
payment behaviour, see e.g. HBD (2002), Jonker (2007), Bolt, Jonker and Renselaar (2008), Jonker, Kosse and
Hernandez (2012), Kosse and Vermeulen (2014) and Bagnall et. al. (2014).
6
Proposed amendments to articles have been taken into account in this paper as far as possible and relevant until
31 January 2015. As the exact final wording of the entire text is not yet officially published, I use the text as
published in 2013 (COM(2013) 547 and 550), unless otherwise indicated.
7
Similar developments take place all over the world. For example Rochet & Wright (2009, p.4-5) report “...more
than 50 lawsuits concerning interchange fees filed by merchants and merchant associations against card networks
in the US, while in about 20 countries public authorities have taken regulatory actions related to interchange fees
and investigations are proceeding in many more”. Hayashi (2013) gives an updated overview and accounts for
over 30 countries where regulatory actions have been taken.
11
after the IFR enters into force (art. 4). These caps have been calculated by MasterCard1 using
a test called the (merchant) avoided-cost test, popularised as the ‘tourist-test’, derived from
Rochet and Tirole (2008). The EC finds it a “reasonable benchmark for assessing a MIF level
that generates benefits to merchants and final consumers”2. To calculate the figures
MasterCard used three empirical cost studies of the central banks of Belgium, Sweden and
the Netherlands (DNB). The latter study was published by Brits and Winder (2005) and
contains data pertaining to the year 20023. MasterCard implemented the caps on their cross-
border MIFs with the EC’s consent as of July 2009 (see also §2.7).
Jonker and Plooij (2013)4 have analysed the tourist-test using the Brits and Winder data and
a second dataset pertaining to 2009 and report in an understatement that their results
indicate the test can have “unintended consequences”: in a country like the Netherlands
where social costs of cash are increasing and the social costs of debit card payments are
decreasing, the IF level may actually more than double from 0.2% to 0.5% of an average debit
card transaction size and still pass the test (in other words: the test would qualify the IF level
as ‘not excessive’). If this rise were to be passed on to merchants completely, they would
experience a surge in their costs by 233%. Moreover, where an IF is intended to cover (part
of) the internal costs of the issuing bank, the calculated tourist-test IF level would be higher
than the issuing and acquiring costs of the banks combined. The researchers therefore
recommend competition authorities to look for a different benchmark.
1.2 Research question
The objective of this study is to examine an alternative model to the tourist-test and analyse
how that model, if applied, would affect the IF level for debit card payments in the
Netherlands over time, using cost data for 2002 and 2009, i.e. the same data sources that the
DNB researchers used for their empirical test of the tourist-test. This alternative model has
some preferable features, in particular a closer resemblance to a debit card favouring country
such as the Netherlands, whereas the Rochet and Tirole’ model does not distinguish between
debit and credit cards. The main question to be answered is:
Does the method present a viable alternative to the tourist-test benchmark?
1
Schwimann (2009).
2
Press release by the EC, Commissioner Kroes, 1 April 2009, memo 09/143.
3
The Belgian and Swedish studies concern the years 2003 and 2002 respectively.
4
Also published by Bolt, Jonker & Plooij (2013).
12
PM: I make a distinction between the formal model and its operational counterpart which I
casually refer to as ‘method’.
1.3 Academic contribution
After Baxter (1983) his contribution, the topic of interchange fees faded from academic view
and policy circles. By the end of the 1990s the topic started to attract academic attention, in
parallel with law suits and regulatory investigations in a number of countries in the world.
Rochet and Tirole (2000) published a seminal paper and since then economic literature on
the topic surged, together with the building of knowledge on the functioning of two-sided
markets. The more formal economic literature in the first decade of this century generally
yields two broad conclusions: 1) IFs are foremost a balancing device, as opposed to a collusive
device, their level depends on much more (if at all) than the costs of producing the payment
alone; 2) there is no apparent need for a regulator to intervene. These findings are of
particular importance considering that regulators and courts usually employ a test based on
issuers’ costs in their assessment of possibly excessive merchant fees. Some more recent
academic contributions however do point to market failures that might need mending by
regulation. Confrontation with empirical data was for long very rare. As far as I am aware,
Brits and Winder (2005) were one of the very first to publish empirical cost data (see also
section 5). Their study concerned the Dutch POS1 payment market in 2002. Similar studies
by other central banks in European countries followed, often adopting their conceptual cost
model. Schmiedel, Kostova and Ruttenberg (2012) present the aggregated results of thirteen
of these empirical studies, most of which with data on 2009. So far, only for the Netherlands,
Sweden and Norway detailed cost studies on two or more years are available2 (year 2009
data on the Netherlands and Sweden were part of the study by Schmiedel c.s.). This offers a
first opportunity to study the development in social costs of card payments. Jonker and Plooij
(2013) use two datasets on the Netherlands in their examination of an application of the
tourist-test. To my knowledge Korsgaard (2014) is the first to construct a formal model that
resembles more that of a debit card favouring market, whereas other papers focus either on
credit cards or on the payment activity, rather than the particular instrument. He also reviews
his model in the light of empirical data for Denmark3 and finds a ground for regulation (as is
1
Point Of Sale.
2
See Jonker (2013) p.9.
3
Most of the data Korsgaard uses is published in Danmark Nationalbank (2011, 2012) which were the basis of
the Danish part of the cost study by Schmiedel c.s.
13
already the case in Denmark). My study is the first to calibrate that model to two different
years and to analyse the predicted notional IF levels over time. By using the same datasets
and data sources as in the empirical test of the tourist-test for the Netherlands, the
predictions of the two models can be compared.
1.4 Thesis structure
The remainder of this paper is structured as follows. Section 2 offers an introduction to the
retail (card) payments industry, introduces basic terms and concepts and touches upon
regulatory developments. Section 3 reviews the literature on interchange fees with a focus
on formal models that allow for a welfare analysis. Section 4 presents the analytical
framework to examine the notional interchange fee levels in the Netherlands, consisting of a
formal model for optimal interchange fees and a cost model that captures the concept of
social costs in payments. Section 5 outline how the data for the two years, 2002 and 2009,
have been collected and presents the most important data. Section 6 then shows how I
calibrate the model with the empirical data, presents the results and extends some
robustness checks. Section 7 discusses the results and offers some recommendations for
future research and policy. Finally, section 8 contains the references.
Reading notes
For illustrative purposes, I sometimes make references to news articles, position papers etc.
that can be found on the internet. I refer to them, mostly in footnotes, with a simple [#], which
is short for: “see References, link [#number]”, for example link ‘[9]’. The references link is
followed by a date indicating when I last accessed it.
Occasionally I place short stories, also meant as illustrative background information, in boxes.
They can be skipped without loss of thread of this study.
14
2 PAYMENTS MARKETS, INTERCHANGE FEES AND REGULATION
This section provides a brief introduction to the retail (card) payments industry and a context
for putting the findings of my research into perspective.
2.1 Basics and terminology
In retail payments there are usually four parties involved: the payer and the payee who
transact with each other, and their respective banks. In their roles of economic agents, the
payer is normally referred to as the “buyer” (of goods) and the payee is the “seller”: the
payee/seller sells goods or services to the payer/buyer and in return the payer offers cash,
writes a cheque, uses his debit or credit card, initiates a credit transfer, mandates the payee
to initiate a direct debit, or uses some other payment instrument1. In the context of (wire)
transfers, the payee is normally referred to as the “beneficiary” (of the funds) and his bank is
the “beneficiary’s bank”. In the context of payment cards, the seller is usually referred to as
the “merchant” and his bank as the “acquirer” or “acquiring bank”; the buyer is then referred
to as the “cardholder” and his bank as the “issuer” or “(card) issuing bank”.
A debit card is issued by the issuing bank to the cardholder and provides him electronic
access to his current account2, either through internet/online banking, to withdraw cash
from an ATM or to initiate a transfer of funds at a POS terminal. A credit card on the other
hand is not directly linked to a current account and in three-party models (see §2.3), the
credit card company is an entirely different firm than the bank where the cardholder holds
his current account.
Retail payments between banks on behalf of their customers, millions per day, are usually
not executed individually but sent to clearing institutions or Clearing and Settlement
Mechanisms (CSMs) who sort, match and net all of the incoming and outgoing payment
instructions and then calculate the net amount to be paid or received per bank participating
in the mechanism. Information on these calculated net amounts is then sent to the settlement
institution, usually the local central bank, who settles the claims. This means that the central
bank, where all domestic banks hold an account, transfers funds between these accounts
corresponding to each bank’s claim, provided there is sufficient liquidity or collateral. The
banks are only liable for these net amounts but the banks receive details of all underlying
1
I use the term “payment instrument” in a general meaning as a way to initiate a transfer of funds. Whether a
paper cheque, a plastic card with a magnet stripe or chip, a chip with NFC inside a mobile phone or an app on a
smartphone, the exact technology or carrier is not of particular relevance in the context of this paper.
2
Synonyms for “current account” are “payment account” and “checking account” (typically US custom).
15
transfers from the CSM, enabling them to credit each customer’s account accordingly. The
settlement is usually a daily process at the end of the business day, but frequently the
participants choose to settle several times a day1.
2.2 Social costs of payments
From the point of view of society, payments are costly: estimations of the costs of retail
payments for the European Union (EU) are roughly 1% of GDP, i.e. € 130 billion. Banks and
interbank infrastructures (e.g. CSMs) incur 50% of these social costs, retailers about 46%,
central banks about 3% and the remaining 1% is incurred by cash-in-transit companies (e.g.
conveyance of banknotes)2. Social costs refer to the costs of the resources employed to
‘produce’ the payment services, that is: the sum of all internal costs by all the agents involved
to execute a payment transaction. Private costs are all the costs a single agent faces and/or
incurs to produce a payment transaction and include by definition internal and external costs.
External costs are the charges (e.g. fees) the agent pays to other agents in the value chain.
Since the external costs to one agent in the chain are revenues to others, these external costs
cancel out in the calculation of total social costs. Equivalently, social costs can be defined as
the sum of all parties their net costs (total costs – revenues)3.
2.3 Challenges in payment markets
Payment markets exhibit interesting dynamics, in part resulting from multiple potential
market failures. I will indicate a few, in no particular order. Banks are in competition with
each other and increasingly with non-banks. Yet a payment transaction cannot take place
without some basic form of cooperation between the bank (or party) acting on behalf of the
payer and the bank (or party) acting on behalf of the payee, at the very least on some standard
of exchange4. The element of cooperation may conflict with the element of competition. An
1
Next to this there are systems to settle individual, typically high-value transactions in real-time, so-called RTGS
systems, for example TARGET2 used by e.g. central banks in Europe. The CSMs in the payment cards business
are usually different from the CSMs used for wire transfers. Indeed, within the payments business these are rather
separate domains. Typically cards settlement is a daily cycle (end to end) whereas wire transfers often take a few
working days. For example an SCT, see §2.7, is regulated to take no more than 2 business days, that is: the
crediting of the beneficiary’s account must happen at the latest at the business day following the business day on
which the bank received and accepted the payer’s payment instruction (as of 1 January 2012 (PSD, art. 69); until
2012 this was max. 4 business days and before the introduction of the PSD, this could be even longer).
2
Schmiedel, Kostova and Ruttenberg (2012), p.6. Numbers refer to the year 2009. This excludes social costs
incurred by households, which would add on average another 0.2% if those were to be included (id. p.7).
3
See e.g. Schmiedel c.s. (2012), Brits & Winder (2005) or Jonker (2013). Section 4 will give more details.
4
This is the case even when the economic transaction is settled in cash (with fiat money): if a merchant sells goods
and accepts some ‘pieces of paper’ in exchange but later finds out other economic agents such as the banks do
not, then he is likely to end up with a real loss. On a larger scale, the economy might then revert to barter trading.
16
industry practise is to separate (as a first step) the set of standards, e.g. the ‘scheme’ and
scheme rules, from the product, i.e. the payment instrument. The latter is the object of
competition, whereas the former defines the space where cooperation is, under certain
conditions, allowed; in the case of payment cards the space has been gradually narrowed
down by courts and regulators where for example certain scheme rules have proven to
hamper competition, see section 3. For a standard to be set effectively requires the
cooperation of a significant part of the market to gain critical mass, which may pose
coordination problems. Similarly, for a card scheme to be successful requires attracting a
sufficient number of merchants as well as consumers (cardholders). Three-party schemes,
for example American Express, Diners Club, JCB and Discover, who issue their cards directly
to cardholders and directly acquire merchants, can set a price structure and price level that
appeals to both sides of the market. Four-party schemes on the other hand, such as
MasterCard, VISA and CUP, do not deal with both sides of the market simultaneously and are
therefore unable to directly balance the costs and benefits to both. This may be a role for
interchange fees. But without the right price structure there may be no payment card product
in the first place. Typical of payment card markets and two-sided markets in general, is the
asymmetrical distribution of costs and benefits, resulting in a (heavily) skewed price
structure: one side of the market bears a significantly larger part of the total costs than the
other. More on these points follows in the next paragraphs and section.
Strong economies of scale may lead to monopolistic platforms with the most efficient cost of
production possible but equally to a risk of abuse of the dominant position and/or a price
setting below socially optimal, thus leading to an underprovisioning of payment cards and
furthermore a lack of incentives to continue to improve efficiency through innovation. The
presence of switching costs to bank customers may reduce banks’ incentives to make their
(ATM) networks interoperable. The same mechanism applies to banks themselves: the
presence of switching costs may reduce interbank networks (e.g. CSMs) to become
With cash, the payment instrument and the standard of exchange coincide. With electronic money there is a
difference: some technological standard for data exchange and the acceptance of these data as a change in debt
obligation (through entry in the bank’s books), that is: liability or ‘money’. This is indeed somewhat ambiguous
because ‘money’ has no clear and unambiguous real-world definition, despite being such a fundamental economic
concept. Recall the definitions of monetary aggregates M0, M1, M2 etc., see e.g. Danielsson (2013) p.36. See
also §2.5.
17
interoperable1, which could ultimately make a transfer of funds between two banks
connected to two different networks, impossible2.
As a last example, I revisit the role of banks as guardians of a safe and secure payment system.
For this they need to share payments information and sometimes information on the
transacting parties, the beneficiary and payer. This sharing of information can conflict with
consumer data protection and privacy rights. These conflicts can occur on national level but
are exacerbated on international level as countries, e.g. the US and the EU, or Member States
within the EU, not always agree or align their positions3.
2.4 The concept of interchange fees
The concept, the mechanics so to say, of interchange fees is straightforward. Following the
setting of payment cards in a stylized setup, picture a consumer (the cardholder) using his
payment card to pay for some goods at a merchant’ store. The cardholder received this card
from his issuer; in case of a debit card this is usually his own bank, in case of a credit card this
can be his own bank, a company affiliated with his or some other bank or a totally unrelated
firm (e.g. American Express and other three-party schemes). This relationship is mirrored at
the other side of the trade: the merchant can accept this payment card as a legitimate way of
receiving payments ever since he has a contract with his acquirer4. When the cardholder
swipes, or nowadays ‘dips’ his card at the POS terminal5, he initiates an authorisation request.
In case of a debit card this request is sent to his bank and if there is enough money on the
account, the transaction will be authorised. In case of a credit card, the request is sent to the
acquirer who authorises the transaction6. The authorisation is observed by the merchant who
1
See also Box 3.
2
Matutes and Padilla (1994) in Freixas and Rochet (2008); for more on the economic effects of switching costs
in general see for example Klemperer (1987).
3
Recall for example the conflict between the EU and the US in the summer of 2006 on the latter’ Terrorist Finance
Tracking Program where they, based on their International Emergency Economic Powers Act of 1978, gathered
information on financial transactions from SWIFT, the world’s largest financial communications network,
following the events of September 11, 2001. The EU forced SWIFT to process intra-EU financial messages within
the EU, removing their backup systems from US territory.
4
The merchant also need a contract with at least a telecommunications network provider for data exchange
between the terminal and a switching centre.
5
For the example it is irrelevant whether it is a POS transaction in the physical world or an online transaction, the
IF works exactly the same way.
6
In practise, there is obviously more to this. For example, the card has to be validated first, checked for not
reported stolen, missing or other fraudulent options and the authorisation request is first sent to a switching centre,
perhaps rerouted to another before it arrives at the issuer. The request can also be handled offline by the terminal.
See e.g. Kokkola (2010) for more on these mechanisms. Notice the difference in risk and liability between a debit
and credit card transaction. Credit card companies will sometimes contractually shift the risk of e.g. eventual non-
payment to the merchant.
18
then can rest assured he will receive a transfer of funds to his bank account in an amount
equal to the selling price of the goods. An IF is simply an amount of money, a fraction of the
value of this funds transfer, that usually goes the opposite direction: from the acquirer to the
issuer1. It is calculated per card-transaction, either some percentage of the value of the
transaction or a fixed fee per transaction.
The IF can be determined either bilaterally between the issuer and acquirer in an arm’s-
length agreement or multilaterally by the scheme. The multilateral interchange fee (MIF)
then applies as a default fee, a fallback in the absence of a bilateral agreement. From a
practical point of view, the MIF accomplishes that the transaction between the merchant and
cardholder can proceed, without the issuer and acquirer ever having made agreements with
each other. The efficiency gain is obvious: with more than 8,000 credit institutions (banks) in
Europe alone, one would need over 32 million bilateral agreements to accommodate all
possible card payment transactions. Disturbing to regulators is that a MIF, in case of VISA and
MasterCard before they were listed (and even thereafter) used to be set by an association of
banks. Parallels with a cartel are apparent. In case of a (pure) three-party model there is no
(explicit) IF as the scheme typically combines and includes the issuing and acquiring
functions.
2.5 Justifications for interchange fees
Whereas the concept of IFs is straightforward, it is the interpretation and justification that
meets scepticism and critique. For a part this is understandable as payment markets are an
instance of two-sided markets and the body of economic knowledge on their functioning is
still evolving. In the earlier literature, economic scholars have gone sometimes to a
considerable extent to emphasise that two-sided markets are nothing exotic but an often
observed and old phenomenon, for example Evans (2003), Rochet and Tirole (2000, 2002,
2005) and Wright (2004a). This has lead others to the remark that therefore no new antitrust
policy should be necessary, at least when dealing with predatory pricing, see e.g. Motta, 2004,
p.452. The main message from these early contributions is that in a two-sided market, it is
perfectly normal and just that one side of the market pays (substantially) more than the other
(instead of each their own ‘fair share’). Consequently, there is no economically justified
reason for a regulator to intervene.
1
The only country I know of where the opposite used to be the case is Australia for eftpos-transactions, a PIN
based debit card scheme (Weiner & Wright, 2005, p. 293 fn.5). However, in 2012 the IF direction was reversed.
19
It should be noted that IFs are not a feature exclusive to payment cards: other retail payment
instruments like cheques, direct debits and typical local instruments like the Dutch
Acceptgiro also carry IFs1. An ATM cash withdrawal by a customer of a different bank than
the one owning the ATM is sometimes mentioned in the literature as another example of an
IF, as it involves a fee from the customer’s bank to the ATM-owning bank. This is however not
a good example as there are only three agents involved: the two banks and a customer of one
of them, but a merchant is missing. In a ‘true’ two-sided market, an issuer (or: bank of the
payer) supplies a service to an acquirer (or: bank of the payee) who supplies a service to the
merchant (payee), but the issuer also supplies a service to the cardholder (payer) to enable
his payment transaction with the merchant. The more merchants accept the payment
instrument (cheques, debit card, credit card, direct debit,…) the higher the benefits of using
it to the payer/cardholder and vice versa. One could therefore reframe the question on the
justification of the IF as: does the issuer get compensation from the acquirer for part of his
internal cost? Or do the issuer and acquirer essentially cooperate to redistribute some of the
benefit the payment instrument usage has to the merchant from the merchant to the
cardholder, in order to induce the latter to use it through a lower fee? This interdependency
and ‘general’ feature adds merit to the argument that IFs are primarily a device to balance a
two-sided market when there is not a single platform to do that implicitly, as for example
with a three-party scheme2.
What complicates matters is that once a line of thought has been chosen, e.g. by a regulator
or the market participants themselves, the line tends to persist. Domestic direct debit
schemes in France, Italy and the Netherlands for example carry an IF but these are strictly
cost based and, as for the Netherlands, set collectively by the Dutch banks based on the per
transaction processing costs of the most efficient debtor bank (issuer). This practise was
1
During the ‘free banking periode’ in the 19th
century, commercial banks could issue their own banknotes and
central banks were typically established to organise their orderly issuance. A central bank issues its own liabilities
for use as ‘central bank money’. Commercial banks too began increasingly to issue liabilities, i.e. ‘commercial
bank money’. A layered structure then emerged, whereby private individuals held their deposits in commercial
banks, and these in turn held theirs in accounts at the central bank. Individuals’ confidence in commercial bank
money lay in the ability of commercial banks to convert their liabilities into liabilities of other banks or central
bank money when requested by their customers. The central banks were in particular responsible for ensuring that
central bank money and commercial bank money could coexist and be ‘interchangeable at par’ (Kokkola, 2010,
p.152).
2
See e.g. Rochet and Tirole, 2003, p.73-76, where they argue that for four-party models, the IF is the only
balancing device available.
20
authorised by the NMa1. For SEPA Direct Debits (SDDs), (M)IFs are prohibited as per
November 2012 for cross-border SDDs and as per February 2017 for domestic SDDs2.
This leads to the question if/when an IF is competitively neutral or socially optimal and
consequently, if there is reason for a regulator to intervene. These questions are addressed
in the review of literature section.
2.6 Some landmarks in interchange fee litigation
Evans and Schmalensee (2005) report the case of National Bancard Corporation (NaBanco)
vs. VISA3, filed in 1979 and decided in 1986. NaBanco claimed the MIF that VISA’s member
banks set, was a per se violation of the Sherman Act. This claim was rejected by the court with
reference to the potential efficiency benefits for a two-sided market. In 1983, William F.
Baxter, then professor of Law at Stanford who had worked on the case, published a paper that
was the first to explain the rationale for interchange fees. After the court’s ruling however “...
interchange fees faded from view in academic and policy circles and was a topic of interest
mainly to industry insiders” (Evans and Schmalensee, 2005, p.75). That changed with the
proliferation of payment cards in the Western world (see charts 1 and 2a below for an
illustration) accompanied with lawsuits and complaints to regulators by retailers who were
confronted with an increase in their costs of receiving payments as a result of the usage
increase.
[charts 1 and 2a about here.]
The high profile case in the US of Wal-Mart, Sears Roebuck, Safeway and a long list of other
merchants versus VISA and MasterCard marks the increase in litigation. The case started in
October 1996 and took seven years of legal combat before it was settled in 20034. In Europe
a similar case was filed in 1997 by EuroCommerce, an association of large European retailers
1
Exemption granted in 2002 for 5 years and since then prolonged. Exemption is based on art. 17 Dutch
Competition law (Mededingingswet) “…promoting technical or economic progress…”, compare art. 101(3)
TFEU.
2
Per Regulation EU 260/2012, with the exception of so-called R-transactions, i.e. returned or rejected DDs. These
IFs too must be “strictly cost based” (art. 8).
3
As do Rochet and Tirole (2000, p.2 fn.9).
4
The case turned into a class-action lawsuit and was finally settled on June 4, 2003 for “the largest antitrust
settlement in history” ($2 billion for VISA and $1 billion for MasterCard). Later that month, the district court
approved the notice of settlement. Eighteen merchants, of around five million in the class, objected to the
settlement; the district court however approved it in January 2004 after a fairness hearing. E.g. [7], 7 Sept. 2014.
21
and national commerce federations, however not before court but with the EC. The difference
in approach between these two quests for relief is noteworthy. An important explanatory
factor is the difference in the implementation of VISA’s and MasterCard’s Honour All Cards
rule (HACR). In the US the two schemes required merchants who accept their credit card to
also accept their (signature-based) debit card, whereas in the EU the merchant is not allowed
to distinguish between issuers of a card brand, that is: the merchant has to accept e.g. all debit
cards but is allowed to refuse the scheme’s credit card. Wal-Mart c.s. argued that the tie-in of
both types of cards, together with other scheme rules was an attempt to monopolize the debit
card market in violation of section 1 of the Sherman Act (anti-cartel rule) and as a
consequence the plaintiffs had been charged excessive interchange fees during October 1992
and June 2003. The courts affirmed this notion. This tie-in of both types of cards has never
been present in the EU1. After the EC’s investigation into VISA’s MIFs and VISA’s offer in 2002
to cap them at the level of relevant costs, VISA’s intra-EEA cross border MIFs were exempted
based on Art. 81(3) of the EC Treaty until the end of 20072. EuroCommerce and First Data, a
cards transactions processor and acquirer, appealed against the exemption. Both eventually
dropped the case but First Data did not do so until VISA had withdrawn another one of their
scheme rules: No Acquiring Without Issuing3. Meanwhile, MasterCard had received a
Statement of Objections (SO) from the EC in October 2003 for the way it sets its MIFs,
basically suggesting MasterCard is acting as a cartel. MasterCard was listed on the NYSE in
May 2006, which led to speculations that this change in its legal structure may have been to
avoid the antitrust allegations, or at least in part; see Schinkel (2010) for an entertaining
account. It nevertheless received a supplementary SO from the EC in June 20064, followed by
its infringement decision in December 20075. MasterCard sought annulment of this decision
but the General Court upheld it in May 2012, including the surprising assessment that
1
See also Weiner & Wright (2005) p.300.
2
OJEC L318, 22.11.2002. VISA will apply flat-rate intra-EEA MIFs before the end of 2002 whose weighted
average will not exceed EUR 0.28 (par. 18) and similarly for credit cards an ad valorem fee of 0.7% before the
end of 2007 (par. 19); exemption granted until 31 December 2007 (par. 109). In short, Art. 81(3) states that certain
restrictions on intra -EU competition may be allowed if the agreement / decision / practice “…contributes to
improving the production or distribution of goods or to promoting technical or economic progress...” provided
certain criteria have been met. Art. 81 of the EC Treaty of 1957 has been renumbered as Art. 101 in TFEU.
3
See e.g. Bos (2007, p.114-115) for more on this case.
4
MEMO/06/260, 30 June 2006. [9], 3 April 2015.
5
Case COMP/34.579, December 19, 2007: the EC views MasterCard’s multilateral intra-EEA IF for cross-border
payment card transactions made with MasterCard (credit card) and Maestro (MasterCard’ debit card brand) a
violation of Art. 81 and the MIFs should therefore be withdrawn within 6 months.
22
MasterCard “… had continued to be an institutionalised form of coordination of the conduct of
the banks…”1. MasterCard’s appeal to the ECJ was rejected in September 20142.
After the expiration of VISA’s exemption and just before the IPO on the NYSE in March 2008
as of when VISA Inc. separated with VISA Europe with the latter continuing to be a
membership organisation, VISA Europe was informed of the EC’s opening of antitrust
proceedings, for which it received an SO in April 2009. Another year later both parties had
come to an agreement: following MasterCard who had introduced caps on its weighted
average cross-border MIFs for debit (0.2%) and credit (0.3%) card transaction in July 2009,
VISA Europe would likewise cap its intra-EEA cross border MIFs for debit cards at a weighted
average of 0.2%; moreover the same cap was agreed to apply to domestic VISA debit card
transactions in nine Member States3. The agreement was made legally binding in December
that year. In May 2013, VISA agreed to adopt the same cap of 0.3% MIF for both its cross-
border intra-EEA and domestic credit card transactions.
As already noted in §1.1, the IFR will take this another step further and apply caps to all four-
party payment card transactions, both cross-border intra-EEA transactions and domestic
transactions, both debit card (0.2%) and credit card (0.3%) transactions, regardless how the
IF is set (multilaterally, bilaterally or otherwise). An exemption is made for pure three-party
schemes, schemes that directly contract both the cardholder and merchant. This may raise
speculations as to whether the four-party schemes will add another chapter to market
oversight games (Schinkel, 2010) and perhaps turn into three-party schemes. As evidenced
from a working document, the EC seems aware of this risk of regulatory circumvention4.
Impact IFR
Had the scope been only cross-border intra-EEA transactions, or only MIFs (as a fallback fee),
then the impact of the IFR on issuing bank’s revenues would not have been that large. Merely
to illustrate this point: most economies are still locally oriented; to take an ‘extreme’: the
1
Case T-111/08, 24 May 2012, par. 259.
2
Case 382/12P, Sept.11, 2014. Note that the decisions relate to MIFs, not bilaterally agreed IFs or IFs set
collectively at national level.
3
See [8], 11 April 2015, for an overview of the EC’s work on MIFs, including links to relevant documents.
4
Commission Services, Working Party on Financial Services, Proposal for a Regulation on interchanges fees for
card based payment transactions, Inclusion of three party card schemes under the scope of Chapter II,
WORKING DOCUMENT #27, MIF, 17 October 2014. This document was shared with payment industry insiders
(such as the author) and may not be available outside the stakeholder group.
23
economy of the Netherlands has a relatively large international orientation, ranking for
example no. #1 in 2014 in the Global Connectedness Index of logistics company DHL1. Still,
only 2.7% of all debit card transactions are cross-border2. The number can reasonably be
taken as an upper limit indication when generalizing to other European economies. However,
with the domestic transactions in scope, the impact of IFR will be much larger. How large
exactly will depend on the actual IF levels in each country, see for an estimation the EC’s
Impact Assessment3. As for the Netherlands, the actual market IF levels are substantially
below the proposed caps, in 2009 between 1 and 2 eurocents per debit card transaction, i.e.
0.04% of an average debit card transaction4. The Dutch government therefore would like the
EC to choose an even lower cap on debit card IFs5.
2.7 A note on SEPA and PSD
The EU and the Member States started a huge endeavour with the integration of the internal
market, including money to transact. First cash (banknotes and coins), then the instruments
to transfer money electronically. However, throughout the years each country had developed
and grown accustomed to their own payment instruments so this proved to be quite a
challenge. The European banking industry created the European Payments Council (EPC) in
2002 to help realise SEPA, a Single Euro Payments Area6. Through its efforts and under its
coordination, the two most commonly used instruments in the EU for electronic payments,
the credit transfer and the direct debit, were redefined in standardising schemes7. Soon
thereafter a scheme for payment cards, the most common alternative to cash, followed. To
dismantle legal barriers between the countries and create a ‘level playing field’, the Payment
Services Directive (PSD) was introduced, taking effect on November 1st, 20098. As already
1
Measured in cross-border traffic of goods, capital, information and people. The Netherlands also ranked #1 in
the previous edition of the Index in 2012. Source: FD (Financieel Dagblad), 5 November 2014.
2
Around 70 million cross-border transactions versus 2.6 billion POS debit transactions in total. Cross-border
includes Dutch cardholders paying abroad and foreigners paying in the Netherlands. Taking credit card
transactions into account, the number is 5.5% (152.6 million cross-border POS transactions versus 2.8 billion in
total). Looking at other retail payments instruments, credit transfers and direct debits, then the numbers are 2.1%
and 0.0% (too low to measure). Source: DNB retail payments statistics 2014.
3
SDW(2013) 288 final, figure 6, p. 21.
4
Average transaction amount with a debit card in 2009 was EUR 39.07. Calculated with an IF of 1.5 eurocents.
5
Fiche 1: Verordening interbancaire vergoedingen, Kamerstuk 22 112, nr. 1705, 4 October 2013
6
The jurisdictional scope of the SEPA Schemes currently consists of the 28 EU Member States plus Iceland,
Norway, Liechtenstein, Switzerland, Monaco and San Marino. Note that this is more than the countries carrying
the Euro as their currency.
7
The SCT (SEPA Credit Transfer) and the SDD (SEPA Direct Debit) were implemented as of 2008 and 2009
respectively.
8
The PSD however does not apply to Switzerland and Monaco. Previously, the jurisdictional ‘misalignment’
between the PSD and SEPA was much larger.
24
noted, the PSD is currently being revised. The proposed changes have caused an intense and
still ongoing debate and lobbying over certain topics. Two are at the heart of the controversy:
1) the introduction into its scope of two new payment services, so-called payment initiation
services and account information services, and the legal provisions regarding the parties
offering these services; and 2) the prohibition of surcharging together with caps on IFs (the
IFR).
Just to briefly illustrate one of the several issues regarding the first controversial topic before
continuing with the second, which is the focalpoint of this paper. The parties offering the
newly introduced services1 will be allowed to gain access to account information of the
consumer and initiate a payment on his behalf using his personal credentials, the same as the
consumer himself uses to log in to his internet/online banking environment. The banks
consider this a major security risk as it may be impossible for them to unambiguously
distinguish the consumer, i.e. their own customer, from a potentially fraudulent party
disguised as the consumer. A contract or agreement between the TPP and the bank will
however not be required2. The concern has even created ‘unnatural allies’ where banks find
regulators3 and consumer’ interest organisations at their side and this has struck a chord
with the EP and several Member States4 but so far not visibly with the EC.
Given the significant reduction in the general IF level across Europe and the expected
consequential reduction in merchant fees, the EC finds surcharging no longer justified for the
regulated payment cards and will therefore prohibit surcharging (PSD2, art. 55(4)). The PSD
left its regulation at the discretion of Member States and this resulted in half the countries
(thirteen) prohibiting it while the other half allowed it5.
1
Called TPPs, Third Party Providers or Third Party PSPs. PSPs are Payment Service Providers, a term introduced
in the PSD for banks and non-banks (“payment institutions”) authorised to provide payment services.
2
ING filed suit against AFAS, a provider of bookkeeping tools for small businesses and individuals, claiming
AFAS seduces ING’ customers to share their personal credentials in breach of ING general terms and conditions.
AFAS used inter alia the upcoming PSD2 in their defence. The court agrees with ING (30 July 2014). [6], 18
August 2014.
3
E.g. BaFin, the German Federal Financial Supervisory Authority; [13], 18 April 2015.
4
For example, France, Denmark and the Netherlands have officially notified the EC of their strong objections
against the provisions requiring the sharing of personal credentials. Austria has even filed a waiver, stating that it
will under no circumstances transpose these provisions into national law.
5
Situation at the time of the PSD2 proposal, July 2013. A few countries have changed their policies since.
Countries that currently allow surcharging include the Netherlands, Denmark and the UK, where regulation
abolished the no-surcharge rule in 1991 (Vickers, 2005, p.233 fn.10). In Denmark, only for international debit and
credit cards surcharging is allowed, but not the domestic Dankort (so-called ‘split model’, see Danmark
Nationalbank, 2012, p.121). For the Netherlands see section 6.
25
Next to these provisions (and a few others), the IFR sets forth in art. 10 the HACR, basically
reconfirming the European interpretation of the HACR (§2.6). It will also enforce a legal and
organisational separation between the scheme and the processing infrastructure.
26
3 REVIEW OF LITERATURE
The cost-based approach, derived from conventional economic wisdom on monopolistic and
oligopolistic ‘one-sided’ market behaviour and benchmark perfect (Cournot and Bertrand)
competition, does not fit a two-sided market. Market authorities and courts using
benchmarks based on issuer’ cost to assess possible excessiveness of merchant charges and
interchange fees, have therefore been strongly and unanimously advised by economists to
readjust their logic. I have borrowed from Wright to summarise in Box 1 some incorrect
applications of ‘one-sided logic’ to two-sided markets. See Wright (2004a) for a discussion of
each fallacy and an entertaining illustration using quotes from investigations into credit card
schemes by three different market authorities.
3.1 Preliminary
At that time however, the economic knowledge on two-sided markets was still rapidly
developing. As Evans and Schmalensee put it in their survey of the economic literature on
interchange fees and their possible regulation in 2005: “Economists have only scratched the
surface of the theoretical and empirical work that will be needed to understand pricing in two-
sided markets in general and the determination of interchange fees in particular” (id. p.104).
The most important conclusions from the survey can be summarised as follows. Next to the
factors that determine socially optimal prices for customer groups in multisided industries,
inter alia a) price elasticities of demand, b) indirect network effects between the customer
groups, and c) marginal costs for providing goods to each group, socially optimal prices in the
payment card industry also depend on other characteristics, including d) the use of fixed and
variable fees, e) competitive conditions among merchants, issuers, and acquirers, and finally
f) the nature of competition from cash, cheques, and three party payment schemes. Thus, the
socially optimal interchange fee is not, in general, equal to any interchange fee based on cost
BOX 1. Eight fallacies - lessons to forget in two-sided markets.
1. An efficient price structure should be set to reflect relative costs (user-pays).
2. A high price-cost margin indicates market power.
3. A price below marginal cost indicates predation.
4. An increase in competition necessarily results in a more efficient structure of prices.
5. An increase in competition necessarily results in a more balanced price structure.
6. In mature markets (or networks), price structures that do not reflect costs are no longer justified.
7. Where one side of a two-sided market receives services below marginal cost, it must be receiving a
cross-subsidy from users on the other side.
8. Regulating prices set by a platform in a two-sided market is competitively neutral.
27
considerations alone and such a solely cost based interchange fee is unlikely to improve social
welfare (id. p.102).
In my selection I have mainly focussed on contributions with formal models that allow for a
welfare analysis, either total surplus or user surplus1. I describe these models in non-
technical terms.
3.2 Models of interchange fees and policy implications
In Baxter’s (1983) model, cardholders use and merchants accept a card payment if the per-
transaction price charged to each of them is less than the per-transaction benefit they derive
from it. Each of their marginal valuation of the payment transaction depends on the other
party accepting, respectively using the card (otherwise the valuation is zero). Cardholders
are assumed to be more price sensitive than merchants and so their costs, that is the price
they pay for the transaction to the issuer, need to be lowered to reach equilibrium. For this, a
transfer – the IF – needs to be made from the acquirers to the issuers who both are assumed
to behave (perfectly) competitively. Therefore the IF does not affect the overall price level (it
could be at any level), only the price structure. In absence of bargaining power, the price the
issuers charge the cardholders is not based on their marginal costs, nor is the price the
acquirers charge the merchants based on their marginal costs (which is what ‘one-sided logic’
would suggest); instead, in equilibrium aggregate joint demand for card payment services of
merchants and cardholders equals the total combined issuers and acquirers costs of
providing them. In this perfectly competitive world, an assumption Baxter acknowledges may
not hold in reality, without e.g. fixed costs or membership fees, the equilibrium is efficient
(the model is not equipped for a welfare analysis). Baxter concludes his article stating that
the characteristics of the payments market, i.e. joint costs and interdependent demand, are
not well understood and the controversy that troubled the US banking industry for more than
five decades around ‘clearance at par’ of cheques2 is likely to repeat in the context of debit
and credit cards (id. p.586). Those words proved prophetic. In light of this anticipated
controversy, Baxter warns that governmental intervention should be resisted. See Box 2 for
a sidestep to Baxter’s account of payment instruments and substitutes.
1
Total welfare, or total surplus, is the sum of consumer surplus and producer surplus. In the context of payment
markets, the producer is (are) the banks and other providers in the chain and the consumer is (are) both the
payer/cardholder and the payee/merchant, both are “users” of the card. Consumer surplus, or consumer welfare,
thus involves simultaneously both users.
2
Baxter uses the US spelling: checks.
28
Schmalensee (2001) allows for imperfect competition among issuers and among acquirers.
He focuses on credit cards and on a four-party, cooperative (not-for-profit) scheme that sets
a MIF. Assuming the scheme is facing a multiplicative demand function and conducting a
welfare analysis, first with a monopolistic issuer and a monopolistic acquirer which he next
generalizes to oligopolistic competition, Schmalensee arrives at similar conclusions as
Baxter: privately optimal IF is also socially optimal and the IF depends on demand conditions,
costs, competition among issuers and acquirers and on externalities between merchants and
cardholders. He sees no cause for a regulator to step in.
BOX 2. On payment means and substitutes.
Baxter’s historical account of US four-party payment systems covering roughly 200 years narrates the
growing popularity of cheques (initially: drafts) and currency (initially: bank notes) as markets developed
from very local to increasingly larger geographies, while inferior country bank notes were driving sounder
city bank notes out of circulation and transporting bank notes was increasingly costly and risky. This
explains the heavy usage of cheques for which later the credit card became the prime substitute. By
contrast, cheques have never gained comparable popularity in other countries such as Sweden, Spain,
Denmark, and the Netherlands (revisit chart 1). Baxter speculates that the ‘clearance at par’ of cheques,
before Federal regulation put an end to this heterogeneous practise by imposing an ‘IF’ of zero, was a result
of a shift in relative demands of purchasers (~cardholders) and merchants for cheque services and a shift
in relative costs in providing them. It is also interesting to note that Baxter treats cheques and credit cards
as main substitutes (thus cheques’ costs are the most appropriate benchmark for credit cards’ costs,
including IFs) whereas in Europe, economists and the EC focus more on cash versus debit cards. Though
these may perhaps be better comparable substitutes , the comparison is not without difficulties as cash
usage is very difficult to measure reliably, comes with “interchange fees at par” and nowadays involves a
monopolist supplier (i.e. the central bank, see §2.5).
Note 1: Although Baxter occasionally mentions debit cards simultaneously with credit cards, the focus of
his article is largely on the latter. Indeed, the debit card is a “poor man’s card” as “…any cardholder entitled
to use a credit card, will always use it rather than a debit card” (p.585), because of the float benefits
attached to the credit card. For this reason, and because of a lower risk of default, Baxter predicts that debit
card transactions will be substantially cheaper than credit card transactions and with different IFs. This
prediction is confirmed in Weiner & Wright (2005) who report credit card IFs typically between 1% to 2%
of transaction value and debit card IFs typically between 0% and 1% for a number of regions and countries
(id. tables 1 – 3).
Note 2: Payments practitioners frequently use a rule of thumb categorisation of payment instruments from
a payer’s point of view: Pay Before, Pay Now and Pay Later. Cheques and credit cards are examples of Pay
Later instruments, while cash and debit cards are Pay Now instruments. Prepaid cards like a gift card or
special purpose card, are often considered Pay Before instruments as it involves a transfer of funds (from
the cardholder’s account to the card or to the administrator of the card) while the goods or service will be
delivered somewhere in the future. Because the card usually cannot be used everywhere and for all
purposes, like cash, consumers tend to view their funds have ‘changed currency’. Examples are the Dutch
Public Transport card (OV-chipkaart) and, according to most of its (former) users, the Chipknip.
29
Rochet and Tirole (2000, published in 2002) introduce in this seminal paper a formal
framework upon which most of the later scholarly contributions build. Like Baxter, they take
into account the (direct) benefits of using and accepting a card payment. However, they also
model merchants and cardholders as strategic players: merchants use the acceptance of
cards to generate higher sales revenue by winning consumers (cardholders) from
competitors and consumers may decide to visit stores based on stores’ card acceptance
policy. Thus merchant resistance to increases in IF is likely to be lower than in Baxter’s model.
Their framework also allows for consumers to hold cards of more than one scheme1, making
merchants’ opportunity costs of card acceptance endogenous. Acknowledging that the
schemes compete for cardholders, merchant resistance to increases in IF may in this respect
be higher than in Baxter’s model. Furthermore, acquirers are assumed to behave
competitively while issuers may enjoy some market power2. Cardholders are assumed to
have a fixed volume of transactions3, which (technically) implies that there is no difference
between an annual fixed fee or a variable per-transaction fee, at least not from the issuer-
cardholder relationship perspective4. Cardholders are modelled to have structural
preferences for using cash or cards. The model makes no specific distinction between debit
and credit cards as it focusses on the payment activity rather than the instrument. Merchants
face no fixed costs for accepting payments. Given this model setup where by assumption all
profits fall at the issuing side of the market, Rochet and Tirole show by applying a total
welfare analysis that the socially optimal IF, which is the one where the total cost of the
marginal transaction equals its total benefit, coincides with the privately set IF (as set by the
scheme, i.e. the issuers) if that IF exceeds the level at which merchants accept the card. This
requires a low cardholder fee. Or the privately set IF exceeds the socially optimal IF, in which
case consumer fees are set too low leading to an overconsumption (overusage) of cards5. No
cost-based regulatory intervention can prevent this. Both equilibria apply under the no-
surcharge rule, a scheme rule that prohibits the merchant to price-discriminate between
payment means (e.g. demand an extra fee from the customer for a card payment). Lifting this
rule creates however ambiguous welfare effects6.
1
This is sometimes also referred to as multi-homing (vs. single-homing).
2
As Rochet and Tirole note on p.5 fn.13, this is closer to reality. As an indication thereof, they refer to the voting
rights of the banks in the US in VISA and MasterCard (before their respective IPOs) which are more sensitive to
issuing than to acquiring volume, suggesting some bargaining power is on that side.
3
In their model this is normalized to one transaction for each customer.
4
For their main model, cardholders are assumed to be charged a fixed annual fee.
5
This situation can be found in countries where regulation prevents banks from charging customers for the use of
cheques, thus leading to an overprovisioning of cheques (Rochet & Tirole, 2000, p.17 fn.23).
6
Rochet & Tirole, 2000, p. 18-20.
30
The main result, including its corollary that there is no equilibrium where a privately set IF
is lower than the socially optimal IF, critically depends on an assumed merchants
homogeneity. Relaxing this assumption can lead to an underprovisioning of cards, depending
on how well informed the cardholder ex ante is of a merchant’s card acceptance.
Interestingly, the way cardholders are charged for card usage does matter as it influences
merchant resistance: if the cardholder would be charged a perfect fixed and variable fee, with
marginal cost pricing for the variable fee, then this would reduce merchant resistance if (and
only if) the IF exceeds the issuer cost. Indeed, the main mechanism at work in this model is
merchant resistance to accept cards. A higher resistance is likely to bring an equilibrium
where private and socially optimal IF coincide; a lower resistance is likely to create an
overconsumption of cards with a higher than socially optimal IF. The question whether the
IF is ‘too high’ is left an empirical one.
Gans and King (2001b, published in 2003) show in a general way that if merchants can
costlessly surcharge, then interchange fees will have no real effects irrespective of the level
of competition at either the banks or the merchants.
Following closely Rochet and Tirole’s main model, Wright (2004b) relaxes the assumption
of identical merchants and introduces heterogeneity on both sides of the market, applying a
standard Hotelling model of competition (of ‘linear cities’). This accounts for the fact that in
some sectors accepting card payments may be more beneficial to merchants than in others.
Cardholders are assumed to pay a per-transaction fee as do merchants and neither pays a
fixed fee or faces fixed costs. Cardholders are supposed to be fully informed about merchant’s
card acceptance policy before they frequent the store, which maximises the merchant’s
incentive to accept cards. Cardholders discover their preference for a cash or card payment
at the moment of purchase of the good1. The results show that the privately set IF may or may
not be equal to the socially optimal IF and either one can be higher than the other. Thus there
may be too many or too few card transactions from a socially optimal point of view. No cost-
based regulation would be able to restore balance in case of a troublesome diversion as it
depends on differences in price elasticities on both sides of the market, competition (among
issuers, among acquirers, among merchants and among schemes) as well as costs. Again, this
leaves the question of a possible excessive IF an empirical one.
1
This makes merchants’ card acceptance decisions independent rather than strategic complements as in Rochet
& Tirole (2000)
31
By this time the literature had rapidly build into several directions focussing on specific
topics such as platform competition1, multiple card membership (multi-homing) and usage,
market two-sidedness and the influence of merchants and cardholders entering into a
Coasean negotiation as to set their own fee directly. Rochet and Tirole (2005) integrate in
particular the findings on multiple membership and multiple usage (in the broader context
of two-sided markets, not only card payments industry) and reinterpret some of the
previously obtained results. In summary: 1) pricing in two-sided market obeys standard
Lerner principles2 with a reinterpretation of marginal costs as ‘opportunity costs’3; 2) a
market is two-sided if the price structure matters (and much less so the price level),
measured by the ability to affect the volume of transactions; however, in the absence of
membership externalities, the market can turn one-sided in the presence of asymmetric
information between the card-users (merchants and cardholders) if the transaction between
them involves a bilaterally negotiated price or monopoly price. A market turns two-sided
when there are transaction costs to the bilateral price negotiation or constraints on this kind
of price-setting (like an imposed no-surcharge rule) or when there are membership fixed fees
(fixed costs).
This latter finding is one found in McAndrews and Wang (2008) who develop a formal
model different from the ones described above in that they 1) ignore benefits consumers
derive from using a payment instrument (the instrument imposes a frictional cost to the
purchase of a good), 2) assume a contestable market for merchants, which simplifies the
welfare analysis, and 3) take merchants and consumers as non-strategic players (as Baxter).
They do take both fixed and variable costs into account. Starting with cash as a benchmark,
they analyse the choices of merchants and consumers when offered a payment instrument
such as a card that comes at a higher fixed cost but offers lower variable costs. Consistent
with empirical studies (id.), they find that large merchants adopt payment cards faster than
smaller merchants and will set a price that is lower than cash customers would experience at
only-cash accepting merchants; smaller merchants on the other hand may accept card
payments or not and those who do, set a price higher than the competing only-cash accepting
1
See in particular Rochet & Tirole (2002) and Guthrie & Wright (2007).
2
The price charged to a side of the market is inversely related to that side’s elasticity of demand.
3
Marginal cost c is in the context of a two-sided market with a platform replaced by (c-vj), i.e. platform cost c per
transaction minus vj, the ‘other’ side j’s willingness to pay to interact with ‘this’ side. See for an accessible account
Tirole’s lecture in accepting the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2014;
[10], 18 January 2015.
32
merchants; finally, the small merchants will not accept cards. The paper adds the insight that
if a merchant serves both cash-preferring and card-preferring consumers, the first group is
facing a lower selling price than they would if the store only accepted cash. This insight
contradicts the findings and intuitions that price coherence1, or the inability to surcharge,
leads to higher selling prices.
In another milestone contribution, Rochet and Tirole (2008) respond to Vickers’ (2005)
“must take card” argument. The then head of UK’s Office of Fair Trading2, Vickers posed that
in the UK retail business it had become a market practise to accept at least the two major
credit card brands (VISA and MasterCard) and non-compliance with this market standard
could jeopardise the retailer’s business as he would risk losing customers to competing
retailers who do. It should be noted that the use of credit cards (and cheques) is very high in
the UK compared to other countries in Europe. See also chart 1 and as a further exhibit:
“Indeed, the United Kingdom has accounted for more than 75 percent of credit card spending in
western Europe in recent years. By contrast, France and Germany each account for less than 1.5
percent” (Vickers, 2005, p.232). Rochet and Tirole operationalise and validate the argument
under different models of merchant competition building on earlier frameworks, in
particular the ones described above. The possibility that a merchant might accept a card
payment even though this would increase his net operating cost3, was already identified in
their earlier paper (2000). The reason for the merchant’s decision lies however not in
competition, i.e. to win over a customer over a competitor: the property holds even when the
merchant is a local monopolist. It lies in the merchant’s improved quality of service by
offering the customer extra payment means, which translate into slightly higher retail prices4.
The paper presents an alternative to the benchmark for excessive IF-setting used by
regulators, which is based on issuer’ costs5. Their alternative is based on the merchant’s
1
A term coined by Frankel (1998) and later adopted by others, which he defines as: “the phenomenon in which
the price paid by a consumer for a product does not vary with modest differences in the costs imposed on the
merchant by the customer’s choice of brands or payment methods” (p.314).
2
The OFT was at that time involved in a several-year-long investigation into MasterCard’s MIFs.
3
The property that has become known as merchant internalization states that merchants accept cards if, and only
if, the merchant fee 𝑝 𝑀 is equal to or less than the sum of the merchant benefit of accepting a card payment 𝑏 𝑀
plus the average net cardholder benefit per card payment 𝑣 𝐶(𝑝 𝐶): 𝑝 𝑀 ≤ 𝑏 𝑀 + 𝑣 𝐶(𝑝 𝐶) This can lead to merchants
accepting card payments even when it increases their net operating costs: 𝑝 𝑀 ≥ 𝑏 𝑀 Rochet & Tirole show that
this property holds under three important models of competition: perfect competition, Hotelling-Lerner-Salop
differentiated products competition and monopolistic competition.
4
This result is also derived in Bedre-Defolie and Calvano (2009).
5
See e.g. the EC COMP/34.579 in 2007 (as mentioned in par. 1.1), the Reserve Bank of Australia in 2003 (Wright
2004a) and the US Federal Reserve by the Durbin Amendment (section 1075, adding section 920 to the Electronic
Funds Transfer Act) to the Dodd-Frank Act as of July 2011.
33
avoided-cost: would he have an incentive to refuse a card payment if an incidental customer,
like a tourist, wants to pay and is also able to pay in cash? (hence the nickname the “tourist-
test”). Since accepting this card transaction would trigger a fee to his acquirer1 - a significant
portion thereof constitutes the IF - the test looks for the merchant’s point of indifference
between the two means of payment. The merchant fee passes the avoided-cost test if, and
only if, the card payment does not increase his net operating cost compared to accepting a
payment in cash. If it does not pass the test, then this should be interpreted as an indication
of an excessive IF. Rochet and Tirole show that the test yields unbiased results if the objective
is short-term total user surplus2, provided that issuers’ margin is constant. However, the IF
that would be optimal from a social point of view lies higher in that case and the same holds
in the case of variable issuers’ margin. The test yields false positives if the objective is total
welfare. Zenger (2011) shows the avoided-cost test and the model of Rochet and Tirole
(2000) under perfect surcharging yield the same outcomes.
Discussion of the avoided-cost test
Before moving on to the last piece of review of literature in this section, I briefly discuss the
test and its implications in some more depth. The avoided-cost test seems reasonable at first
sight but on second thought appears problematic. The test makes no distinction between a
debit and credit card transaction, while the proposed benchmark is cash. As noted before,
from a payer’s point of view, cash and debit cards are more or less substitutable payment
instruments, but credit cards and cash are much less of substitutes3. More importantly, as the
test looks for the merchant’s point of indifference between the means of payment, it thus
removes incentives to choose the most efficient one from a social point of view4. If the aim of
1
The fee paid by the merchant to the acquirer is usually called the Merchant Service Charge in the credit card
business (MasterCard and others). Rochet & Tirole have consistently used the term “merchant discount”,
indicating that the merchant does not receive the full amount, i.e. the sales price for the goods sold, from the
cardholder (through the issuer and acquirer) but that the acquirer subtracts (“discounts”) a fee from this amount.
From the merchant’ point of view however, he does not get a “discount” in the meaning of common language. To
avoid confusion, I use the more neutral term “merchant fee” throughout this paper. The practise to actually subtract
a fee from the full amount was abolished by the PSD, as of then the merchant can still be charged such a fee but
it has to be invoiced and paid separately.
2
As indicated before, in a two-sided payment market, the ‘one-sided’ consumer surplus includes both the
cardholder’s and merchants’ surplus. I will use the term “total user surplus” from this point onwards, as do Rochet
and Tirole. The term “consumer surplus” is then reserved for the card and cash using consumers (their surplus),
i.e. ex. merchant surplus. In their (2008) model, all consumers hold cards, so there is no distinction between a
consumer paying in cash and a consumer paying with a card in the context of welfare analysis. The “short term”
reflects a situation without market entry.
3
A similar remark is made by Rochet & Wright (2009) who define the avoided-cost test with respect to “store
credit” (credit supplied directly by the merchant to the consumer) instead of cash.
4
A point Rochet and Tirole are aware of (id. p.8).
34
a regulator is to correct a market that is failing in producing economic efficient outcomes,
then adopting this test will at best result in maintaining the status quo if the benchmark, cash,
is the correct one. But empirical studies show that cash payments are relatively expensive to
‘produce’ compared to electronic payments, debit card payments in particular, and the price
to use cash does often not reflect the full (social) costs; see Schmiedel c.s. (2012), Brits and
Winder (2005), Leinonen (2011) and Humphrey, Willesson, Lindblom and Bergendahl
(2003). Indeed, putting the avoided-cost test to an empirical test, Bolt, Jonker and Plooij
(2013) show that it creates the wrong incentives when a country’ social costs of cash
payments are rising and those of debit card payments are declining; as is the case for e.g. the
Netherlands1. These findings are particularly relevant since the EC accepted the tourist-test
as a “reasonable benchmark for assessing a MIF level that generates benefits to merchants and
final consumers”2 and the caps proposed in the IFR are the result of MasterCard’s calculations
using the test and three empirical cost studies3 which MasterCard implemented with the EC’s
consent as of July 2009.
A final remark in this sidestep concerns the welfare objective. Most economists, and I agree,
argue that competition authorities should adopt a total welfare standard instead of a
consumer welfare standard. See e.g. Motta (2004, par. 1.3) and Rochet & Tirole (2003, p.77).
Whether courts and regulators favour one over the other is difficult to say, wording such as
in art. 101(3) TFEU4 merely seem to induce its explicit inclusion in court’s rulings, not
prioritising consumer welfare over total welfare. However, as evidenced by the adoption of
the avoided-cost test and the quoted memo, in the case of the IFR the EC seems to adopt a
strict consumer welfare standard5. Such a standard is unlikely to solve a market failure, more
likely the opposite (see also Rochet and Wright, 2009). In particular, capping the IFs may
initially lead to lower prices but in the longer run are likely to deprive the ‘producers’ of
1
Bolt cs. calculate that the tourist-test may allow the IF level to increase from 0.2% to 0.5% of the transaction
amount of an average debit card payment. In case of a full pass-through this would increase the notional merchant
fee by 233%. For other debit card favouring countries similar results should be expected, see also Leinonen (2011)
and Danmark Nationalbank (2011).
2
Press release by the EC, Commissioner Kroes, 1 April 2009, memo 09/143.
3
These concern the cost studies of the central banks of Belgium, Sweden and the Netherlands. The latter is the
study published by Brits and Winder (2005), with data from 2002. The Belgian and Swedish study contain data
from 2003 and 2002 respectively. See also Bolt c.s. (2013).
4
Certain restrictions on intra-EU competition may be allowed if the agreement / decision / practice “… contributes
to improving the production or distribution of goods or to promoting technical or economic progress, while
allowing consumers a fair share of the resulting benefit...” [underline font added].
5
Another indication can be found in the EC’s Impact Assessment: “Rochet and Tirole (2002) find that the
privately set interchange fee either is socially optimal (but only total welfare has been analysed)…” (p.102)
[underline font added]. Also Rochet and Wright (2009) have noticed and seem to disapprove: “Thus, if regulators
only care about (short-run) consumer surplus, our theory can provide a rationalization for placing a cap on
interchange fees.” (id. p.6).
35
incentives to innovate. Note that the issuer and acquirer function can be fulfilled by others
than banks and the incentives concern both incumbents and potential new entrants, so this
may negatively affect banks, other PSPs, merchants as well as consumers. Thus, at the very
least the consumer welfare objective should be framed in dynamic terms. Evans (2011)
analyses the relationship of IF levels and innovation and argues that drastic reductions of IFs
may invert the skewness of the price structure on the two sides of the market, resulting in a
reduction in the overall level of innovation in the industry and a discouragement of new
entrants1. Circumstantial evidence of this point can be found in Australia for eftpos-
transactions, a PIN based debit card scheme and as far as I am aware, the only country where
IFs used to go from the issuer to the acquirer. In 2012 however, the IF direction reversed “…to
encourage investment in innovation and enhanced functionality for eftpos, so that it can
continue to compete effectively in a fast changing payments landscape”2.
Whereas Rochet and Tirole confirm some validity of Vickers’ “must take card” argument but
conclude that this need not be harmful to social welfare as there is no systematic bias in the
price structure, Wright (2012) now arrives at the opposite conclusion. Using basically the
same model as Rochet and Tirole (2000, 2008) and Wright (2004b), he finds that a profit
maximizing card scheme sets a MIF or price structure that will lead to an overprovisioning
(overconsumption) of cards with merchants paying too high a fee. This possibility is present
in the earlier work and arises from the fact that a monopolist scheme only focuses on
marginal users and not on average users, thus ignoring the effect of IFs on the surplus of the
average merchant or consumer. As long as this happens on both sides of the market, there is
no particular bias either way. But with the property of merchants internalizing the consumer
benefits (meaning they offer their customers an improved quality of service with extra
payment means) for which the extra costs are uniformly included in the prices of the goods
sold (and in the absence of surcharging, i.e. price coherence), this introduces a systematic
bias in favour of cardholders. Therefore, Wright advises a regulatory intervention, not based
on any antitrust considerations since the bias is not the result of any shortcomings in
competitive market behaviour, including the argument that an IF puts a floor in what
competing acquirers can charge merchants. Such an intervention is to be dealt with
cautiously, he warns: the main mechanism at work in the model is merchant internalization
1
As Evans (2011) summarises it: “getting innovation right is likely to be far more important than getting
prices right” (p.2).
2
[11], 5 September 2014.
36
but to which extent this holds in a particular case, is an unanswered empirical question. As
to the proper level of IF, he stresses that economists have reached near unanimity against a
fee based on issuer’ costs and instead proposes a direct cap on merchant fees (which should
include three-party schemes) or to apply the merchant’ avoided-cost test. This I already
discussed.
As far as I know, Korsgaard (2014) is the first to discuss interchange fees in a setting that
resembles more that of a debit card payments market. As I use this model, more details are
given in the next section. Using basically the same model as Rochet and Tirole (2000, 2008)
and Wright (2004b, 2012), he shows that the level of the IF only influences merchant
acceptance of cards (the higher the former, the lower the latter) and banks will set an IF that
exceeds or equals the socially optimal IF. Two critical assumptions underlie this result: 1) all
consumers already possess a payment card and do not incur fees per transaction (cash nor
card), 2) merchants nor banks face fixed costs. In essence, the two-sided market then turns
one-sided; this echoes a finding earlier stated, see Rochet and Tirole (2005). As the only
condition under which the banks set an IF equal to the socially optimal IF, is when merchants
are assumed to be homogeneous (else the banks set an IF that is higher), Korsgaard advocates
a cost-based regulation. His model features consumers, merchants and banks; the latter in
their capacity of producers of payment services1. Issuing and acquiring banks are not
distinguished individually, which is motivated by the assumption that acquiring banks are
assumed to behave perfectly competitively and so the merchant fee consists only of the IF
plus the acquiring bank’s marginal cost of producing the card payment2. Banks are assumed
to maximise their joint profit, which depends on the fraction of consumers, respectively
merchants using and accepting cards. Both consumers and merchants are heterogeneous and
they may face fixed costs for accepting payments.
A striking result of this model is that the socially optimal IF can turn out to depend solely on
costs. More precise, if merchant fixed costs are assumed to be zero, consumers face no
adoption costs and in the absence of surcharging, then the socially optimal IF equals the
difference in bank’s marginal costs of producing card payments and cash payments. Under
surcharging however, the welfare outcomes are ambiguous and the optimal IF no longer
depends on costs alone (same is true when fixed costs are introduced).
1
From a modelling perspective this is akin to a proprietary platform, i.e. a three-party scheme.
2
The merchant fee and the IF are therefore assumed to have an one-to-one relationship.
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final
Thesis DSF Kalina final

More Related Content

Viewers also liked

Thesis: The Information Content Of Credit Ratings
Thesis: The Information Content Of Credit RatingsThesis: The Information Content Of Credit Ratings
Thesis: The Information Content Of Credit RatingsVincentvanMeeuwen
 
Credit Card
Credit CardCredit Card
Credit Cardvinueg
 
Credit card special preference to hdfc bank
Credit card special preference to hdfc bankCredit card special preference to hdfc bank
Credit card special preference to hdfc bankshweta bhosale
 
43888714 plastic-money-full-project
43888714 plastic-money-full-project43888714 plastic-money-full-project
43888714 plastic-money-full-projectJeet Yadavv
 

Viewers also liked (6)

Thesis: The Information Content Of Credit Ratings
Thesis: The Information Content Of Credit RatingsThesis: The Information Content Of Credit Ratings
Thesis: The Information Content Of Credit Ratings
 
Credit Card
Credit CardCredit Card
Credit Card
 
Credit Card Usage Study
Credit Card Usage StudyCredit Card Usage Study
Credit Card Usage Study
 
Credit card special preference to hdfc bank
Credit card special preference to hdfc bankCredit card special preference to hdfc bank
Credit card special preference to hdfc bank
 
43888714 plastic-money-full-project
43888714 plastic-money-full-project43888714 plastic-money-full-project
43888714 plastic-money-full-project
 
Credit Card
Credit Card Credit Card
Credit Card
 

Similar to Thesis DSF Kalina final

Esrc policy forum Big Society and public procurement
Esrc policy forum Big Society and public procurementEsrc policy forum Big Society and public procurement
Esrc policy forum Big Society and public procurementTim Curtis
 
Central bank-digital-currency-opportunities-challenges-and-design
Central bank-digital-currency-opportunities-challenges-and-designCentral bank-digital-currency-opportunities-challenges-and-design
Central bank-digital-currency-opportunities-challenges-and-designRein Mahatma
 
SMi Group's 14th annual Nordic Cards conference & exhibition
SMi Group's 14th annual Nordic Cards conference & exhibitionSMi Group's 14th annual Nordic Cards conference & exhibition
SMi Group's 14th annual Nordic Cards conference & exhibitionDale Butler
 
How do Dutch consumers pay in 2020 - trends and scenarios
How do Dutch consumers pay in 2020 - trends and scenariosHow do Dutch consumers pay in 2020 - trends and scenarios
How do Dutch consumers pay in 2020 - trends and scenariosInnopay
 
Current State of Crowdfunding in Europe - review 2016
Current State of Crowdfunding in Europe - review 2016Current State of Crowdfunding in Europe - review 2016
Current State of Crowdfunding in Europe - review 2016MichalGromek
 
European E-Commerce Alternative Payment Providers Gaining Momentum
European E-Commerce Alternative Payment Providers Gaining MomentumEuropean E-Commerce Alternative Payment Providers Gaining Momentum
European E-Commerce Alternative Payment Providers Gaining Momentummercatoradvisory
 
Examining Country: Specific Regulations Related to Money Transfer
Examining Country: Specific Regulations Related to Money TransferExamining Country: Specific Regulations Related to Money Transfer
Examining Country: Specific Regulations Related to Money TransferArief Gunawan
 
The european union payments landscape in perspective
The european union payments landscape in perspectiveThe european union payments landscape in perspective
The european union payments landscape in perspectivePaperjam_redaction
 
The adoption of e-invoicing in public procurement - Guidance paper for eu pub...
The adoption of e-invoicing in public procurement - Guidance paper for eu pub...The adoption of e-invoicing in public procurement - Guidance paper for eu pub...
The adoption of e-invoicing in public procurement - Guidance paper for eu pub...Andrea Caccia
 
Euro-FEM Module 6 : Financial Planning
Euro-FEM Module 6 : Financial PlanningEuro-FEM Module 6 : Financial Planning
Euro-FEM Module 6 : Financial PlanningAthanasiaIoannidou
 
How To Put Sources In Your Essay. Online assignment writing service.
How To Put Sources In Your Essay. Online assignment writing service.How To Put Sources In Your Essay. Online assignment writing service.
How To Put Sources In Your Essay. Online assignment writing service.Lisa Long
 
College of Europe Student Case Study Final Report
College of Europe Student Case Study Final ReportCollege of Europe Student Case Study Final Report
College of Europe Student Case Study Final ReportIlenia Ventroni
 
Ce fintech-in-cee-region-2016
Ce fintech-in-cee-region-2016Ce fintech-in-cee-region-2016
Ce fintech-in-cee-region-2016SRI HARSHA JETTI
 
Aktifitas Bank Sentral di Blockchain
Aktifitas Bank Sentral di BlockchainAktifitas Bank Sentral di Blockchain
Aktifitas Bank Sentral di BlockchainRein Mahatma
 
Authors Tutor ExaminerSubject Level and semest.docx
Authors Tutor ExaminerSubject Level and semest.docxAuthors Tutor ExaminerSubject Level and semest.docx
Authors Tutor ExaminerSubject Level and semest.docxikirkton
 
Galobal Payments Raising Wave -WP
Galobal Payments Raising Wave -WPGalobal Payments Raising Wave -WP
Galobal Payments Raising Wave -WPRamadas MV
 
Exploring Open Finance .pdf
Exploring Open Finance .pdfExploring Open Finance .pdf
Exploring Open Finance .pdfSoodayJhaveri2
 
How to pay for news?
How to pay for news?How to pay for news?
How to pay for news?sndrspk
 

Similar to Thesis DSF Kalina final (20)

CASE Network Report 106 - The Costs of VAT: A Review of the Literature
CASE Network Report 106 - The Costs of VAT: A Review of the LiteratureCASE Network Report 106 - The Costs of VAT: A Review of the Literature
CASE Network Report 106 - The Costs of VAT: A Review of the Literature
 
Esrc policy forum Big Society and public procurement
Esrc policy forum Big Society and public procurementEsrc policy forum Big Society and public procurement
Esrc policy forum Big Society and public procurement
 
Central bank-digital-currency-opportunities-challenges-and-design
Central bank-digital-currency-opportunities-challenges-and-designCentral bank-digital-currency-opportunities-challenges-and-design
Central bank-digital-currency-opportunities-challenges-and-design
 
SMi Group's 14th annual Nordic Cards conference & exhibition
SMi Group's 14th annual Nordic Cards conference & exhibitionSMi Group's 14th annual Nordic Cards conference & exhibition
SMi Group's 14th annual Nordic Cards conference & exhibition
 
1878940
18789401878940
1878940
 
How do Dutch consumers pay in 2020 - trends and scenarios
How do Dutch consumers pay in 2020 - trends and scenariosHow do Dutch consumers pay in 2020 - trends and scenarios
How do Dutch consumers pay in 2020 - trends and scenarios
 
Current State of Crowdfunding in Europe - review 2016
Current State of Crowdfunding in Europe - review 2016Current State of Crowdfunding in Europe - review 2016
Current State of Crowdfunding in Europe - review 2016
 
European E-Commerce Alternative Payment Providers Gaining Momentum
European E-Commerce Alternative Payment Providers Gaining MomentumEuropean E-Commerce Alternative Payment Providers Gaining Momentum
European E-Commerce Alternative Payment Providers Gaining Momentum
 
Examining Country: Specific Regulations Related to Money Transfer
Examining Country: Specific Regulations Related to Money TransferExamining Country: Specific Regulations Related to Money Transfer
Examining Country: Specific Regulations Related to Money Transfer
 
The european union payments landscape in perspective
The european union payments landscape in perspectiveThe european union payments landscape in perspective
The european union payments landscape in perspective
 
The adoption of e-invoicing in public procurement - Guidance paper for eu pub...
The adoption of e-invoicing in public procurement - Guidance paper for eu pub...The adoption of e-invoicing in public procurement - Guidance paper for eu pub...
The adoption of e-invoicing in public procurement - Guidance paper for eu pub...
 
Euro-FEM Module 6 : Financial Planning
Euro-FEM Module 6 : Financial PlanningEuro-FEM Module 6 : Financial Planning
Euro-FEM Module 6 : Financial Planning
 
How To Put Sources In Your Essay. Online assignment writing service.
How To Put Sources In Your Essay. Online assignment writing service.How To Put Sources In Your Essay. Online assignment writing service.
How To Put Sources In Your Essay. Online assignment writing service.
 
College of Europe Student Case Study Final Report
College of Europe Student Case Study Final ReportCollege of Europe Student Case Study Final Report
College of Europe Student Case Study Final Report
 
Ce fintech-in-cee-region-2016
Ce fintech-in-cee-region-2016Ce fintech-in-cee-region-2016
Ce fintech-in-cee-region-2016
 
Aktifitas Bank Sentral di Blockchain
Aktifitas Bank Sentral di BlockchainAktifitas Bank Sentral di Blockchain
Aktifitas Bank Sentral di Blockchain
 
Authors Tutor ExaminerSubject Level and semest.docx
Authors Tutor ExaminerSubject Level and semest.docxAuthors Tutor ExaminerSubject Level and semest.docx
Authors Tutor ExaminerSubject Level and semest.docx
 
Galobal Payments Raising Wave -WP
Galobal Payments Raising Wave -WPGalobal Payments Raising Wave -WP
Galobal Payments Raising Wave -WP
 
Exploring Open Finance .pdf
Exploring Open Finance .pdfExploring Open Finance .pdf
Exploring Open Finance .pdf
 
How to pay for news?
How to pay for news?How to pay for news?
How to pay for news?
 

Thesis DSF Kalina final

  • 1. Interchange fees, a balancing act An empirical examination of a possible regulatory benchmark Michal Kalina1 Duisenberg School of Finance April, 2015 Thesis, LLM Finance & Law Supervisor: prof. dr. Maarten Pieter Schinkel 1 Author can be contacted through: michal@savnar.nl
  • 2. 2 Interchange fees, a balancing act An empirical examination of a possible regulatory benchmark Michal Kalina1 (20120064) Duisenberg School of Finance, LLM Finance & Law Programme Master thesis, 20 April 2015 Abstract In this paper I show the notional development of the level of the interchange fee for debit card payments in the Netherlands, based on an alternative method to the tourist- test. The alternative has a few a priori attractive features, such as a closer resemblance to a debit card favouring market, whereas most of the literature does not distinguish between debit and credit cards or focusses on the latter. Countries in Europe with high card usage per capita and low interchange fee levels, are typically debit card markets. However, using cost data for 2002 and 2009 I show that the method is likely to exhibit market disturbing effects by reversing the market structure through a negative and much larger interchange fee than in reality. In the long run, calibrated to actual institutional settings, the method would create the wrong incentives, stimulating the use of the more expensive payment instrument and discouraging the more efficient. The results indicate that application of this alternative method would be ill-advised for countries such as the Netherlands, with rising costs for cash and declining costs for debit card; thus, regulators should be looking for other benchmarks. 1 I would like to express my gratitude to a few people without whose support or doings, this thesis would not have been accomplished: Joe McCahery for directing a superb programme in Finance & Law; Maarten Pieter Schinkel for sharing his insights on the subject and support to set the scope; Søren Korsgaard for elaborating on his model, enabling me to properly calibrate it; Nicole Jonker for showing me the adjustments in cost item classifications in the 2009 data set compared to the 2002; Frank Leerssen at Rabobank for his support and flexibility to combine my work and study; and foremost my wife, Caroline Spoor-Kalina, for the best support I could wish for throughout the course, this thesis and beyond. The views expressed in this study are my own, as are any remaining errors.
  • 3. 3 Table of Contents 1 INTRODUCTION..............................................................................................................................................8 1.1 Background and motivation.............................................................................................................8 1.2 Research question .............................................................................................................................11 1.3 Academic contribution.....................................................................................................................12 1.4 Thesis structure..................................................................................................................................13 2 PAYMENTS MARKETS, INTERCHANGE FEES AND REGULATION ..........................................14 2.1 Basics and terminology ...................................................................................................................14 2.2 Social costs of payments..................................................................................................................15 2.3 Challenges in payment markets...................................................................................................15 2.4 The concept of interchange fees ..................................................................................................17 2.5 Justifications for interchange fees...............................................................................................18 2.6 Some landmarks in interchange fee litigation........................................................................20 2.7 A note on SEPA and PSD..................................................................................................................23 3 REVIEW OF LITERATURE........................................................................................................................26 3.1 Preliminary...........................................................................................................................................26 3.2 Models of interchange fees and policy implications............................................................27 4 ANALYTICAL FRAMEWORK ...................................................................................................................37 4.1 Interchange Fee model ....................................................................................................................37 4.2 Costs model ..........................................................................................................................................41 5 DATA ................................................................................................................................................................45 5.1 Data collection 2002.........................................................................................................................45 5.2 Data collection 2009.........................................................................................................................46 5.3 Data..........................................................................................................................................................47 6 ESTIMATION.................................................................................................................................................49 6.1 Payments developments in the Netherlands 2002 - 2009................................................49 6.2 Calibration ............................................................................................................................................53
  • 4. 4 6.3 Obtained results .................................................................................................................................56 6.4 Robustness............................................................................................................................................57 7 CONCLUSIONS..............................................................................................................................................60 7.1 Discussion of the results .................................................................................................................60 7.2 Limitations of the study...................................................................................................................61 7.3 Afterword..............................................................................................................................................61 8 REFERENCES ................................................................................................................................................62 8.1 Bibliography.........................................................................................................................................62 8.2 (Pre-) Legislation, Cases and Miscellaneous Links...............................................................66 9 APPENDICES .................................................................................................................................................69
  • 5. 5 LIST OF ACRONYMS AND ABBREVIATIONS ABC Activity Based Costing ACH Automated Clearing House, see CSM. ACM Autoriteit Consument & Markt. Founded in April 2013, merger of Dutch Telecom regulator (OPTA), Dutch competition authority (NMa) and Dutch Consumer Rights supervisor. AFM Autoriteit Financiële Markten. Dutch supervisor of financial markets conduct. AML Anti-Money Laundering ATM Automated (or automatic) Teller Machine. BBAN Basic Bank Account Number. BGC BankGiroCentrale. Former interbank payments processor (CSM) in the Netherlands. CSM Clearing and Settlement Mechanism. In the context of this paper, CSM can be deemed synonym for ACH. More info: e.g. Kokkola (2010). CUP China Union Pay, a four-party scheme card brand based in Shanghai. DNB De Nederlandse Bank, Dutch central bank. EC European Commission. ECB European Central Bank. ECJ European Court of Justice. EEA European Economic Area: all Member States of the EU plus Iceland, Norway and Liechtenstein. EIM Economisch Instituut voor het Midden- en Kleinbedrijf (~Economic Institution for the SME sector). Now part of Panteia. EP European Parliament. EPC European Payments Council. More info: http://www.europeanpaymentscouncil.eu/ FD Financieel Dagblad, Dutch leading daily financial newspaper. HACR Honour All Cards Rule. A business rule set by a scheme prohibiting the merchant who accepts a particular card brand to distinguish between issuers (general implementation of the rule in Europe, differs from the US). IBAN International Bank Account Number.
  • 6. 6 IF Interchange Fee. JCB A three-party scheme credit card company based in Tokyo. MIF Multilateral Interchange Fee. MinFin Either the Dutch Minister of Finance or the Ministry of Finance. MOB Maatschappelijk Overleg Betalingsverkeer (~Societal Consultation on Payments), by order of MinFin in 2002. Chaired by DNB. More info: http://www.dnb.nl/binaries/Oprichting%20MOB_tcm46-274623.pdf NBC Nationale BetalingsCircuit. See Box 3. NFC Near Field Communication or Near Field Communication technology. NMa Nederlandse Mededingingsautoriteit. Dutch competition authority. Now part of ACM. NVB Nederlandse Vereniging van Banken (Dutch Banking Association) NYSE New York Stock Exchange OFT Office of Fair Trading. UK’s consumer protection and competition authority. OJEC Official Journal of the European Communities. PIN Personal Identification Number. Also the brand and scheme of the former Dutch domestic debit card. Became part of Dutch vocabulary: to pin is to make a debit card payment or ATM cash withdrawal. POS Point Of Sale. PSD Payment Service Directive (Directive 2007/64/EC). See References. PSP Payment Service Provider, term introduced in PSD RTGS Real Time Gross Settlement (system) SCT SEPA Credit Transfer (scheme) SDD SEPA Direct Debit (scheme) SEPA Single Euro Payments Area. The jurisdictional scope of the SEPA Schemes currently consists of the 28 EU Member States plus Iceland, Norway, Liechtenstein, Switzerland, Monaco and San Marino. Note that this is a larger area than the Euro-countries. More info: http://www.europeanpaymentscouncil.eu/index.cfm/knowledge- bank/epc-documents/epc-list-of-sepa-scheme-countries/ SME Small and Medium-sized Enterprise SO Statement of Objections
  • 7. 7 St.BEB Stichting Bevorderen Efficient Betalen (Foundation to Promote Efficient Payments). Founded in Nov. 2005 to manage EUR 10 million donated by the banks as agreed in the Convenant Betalingsverkeer. More info: http://www.efficientbetalen.nl/ SWIFT Society for Worldwide Interbank Financial Telecommunication. A member- owned cooperative, predominantly owned by banks. TFEU Treaty on the Functioning of the European Union (2012). TPP Third Party PSPs, term introduced in PSD2
  • 8. 8 1 INTRODUCTION “The most important financial innovation that I have seen the past 20 years is the automatic teller machine, that really helps people and prevents visits to the bank and it is a real convenience. How many other innovations can you tell me of that have been as important to the individual as the automatic teller machine, which is more of a mechanical innovation than a financial one?” Paul Volcker, former FED Chairman1 1.1 Background and motivation2 The history of the payments industry is the history of banks and money. And it is full of “mechanical innovations”, some of which successful like the ATM, while many others were not or only briefly so. For example, after almost 20 years the Dutch banks have decommissioned the Chipknip, a domestic e-wallet type payment product3, per the beginning of this year. An empirical cost study in 2002 showed that a payment made by Chipknip had already then by far the lowest variable social costs compared to cash, debit card or credit card4. For the longest time most consumers had one in their wallet and the system had worked without an interchange fee, yet as a payment product it never became a real success5. One may wonder why. Clearly, the absence of an interchange fee is no guarantee for a successful payment instrument. Nor is the presence of one. This paper is however not a post- mortem on the product. The worldwide success of payment cards is undisputed and it provides a fruitful ground to academics for study. And for regulators to intervene, so it seems. For long, offering payment services has been the exclusive domain of banks and their legal monopoly on providing current accounts or creating money remains until today. Being mostly private institutions entrusted with the facilitation of a public good6, banks are obligated to comply with a sizeable and increasing set of rules and regulations. On the one hand, this may ensure the security and 1 Speech at the occasion of addressing ideas for reforming financial services at Wall Street Journal Future of Finance Initiative in the UK, December 2009. [1], 21 August 2014. 2 Several terms and concepts introduced in this paragraph will be properly explained later on in this paper. 3 e-wallets currently on the market are more advanced and can typically contain several virtual payment ‘cards’. The Chipknip was designed as an alternative to coins and low-denominated banknotes. 4 3 eurocents, compared to 11, 19 and 80 eurocents for a cash, a debit card and a credit card transaction respectively. Source: (DNB) Brits & Winder (2005), table 4.3, p.26. 5 The officially first Chipknip transaction was done by Wim Duisenberg, then president of DNB on October 26, 1995. It reached its peak in 2010 with 178 million transactions that year; by comparison, there were 2,154 million debit card transactions (source: Currence Annual Report 2011, p.8). The decommissioning was announced by the banks and the brand owner Currence in March 2013, to take effect as of 1 January 2015. 6 See also e.g. Freixas & Rochet (2008) par. 1.1.
  • 9. 9 safety of the entire payment system. For example, banks are expected to take appropriate measures to fight illegal activities such as fraud, money laundering, terrorism financing or tax evasion, or prevent payments to entities (e.g. persons, institutions and countries) that have been sanctioned1. On the other hand the regulatory burden amounts to a significant cost level and since the efficiency of the payment system is also of fundamental concern to society, it constitutes another justification for public intervention. A frequently taken route by governments to increase efficiency, is to create or let market participants create (regulated) monopolies that can then capture the economies of scale that so typically characterise payments processing volumes2. Another way to increase efficiency is to increase competition, or so the contemporary paradigm dictates. In the payments industry however this can lead to different or even opposite results3. To illustrate this point from theory: Matutes and Padilla (1994)4 model a case of three competing banks who choose membership to some ATM network and find that if equilibrium exists, it will not be efficient as there will be either three incompatible ATM networks (so the cardholder can only withdraw cash at ATMs from his own bank) or two banks sharing their ATM network and leave the third bank out. To illustrate the point from practise: it is generally posited that the US banking sector and the US payments industry is competitive, and more competitive than for example (most countries in) Europe. Yet the interchange fees for credit cards are significantly higher in the US (around 50%) than in Europe, the UK or Australia and that was even before regulatory action brought it further down in the latter three regions5. One explanation offered for this phenomenon is known as reverse competition and can be summarized as follows: card schemes compete with each 1 Economic and political sanctions on territories and/or specific persons therein; current examples include Russia, the Crimean part of the Ukraine, Iran, Cuba (although these are now in the process of being (partially) lifted), Zimbabwe, North Korea, and others. There are multiple sanction lists, most importantly the US OFAC list, the EU-sanctions list and lists maintained by designated agencies (e.g. Interpol). Many banks have recently found themselves facing criminal charges and huge penalties for not taking appropriate measures and/or failing to report suspicious transactions. For example: JP Morgan Chase $1.7 billion penalty regarding Madoff’s Ponzi scheme; HSBC for facilitating transactions to and from Cuba, Libya, Iran (and others) and for drug cartels in Mexico and Colombia ($1.9 billion), similarly BNP Paribas ($8.9 billion); UBS ($780 million) and Credit Suisse ($2.6 billion) for facilitating tax evasion; Commerzbank for “unsafe and unsound practises” violating US sanctions and AML laws ($1.7 billion). [2], [3], [4], [5], 12 March 2015. 2 Beijnen and Bolt (2009), in Bolt (2013) p.75, estimate economies of scale for card payments about 0.25–0.30, based on cost data of 8 European processors spanning 15 years. This means that a 100% increase in payment cards volume leads to just a 25–30% increase in total costs. 3 Opposite to conventional economic wisdom. Economic history seems apt in bringing forward counter-intuitive results. For example Gresham’s Law (“bad money drives out good money”) or Akerlof’s lemon (“poor quality second-hand cars drive out good quality second-hand cars”). Some scholars regard government intervention as a cause of counter-intuitive results, e.g. Fisher Black on systemic risk: “… It means that they [the government] want you to pay more taxes for more regulations, which are likely to create systemic risk by interfering with private contracting…” (in Danielsson, 2013, p.35). 4 Taken from Freixas & Rochet (2008) p.87. 5 See Hayashi (2004), p.3, chart 1, also to be found in Weiner & Wright (2005), p.299, chart 2.
  • 10. 10 other by offering ever higher interchange fees to banks that issue their cards. These IFs constitute revenues to these bank and this results in higher fees for card payments for merchants who pass these costs on to consumers, either directly or indirectly1. The European Commission (EC) uses this argument in their assessment of the economic effects of interchange fees2. A second effect of reverse competition is that the increased interchange fee (IF) level then may serve as a protection for incumbents from market entry of new card schemes as these will at least have to match the existing IF level to get a foothold. Similarly, the increasing regulatory burden can also work as a barrier to market entry as a large part thereof are sunk costs. In this sense it may perhaps seem ironic that the EC has recently proposed more regulation to increase competition and foster innovation on the payments market. I’m referring in particular to the ‘Payments Legislative Package’3, consisting of an Interchange Fee Regulation (IFR) and a revised Payments Services Directive (PSD). Years of legal combat and emotional debate have preceded the IFR4, in courts and in political and academic circles5. The IFR has largely been voted for on 10 March 20156 in the Plenary of the European Parliament7. Once official, a maximum IF will apply for debit cards of 0.2% and for credit cards of 0.3% for cross-border consumer transactions as of 2 months after the IFR enters into force (art. 3) and equally for all consumer transactions as of 2 years 1 By applying a surcharge on the card payment or by slightly increasing overall prices of the goods sold. 2 See for example SDW(2013) 288 final, the EC’s Impact Assessment accompanying the proposals for PSD2 and IFR (see footnote 3 below). Europe Economics, a London based consultancy firm, in a report commissioned by MasterCard finds this assessment to be “plainly wrong” (Europe Economics, 2014, p.11). 3 Presented by the EC on 24 July 2013. Core pieces are COM(2013) 547 final, hereafter PSD2 (Payment Services Directive 2); and COM(2013) 550 final, hereafter IFR (Interchange Fee Regulation). The EC currently expects adoption in May 2015. 4 Parts of the PSD2 are also still under heavy debate, see §2.7. 5 As another example of possible adverse outcomes: several banks have voiced that the EC’s choice to let the cardholder choose the payment card brand at the moment of the transaction when more than one brand is accepted at the POS (IFR art. 8(18)), will lead to higher costs for the users and strengthen the MasterCard/VISA duopoly in Europe. The argument is that the cardholder will choose the brand that’s on top of mind which is most likely one of the two international brands and not some cheaper yet less known local brand; thus, this will start an ever intensifying advertising campaign, a race ultimately to be won by the two deep-pocketed companies. See for example [12], 9 April 2015. The hypothesis bears similarity to the reverse competition argument, formal economic literature however points in the opposite direction, see e.g. Rochet & Tirole (2002). For more on consumer payment behaviour, see e.g. HBD (2002), Jonker (2007), Bolt, Jonker and Renselaar (2008), Jonker, Kosse and Hernandez (2012), Kosse and Vermeulen (2014) and Bagnall et. al. (2014). 6 Proposed amendments to articles have been taken into account in this paper as far as possible and relevant until 31 January 2015. As the exact final wording of the entire text is not yet officially published, I use the text as published in 2013 (COM(2013) 547 and 550), unless otherwise indicated. 7 Similar developments take place all over the world. For example Rochet & Wright (2009, p.4-5) report “...more than 50 lawsuits concerning interchange fees filed by merchants and merchant associations against card networks in the US, while in about 20 countries public authorities have taken regulatory actions related to interchange fees and investigations are proceeding in many more”. Hayashi (2013) gives an updated overview and accounts for over 30 countries where regulatory actions have been taken.
  • 11. 11 after the IFR enters into force (art. 4). These caps have been calculated by MasterCard1 using a test called the (merchant) avoided-cost test, popularised as the ‘tourist-test’, derived from Rochet and Tirole (2008). The EC finds it a “reasonable benchmark for assessing a MIF level that generates benefits to merchants and final consumers”2. To calculate the figures MasterCard used three empirical cost studies of the central banks of Belgium, Sweden and the Netherlands (DNB). The latter study was published by Brits and Winder (2005) and contains data pertaining to the year 20023. MasterCard implemented the caps on their cross- border MIFs with the EC’s consent as of July 2009 (see also §2.7). Jonker and Plooij (2013)4 have analysed the tourist-test using the Brits and Winder data and a second dataset pertaining to 2009 and report in an understatement that their results indicate the test can have “unintended consequences”: in a country like the Netherlands where social costs of cash are increasing and the social costs of debit card payments are decreasing, the IF level may actually more than double from 0.2% to 0.5% of an average debit card transaction size and still pass the test (in other words: the test would qualify the IF level as ‘not excessive’). If this rise were to be passed on to merchants completely, they would experience a surge in their costs by 233%. Moreover, where an IF is intended to cover (part of) the internal costs of the issuing bank, the calculated tourist-test IF level would be higher than the issuing and acquiring costs of the banks combined. The researchers therefore recommend competition authorities to look for a different benchmark. 1.2 Research question The objective of this study is to examine an alternative model to the tourist-test and analyse how that model, if applied, would affect the IF level for debit card payments in the Netherlands over time, using cost data for 2002 and 2009, i.e. the same data sources that the DNB researchers used for their empirical test of the tourist-test. This alternative model has some preferable features, in particular a closer resemblance to a debit card favouring country such as the Netherlands, whereas the Rochet and Tirole’ model does not distinguish between debit and credit cards. The main question to be answered is: Does the method present a viable alternative to the tourist-test benchmark? 1 Schwimann (2009). 2 Press release by the EC, Commissioner Kroes, 1 April 2009, memo 09/143. 3 The Belgian and Swedish studies concern the years 2003 and 2002 respectively. 4 Also published by Bolt, Jonker & Plooij (2013).
  • 12. 12 PM: I make a distinction between the formal model and its operational counterpart which I casually refer to as ‘method’. 1.3 Academic contribution After Baxter (1983) his contribution, the topic of interchange fees faded from academic view and policy circles. By the end of the 1990s the topic started to attract academic attention, in parallel with law suits and regulatory investigations in a number of countries in the world. Rochet and Tirole (2000) published a seminal paper and since then economic literature on the topic surged, together with the building of knowledge on the functioning of two-sided markets. The more formal economic literature in the first decade of this century generally yields two broad conclusions: 1) IFs are foremost a balancing device, as opposed to a collusive device, their level depends on much more (if at all) than the costs of producing the payment alone; 2) there is no apparent need for a regulator to intervene. These findings are of particular importance considering that regulators and courts usually employ a test based on issuers’ costs in their assessment of possibly excessive merchant fees. Some more recent academic contributions however do point to market failures that might need mending by regulation. Confrontation with empirical data was for long very rare. As far as I am aware, Brits and Winder (2005) were one of the very first to publish empirical cost data (see also section 5). Their study concerned the Dutch POS1 payment market in 2002. Similar studies by other central banks in European countries followed, often adopting their conceptual cost model. Schmiedel, Kostova and Ruttenberg (2012) present the aggregated results of thirteen of these empirical studies, most of which with data on 2009. So far, only for the Netherlands, Sweden and Norway detailed cost studies on two or more years are available2 (year 2009 data on the Netherlands and Sweden were part of the study by Schmiedel c.s.). This offers a first opportunity to study the development in social costs of card payments. Jonker and Plooij (2013) use two datasets on the Netherlands in their examination of an application of the tourist-test. To my knowledge Korsgaard (2014) is the first to construct a formal model that resembles more that of a debit card favouring market, whereas other papers focus either on credit cards or on the payment activity, rather than the particular instrument. He also reviews his model in the light of empirical data for Denmark3 and finds a ground for regulation (as is 1 Point Of Sale. 2 See Jonker (2013) p.9. 3 Most of the data Korsgaard uses is published in Danmark Nationalbank (2011, 2012) which were the basis of the Danish part of the cost study by Schmiedel c.s.
  • 13. 13 already the case in Denmark). My study is the first to calibrate that model to two different years and to analyse the predicted notional IF levels over time. By using the same datasets and data sources as in the empirical test of the tourist-test for the Netherlands, the predictions of the two models can be compared. 1.4 Thesis structure The remainder of this paper is structured as follows. Section 2 offers an introduction to the retail (card) payments industry, introduces basic terms and concepts and touches upon regulatory developments. Section 3 reviews the literature on interchange fees with a focus on formal models that allow for a welfare analysis. Section 4 presents the analytical framework to examine the notional interchange fee levels in the Netherlands, consisting of a formal model for optimal interchange fees and a cost model that captures the concept of social costs in payments. Section 5 outline how the data for the two years, 2002 and 2009, have been collected and presents the most important data. Section 6 then shows how I calibrate the model with the empirical data, presents the results and extends some robustness checks. Section 7 discusses the results and offers some recommendations for future research and policy. Finally, section 8 contains the references. Reading notes For illustrative purposes, I sometimes make references to news articles, position papers etc. that can be found on the internet. I refer to them, mostly in footnotes, with a simple [#], which is short for: “see References, link [#number]”, for example link ‘[9]’. The references link is followed by a date indicating when I last accessed it. Occasionally I place short stories, also meant as illustrative background information, in boxes. They can be skipped without loss of thread of this study.
  • 14. 14 2 PAYMENTS MARKETS, INTERCHANGE FEES AND REGULATION This section provides a brief introduction to the retail (card) payments industry and a context for putting the findings of my research into perspective. 2.1 Basics and terminology In retail payments there are usually four parties involved: the payer and the payee who transact with each other, and their respective banks. In their roles of economic agents, the payer is normally referred to as the “buyer” (of goods) and the payee is the “seller”: the payee/seller sells goods or services to the payer/buyer and in return the payer offers cash, writes a cheque, uses his debit or credit card, initiates a credit transfer, mandates the payee to initiate a direct debit, or uses some other payment instrument1. In the context of (wire) transfers, the payee is normally referred to as the “beneficiary” (of the funds) and his bank is the “beneficiary’s bank”. In the context of payment cards, the seller is usually referred to as the “merchant” and his bank as the “acquirer” or “acquiring bank”; the buyer is then referred to as the “cardholder” and his bank as the “issuer” or “(card) issuing bank”. A debit card is issued by the issuing bank to the cardholder and provides him electronic access to his current account2, either through internet/online banking, to withdraw cash from an ATM or to initiate a transfer of funds at a POS terminal. A credit card on the other hand is not directly linked to a current account and in three-party models (see §2.3), the credit card company is an entirely different firm than the bank where the cardholder holds his current account. Retail payments between banks on behalf of their customers, millions per day, are usually not executed individually but sent to clearing institutions or Clearing and Settlement Mechanisms (CSMs) who sort, match and net all of the incoming and outgoing payment instructions and then calculate the net amount to be paid or received per bank participating in the mechanism. Information on these calculated net amounts is then sent to the settlement institution, usually the local central bank, who settles the claims. This means that the central bank, where all domestic banks hold an account, transfers funds between these accounts corresponding to each bank’s claim, provided there is sufficient liquidity or collateral. The banks are only liable for these net amounts but the banks receive details of all underlying 1 I use the term “payment instrument” in a general meaning as a way to initiate a transfer of funds. Whether a paper cheque, a plastic card with a magnet stripe or chip, a chip with NFC inside a mobile phone or an app on a smartphone, the exact technology or carrier is not of particular relevance in the context of this paper. 2 Synonyms for “current account” are “payment account” and “checking account” (typically US custom).
  • 15. 15 transfers from the CSM, enabling them to credit each customer’s account accordingly. The settlement is usually a daily process at the end of the business day, but frequently the participants choose to settle several times a day1. 2.2 Social costs of payments From the point of view of society, payments are costly: estimations of the costs of retail payments for the European Union (EU) are roughly 1% of GDP, i.e. € 130 billion. Banks and interbank infrastructures (e.g. CSMs) incur 50% of these social costs, retailers about 46%, central banks about 3% and the remaining 1% is incurred by cash-in-transit companies (e.g. conveyance of banknotes)2. Social costs refer to the costs of the resources employed to ‘produce’ the payment services, that is: the sum of all internal costs by all the agents involved to execute a payment transaction. Private costs are all the costs a single agent faces and/or incurs to produce a payment transaction and include by definition internal and external costs. External costs are the charges (e.g. fees) the agent pays to other agents in the value chain. Since the external costs to one agent in the chain are revenues to others, these external costs cancel out in the calculation of total social costs. Equivalently, social costs can be defined as the sum of all parties their net costs (total costs – revenues)3. 2.3 Challenges in payment markets Payment markets exhibit interesting dynamics, in part resulting from multiple potential market failures. I will indicate a few, in no particular order. Banks are in competition with each other and increasingly with non-banks. Yet a payment transaction cannot take place without some basic form of cooperation between the bank (or party) acting on behalf of the payer and the bank (or party) acting on behalf of the payee, at the very least on some standard of exchange4. The element of cooperation may conflict with the element of competition. An 1 Next to this there are systems to settle individual, typically high-value transactions in real-time, so-called RTGS systems, for example TARGET2 used by e.g. central banks in Europe. The CSMs in the payment cards business are usually different from the CSMs used for wire transfers. Indeed, within the payments business these are rather separate domains. Typically cards settlement is a daily cycle (end to end) whereas wire transfers often take a few working days. For example an SCT, see §2.7, is regulated to take no more than 2 business days, that is: the crediting of the beneficiary’s account must happen at the latest at the business day following the business day on which the bank received and accepted the payer’s payment instruction (as of 1 January 2012 (PSD, art. 69); until 2012 this was max. 4 business days and before the introduction of the PSD, this could be even longer). 2 Schmiedel, Kostova and Ruttenberg (2012), p.6. Numbers refer to the year 2009. This excludes social costs incurred by households, which would add on average another 0.2% if those were to be included (id. p.7). 3 See e.g. Schmiedel c.s. (2012), Brits & Winder (2005) or Jonker (2013). Section 4 will give more details. 4 This is the case even when the economic transaction is settled in cash (with fiat money): if a merchant sells goods and accepts some ‘pieces of paper’ in exchange but later finds out other economic agents such as the banks do not, then he is likely to end up with a real loss. On a larger scale, the economy might then revert to barter trading.
  • 16. 16 industry practise is to separate (as a first step) the set of standards, e.g. the ‘scheme’ and scheme rules, from the product, i.e. the payment instrument. The latter is the object of competition, whereas the former defines the space where cooperation is, under certain conditions, allowed; in the case of payment cards the space has been gradually narrowed down by courts and regulators where for example certain scheme rules have proven to hamper competition, see section 3. For a standard to be set effectively requires the cooperation of a significant part of the market to gain critical mass, which may pose coordination problems. Similarly, for a card scheme to be successful requires attracting a sufficient number of merchants as well as consumers (cardholders). Three-party schemes, for example American Express, Diners Club, JCB and Discover, who issue their cards directly to cardholders and directly acquire merchants, can set a price structure and price level that appeals to both sides of the market. Four-party schemes on the other hand, such as MasterCard, VISA and CUP, do not deal with both sides of the market simultaneously and are therefore unable to directly balance the costs and benefits to both. This may be a role for interchange fees. But without the right price structure there may be no payment card product in the first place. Typical of payment card markets and two-sided markets in general, is the asymmetrical distribution of costs and benefits, resulting in a (heavily) skewed price structure: one side of the market bears a significantly larger part of the total costs than the other. More on these points follows in the next paragraphs and section. Strong economies of scale may lead to monopolistic platforms with the most efficient cost of production possible but equally to a risk of abuse of the dominant position and/or a price setting below socially optimal, thus leading to an underprovisioning of payment cards and furthermore a lack of incentives to continue to improve efficiency through innovation. The presence of switching costs to bank customers may reduce banks’ incentives to make their (ATM) networks interoperable. The same mechanism applies to banks themselves: the presence of switching costs may reduce interbank networks (e.g. CSMs) to become With cash, the payment instrument and the standard of exchange coincide. With electronic money there is a difference: some technological standard for data exchange and the acceptance of these data as a change in debt obligation (through entry in the bank’s books), that is: liability or ‘money’. This is indeed somewhat ambiguous because ‘money’ has no clear and unambiguous real-world definition, despite being such a fundamental economic concept. Recall the definitions of monetary aggregates M0, M1, M2 etc., see e.g. Danielsson (2013) p.36. See also §2.5.
  • 17. 17 interoperable1, which could ultimately make a transfer of funds between two banks connected to two different networks, impossible2. As a last example, I revisit the role of banks as guardians of a safe and secure payment system. For this they need to share payments information and sometimes information on the transacting parties, the beneficiary and payer. This sharing of information can conflict with consumer data protection and privacy rights. These conflicts can occur on national level but are exacerbated on international level as countries, e.g. the US and the EU, or Member States within the EU, not always agree or align their positions3. 2.4 The concept of interchange fees The concept, the mechanics so to say, of interchange fees is straightforward. Following the setting of payment cards in a stylized setup, picture a consumer (the cardholder) using his payment card to pay for some goods at a merchant’ store. The cardholder received this card from his issuer; in case of a debit card this is usually his own bank, in case of a credit card this can be his own bank, a company affiliated with his or some other bank or a totally unrelated firm (e.g. American Express and other three-party schemes). This relationship is mirrored at the other side of the trade: the merchant can accept this payment card as a legitimate way of receiving payments ever since he has a contract with his acquirer4. When the cardholder swipes, or nowadays ‘dips’ his card at the POS terminal5, he initiates an authorisation request. In case of a debit card this request is sent to his bank and if there is enough money on the account, the transaction will be authorised. In case of a credit card, the request is sent to the acquirer who authorises the transaction6. The authorisation is observed by the merchant who 1 See also Box 3. 2 Matutes and Padilla (1994) in Freixas and Rochet (2008); for more on the economic effects of switching costs in general see for example Klemperer (1987). 3 Recall for example the conflict between the EU and the US in the summer of 2006 on the latter’ Terrorist Finance Tracking Program where they, based on their International Emergency Economic Powers Act of 1978, gathered information on financial transactions from SWIFT, the world’s largest financial communications network, following the events of September 11, 2001. The EU forced SWIFT to process intra-EU financial messages within the EU, removing their backup systems from US territory. 4 The merchant also need a contract with at least a telecommunications network provider for data exchange between the terminal and a switching centre. 5 For the example it is irrelevant whether it is a POS transaction in the physical world or an online transaction, the IF works exactly the same way. 6 In practise, there is obviously more to this. For example, the card has to be validated first, checked for not reported stolen, missing or other fraudulent options and the authorisation request is first sent to a switching centre, perhaps rerouted to another before it arrives at the issuer. The request can also be handled offline by the terminal. See e.g. Kokkola (2010) for more on these mechanisms. Notice the difference in risk and liability between a debit and credit card transaction. Credit card companies will sometimes contractually shift the risk of e.g. eventual non- payment to the merchant.
  • 18. 18 then can rest assured he will receive a transfer of funds to his bank account in an amount equal to the selling price of the goods. An IF is simply an amount of money, a fraction of the value of this funds transfer, that usually goes the opposite direction: from the acquirer to the issuer1. It is calculated per card-transaction, either some percentage of the value of the transaction or a fixed fee per transaction. The IF can be determined either bilaterally between the issuer and acquirer in an arm’s- length agreement or multilaterally by the scheme. The multilateral interchange fee (MIF) then applies as a default fee, a fallback in the absence of a bilateral agreement. From a practical point of view, the MIF accomplishes that the transaction between the merchant and cardholder can proceed, without the issuer and acquirer ever having made agreements with each other. The efficiency gain is obvious: with more than 8,000 credit institutions (banks) in Europe alone, one would need over 32 million bilateral agreements to accommodate all possible card payment transactions. Disturbing to regulators is that a MIF, in case of VISA and MasterCard before they were listed (and even thereafter) used to be set by an association of banks. Parallels with a cartel are apparent. In case of a (pure) three-party model there is no (explicit) IF as the scheme typically combines and includes the issuing and acquiring functions. 2.5 Justifications for interchange fees Whereas the concept of IFs is straightforward, it is the interpretation and justification that meets scepticism and critique. For a part this is understandable as payment markets are an instance of two-sided markets and the body of economic knowledge on their functioning is still evolving. In the earlier literature, economic scholars have gone sometimes to a considerable extent to emphasise that two-sided markets are nothing exotic but an often observed and old phenomenon, for example Evans (2003), Rochet and Tirole (2000, 2002, 2005) and Wright (2004a). This has lead others to the remark that therefore no new antitrust policy should be necessary, at least when dealing with predatory pricing, see e.g. Motta, 2004, p.452. The main message from these early contributions is that in a two-sided market, it is perfectly normal and just that one side of the market pays (substantially) more than the other (instead of each their own ‘fair share’). Consequently, there is no economically justified reason for a regulator to intervene. 1 The only country I know of where the opposite used to be the case is Australia for eftpos-transactions, a PIN based debit card scheme (Weiner & Wright, 2005, p. 293 fn.5). However, in 2012 the IF direction was reversed.
  • 19. 19 It should be noted that IFs are not a feature exclusive to payment cards: other retail payment instruments like cheques, direct debits and typical local instruments like the Dutch Acceptgiro also carry IFs1. An ATM cash withdrawal by a customer of a different bank than the one owning the ATM is sometimes mentioned in the literature as another example of an IF, as it involves a fee from the customer’s bank to the ATM-owning bank. This is however not a good example as there are only three agents involved: the two banks and a customer of one of them, but a merchant is missing. In a ‘true’ two-sided market, an issuer (or: bank of the payer) supplies a service to an acquirer (or: bank of the payee) who supplies a service to the merchant (payee), but the issuer also supplies a service to the cardholder (payer) to enable his payment transaction with the merchant. The more merchants accept the payment instrument (cheques, debit card, credit card, direct debit,…) the higher the benefits of using it to the payer/cardholder and vice versa. One could therefore reframe the question on the justification of the IF as: does the issuer get compensation from the acquirer for part of his internal cost? Or do the issuer and acquirer essentially cooperate to redistribute some of the benefit the payment instrument usage has to the merchant from the merchant to the cardholder, in order to induce the latter to use it through a lower fee? This interdependency and ‘general’ feature adds merit to the argument that IFs are primarily a device to balance a two-sided market when there is not a single platform to do that implicitly, as for example with a three-party scheme2. What complicates matters is that once a line of thought has been chosen, e.g. by a regulator or the market participants themselves, the line tends to persist. Domestic direct debit schemes in France, Italy and the Netherlands for example carry an IF but these are strictly cost based and, as for the Netherlands, set collectively by the Dutch banks based on the per transaction processing costs of the most efficient debtor bank (issuer). This practise was 1 During the ‘free banking periode’ in the 19th century, commercial banks could issue their own banknotes and central banks were typically established to organise their orderly issuance. A central bank issues its own liabilities for use as ‘central bank money’. Commercial banks too began increasingly to issue liabilities, i.e. ‘commercial bank money’. A layered structure then emerged, whereby private individuals held their deposits in commercial banks, and these in turn held theirs in accounts at the central bank. Individuals’ confidence in commercial bank money lay in the ability of commercial banks to convert their liabilities into liabilities of other banks or central bank money when requested by their customers. The central banks were in particular responsible for ensuring that central bank money and commercial bank money could coexist and be ‘interchangeable at par’ (Kokkola, 2010, p.152). 2 See e.g. Rochet and Tirole, 2003, p.73-76, where they argue that for four-party models, the IF is the only balancing device available.
  • 20. 20 authorised by the NMa1. For SEPA Direct Debits (SDDs), (M)IFs are prohibited as per November 2012 for cross-border SDDs and as per February 2017 for domestic SDDs2. This leads to the question if/when an IF is competitively neutral or socially optimal and consequently, if there is reason for a regulator to intervene. These questions are addressed in the review of literature section. 2.6 Some landmarks in interchange fee litigation Evans and Schmalensee (2005) report the case of National Bancard Corporation (NaBanco) vs. VISA3, filed in 1979 and decided in 1986. NaBanco claimed the MIF that VISA’s member banks set, was a per se violation of the Sherman Act. This claim was rejected by the court with reference to the potential efficiency benefits for a two-sided market. In 1983, William F. Baxter, then professor of Law at Stanford who had worked on the case, published a paper that was the first to explain the rationale for interchange fees. After the court’s ruling however “... interchange fees faded from view in academic and policy circles and was a topic of interest mainly to industry insiders” (Evans and Schmalensee, 2005, p.75). That changed with the proliferation of payment cards in the Western world (see charts 1 and 2a below for an illustration) accompanied with lawsuits and complaints to regulators by retailers who were confronted with an increase in their costs of receiving payments as a result of the usage increase. [charts 1 and 2a about here.] The high profile case in the US of Wal-Mart, Sears Roebuck, Safeway and a long list of other merchants versus VISA and MasterCard marks the increase in litigation. The case started in October 1996 and took seven years of legal combat before it was settled in 20034. In Europe a similar case was filed in 1997 by EuroCommerce, an association of large European retailers 1 Exemption granted in 2002 for 5 years and since then prolonged. Exemption is based on art. 17 Dutch Competition law (Mededingingswet) “…promoting technical or economic progress…”, compare art. 101(3) TFEU. 2 Per Regulation EU 260/2012, with the exception of so-called R-transactions, i.e. returned or rejected DDs. These IFs too must be “strictly cost based” (art. 8). 3 As do Rochet and Tirole (2000, p.2 fn.9). 4 The case turned into a class-action lawsuit and was finally settled on June 4, 2003 for “the largest antitrust settlement in history” ($2 billion for VISA and $1 billion for MasterCard). Later that month, the district court approved the notice of settlement. Eighteen merchants, of around five million in the class, objected to the settlement; the district court however approved it in January 2004 after a fairness hearing. E.g. [7], 7 Sept. 2014.
  • 21. 21 and national commerce federations, however not before court but with the EC. The difference in approach between these two quests for relief is noteworthy. An important explanatory factor is the difference in the implementation of VISA’s and MasterCard’s Honour All Cards rule (HACR). In the US the two schemes required merchants who accept their credit card to also accept their (signature-based) debit card, whereas in the EU the merchant is not allowed to distinguish between issuers of a card brand, that is: the merchant has to accept e.g. all debit cards but is allowed to refuse the scheme’s credit card. Wal-Mart c.s. argued that the tie-in of both types of cards, together with other scheme rules was an attempt to monopolize the debit card market in violation of section 1 of the Sherman Act (anti-cartel rule) and as a consequence the plaintiffs had been charged excessive interchange fees during October 1992 and June 2003. The courts affirmed this notion. This tie-in of both types of cards has never been present in the EU1. After the EC’s investigation into VISA’s MIFs and VISA’s offer in 2002 to cap them at the level of relevant costs, VISA’s intra-EEA cross border MIFs were exempted based on Art. 81(3) of the EC Treaty until the end of 20072. EuroCommerce and First Data, a cards transactions processor and acquirer, appealed against the exemption. Both eventually dropped the case but First Data did not do so until VISA had withdrawn another one of their scheme rules: No Acquiring Without Issuing3. Meanwhile, MasterCard had received a Statement of Objections (SO) from the EC in October 2003 for the way it sets its MIFs, basically suggesting MasterCard is acting as a cartel. MasterCard was listed on the NYSE in May 2006, which led to speculations that this change in its legal structure may have been to avoid the antitrust allegations, or at least in part; see Schinkel (2010) for an entertaining account. It nevertheless received a supplementary SO from the EC in June 20064, followed by its infringement decision in December 20075. MasterCard sought annulment of this decision but the General Court upheld it in May 2012, including the surprising assessment that 1 See also Weiner & Wright (2005) p.300. 2 OJEC L318, 22.11.2002. VISA will apply flat-rate intra-EEA MIFs before the end of 2002 whose weighted average will not exceed EUR 0.28 (par. 18) and similarly for credit cards an ad valorem fee of 0.7% before the end of 2007 (par. 19); exemption granted until 31 December 2007 (par. 109). In short, Art. 81(3) states that certain restrictions on intra -EU competition may be allowed if the agreement / decision / practice “…contributes to improving the production or distribution of goods or to promoting technical or economic progress...” provided certain criteria have been met. Art. 81 of the EC Treaty of 1957 has been renumbered as Art. 101 in TFEU. 3 See e.g. Bos (2007, p.114-115) for more on this case. 4 MEMO/06/260, 30 June 2006. [9], 3 April 2015. 5 Case COMP/34.579, December 19, 2007: the EC views MasterCard’s multilateral intra-EEA IF for cross-border payment card transactions made with MasterCard (credit card) and Maestro (MasterCard’ debit card brand) a violation of Art. 81 and the MIFs should therefore be withdrawn within 6 months.
  • 22. 22 MasterCard “… had continued to be an institutionalised form of coordination of the conduct of the banks…”1. MasterCard’s appeal to the ECJ was rejected in September 20142. After the expiration of VISA’s exemption and just before the IPO on the NYSE in March 2008 as of when VISA Inc. separated with VISA Europe with the latter continuing to be a membership organisation, VISA Europe was informed of the EC’s opening of antitrust proceedings, for which it received an SO in April 2009. Another year later both parties had come to an agreement: following MasterCard who had introduced caps on its weighted average cross-border MIFs for debit (0.2%) and credit (0.3%) card transaction in July 2009, VISA Europe would likewise cap its intra-EEA cross border MIFs for debit cards at a weighted average of 0.2%; moreover the same cap was agreed to apply to domestic VISA debit card transactions in nine Member States3. The agreement was made legally binding in December that year. In May 2013, VISA agreed to adopt the same cap of 0.3% MIF for both its cross- border intra-EEA and domestic credit card transactions. As already noted in §1.1, the IFR will take this another step further and apply caps to all four- party payment card transactions, both cross-border intra-EEA transactions and domestic transactions, both debit card (0.2%) and credit card (0.3%) transactions, regardless how the IF is set (multilaterally, bilaterally or otherwise). An exemption is made for pure three-party schemes, schemes that directly contract both the cardholder and merchant. This may raise speculations as to whether the four-party schemes will add another chapter to market oversight games (Schinkel, 2010) and perhaps turn into three-party schemes. As evidenced from a working document, the EC seems aware of this risk of regulatory circumvention4. Impact IFR Had the scope been only cross-border intra-EEA transactions, or only MIFs (as a fallback fee), then the impact of the IFR on issuing bank’s revenues would not have been that large. Merely to illustrate this point: most economies are still locally oriented; to take an ‘extreme’: the 1 Case T-111/08, 24 May 2012, par. 259. 2 Case 382/12P, Sept.11, 2014. Note that the decisions relate to MIFs, not bilaterally agreed IFs or IFs set collectively at national level. 3 See [8], 11 April 2015, for an overview of the EC’s work on MIFs, including links to relevant documents. 4 Commission Services, Working Party on Financial Services, Proposal for a Regulation on interchanges fees for card based payment transactions, Inclusion of three party card schemes under the scope of Chapter II, WORKING DOCUMENT #27, MIF, 17 October 2014. This document was shared with payment industry insiders (such as the author) and may not be available outside the stakeholder group.
  • 23. 23 economy of the Netherlands has a relatively large international orientation, ranking for example no. #1 in 2014 in the Global Connectedness Index of logistics company DHL1. Still, only 2.7% of all debit card transactions are cross-border2. The number can reasonably be taken as an upper limit indication when generalizing to other European economies. However, with the domestic transactions in scope, the impact of IFR will be much larger. How large exactly will depend on the actual IF levels in each country, see for an estimation the EC’s Impact Assessment3. As for the Netherlands, the actual market IF levels are substantially below the proposed caps, in 2009 between 1 and 2 eurocents per debit card transaction, i.e. 0.04% of an average debit card transaction4. The Dutch government therefore would like the EC to choose an even lower cap on debit card IFs5. 2.7 A note on SEPA and PSD The EU and the Member States started a huge endeavour with the integration of the internal market, including money to transact. First cash (banknotes and coins), then the instruments to transfer money electronically. However, throughout the years each country had developed and grown accustomed to their own payment instruments so this proved to be quite a challenge. The European banking industry created the European Payments Council (EPC) in 2002 to help realise SEPA, a Single Euro Payments Area6. Through its efforts and under its coordination, the two most commonly used instruments in the EU for electronic payments, the credit transfer and the direct debit, were redefined in standardising schemes7. Soon thereafter a scheme for payment cards, the most common alternative to cash, followed. To dismantle legal barriers between the countries and create a ‘level playing field’, the Payment Services Directive (PSD) was introduced, taking effect on November 1st, 20098. As already 1 Measured in cross-border traffic of goods, capital, information and people. The Netherlands also ranked #1 in the previous edition of the Index in 2012. Source: FD (Financieel Dagblad), 5 November 2014. 2 Around 70 million cross-border transactions versus 2.6 billion POS debit transactions in total. Cross-border includes Dutch cardholders paying abroad and foreigners paying in the Netherlands. Taking credit card transactions into account, the number is 5.5% (152.6 million cross-border POS transactions versus 2.8 billion in total). Looking at other retail payments instruments, credit transfers and direct debits, then the numbers are 2.1% and 0.0% (too low to measure). Source: DNB retail payments statistics 2014. 3 SDW(2013) 288 final, figure 6, p. 21. 4 Average transaction amount with a debit card in 2009 was EUR 39.07. Calculated with an IF of 1.5 eurocents. 5 Fiche 1: Verordening interbancaire vergoedingen, Kamerstuk 22 112, nr. 1705, 4 October 2013 6 The jurisdictional scope of the SEPA Schemes currently consists of the 28 EU Member States plus Iceland, Norway, Liechtenstein, Switzerland, Monaco and San Marino. Note that this is more than the countries carrying the Euro as their currency. 7 The SCT (SEPA Credit Transfer) and the SDD (SEPA Direct Debit) were implemented as of 2008 and 2009 respectively. 8 The PSD however does not apply to Switzerland and Monaco. Previously, the jurisdictional ‘misalignment’ between the PSD and SEPA was much larger.
  • 24. 24 noted, the PSD is currently being revised. The proposed changes have caused an intense and still ongoing debate and lobbying over certain topics. Two are at the heart of the controversy: 1) the introduction into its scope of two new payment services, so-called payment initiation services and account information services, and the legal provisions regarding the parties offering these services; and 2) the prohibition of surcharging together with caps on IFs (the IFR). Just to briefly illustrate one of the several issues regarding the first controversial topic before continuing with the second, which is the focalpoint of this paper. The parties offering the newly introduced services1 will be allowed to gain access to account information of the consumer and initiate a payment on his behalf using his personal credentials, the same as the consumer himself uses to log in to his internet/online banking environment. The banks consider this a major security risk as it may be impossible for them to unambiguously distinguish the consumer, i.e. their own customer, from a potentially fraudulent party disguised as the consumer. A contract or agreement between the TPP and the bank will however not be required2. The concern has even created ‘unnatural allies’ where banks find regulators3 and consumer’ interest organisations at their side and this has struck a chord with the EP and several Member States4 but so far not visibly with the EC. Given the significant reduction in the general IF level across Europe and the expected consequential reduction in merchant fees, the EC finds surcharging no longer justified for the regulated payment cards and will therefore prohibit surcharging (PSD2, art. 55(4)). The PSD left its regulation at the discretion of Member States and this resulted in half the countries (thirteen) prohibiting it while the other half allowed it5. 1 Called TPPs, Third Party Providers or Third Party PSPs. PSPs are Payment Service Providers, a term introduced in the PSD for banks and non-banks (“payment institutions”) authorised to provide payment services. 2 ING filed suit against AFAS, a provider of bookkeeping tools for small businesses and individuals, claiming AFAS seduces ING’ customers to share their personal credentials in breach of ING general terms and conditions. AFAS used inter alia the upcoming PSD2 in their defence. The court agrees with ING (30 July 2014). [6], 18 August 2014. 3 E.g. BaFin, the German Federal Financial Supervisory Authority; [13], 18 April 2015. 4 For example, France, Denmark and the Netherlands have officially notified the EC of their strong objections against the provisions requiring the sharing of personal credentials. Austria has even filed a waiver, stating that it will under no circumstances transpose these provisions into national law. 5 Situation at the time of the PSD2 proposal, July 2013. A few countries have changed their policies since. Countries that currently allow surcharging include the Netherlands, Denmark and the UK, where regulation abolished the no-surcharge rule in 1991 (Vickers, 2005, p.233 fn.10). In Denmark, only for international debit and credit cards surcharging is allowed, but not the domestic Dankort (so-called ‘split model’, see Danmark Nationalbank, 2012, p.121). For the Netherlands see section 6.
  • 25. 25 Next to these provisions (and a few others), the IFR sets forth in art. 10 the HACR, basically reconfirming the European interpretation of the HACR (§2.6). It will also enforce a legal and organisational separation between the scheme and the processing infrastructure.
  • 26. 26 3 REVIEW OF LITERATURE The cost-based approach, derived from conventional economic wisdom on monopolistic and oligopolistic ‘one-sided’ market behaviour and benchmark perfect (Cournot and Bertrand) competition, does not fit a two-sided market. Market authorities and courts using benchmarks based on issuer’ cost to assess possible excessiveness of merchant charges and interchange fees, have therefore been strongly and unanimously advised by economists to readjust their logic. I have borrowed from Wright to summarise in Box 1 some incorrect applications of ‘one-sided logic’ to two-sided markets. See Wright (2004a) for a discussion of each fallacy and an entertaining illustration using quotes from investigations into credit card schemes by three different market authorities. 3.1 Preliminary At that time however, the economic knowledge on two-sided markets was still rapidly developing. As Evans and Schmalensee put it in their survey of the economic literature on interchange fees and their possible regulation in 2005: “Economists have only scratched the surface of the theoretical and empirical work that will be needed to understand pricing in two- sided markets in general and the determination of interchange fees in particular” (id. p.104). The most important conclusions from the survey can be summarised as follows. Next to the factors that determine socially optimal prices for customer groups in multisided industries, inter alia a) price elasticities of demand, b) indirect network effects between the customer groups, and c) marginal costs for providing goods to each group, socially optimal prices in the payment card industry also depend on other characteristics, including d) the use of fixed and variable fees, e) competitive conditions among merchants, issuers, and acquirers, and finally f) the nature of competition from cash, cheques, and three party payment schemes. Thus, the socially optimal interchange fee is not, in general, equal to any interchange fee based on cost BOX 1. Eight fallacies - lessons to forget in two-sided markets. 1. An efficient price structure should be set to reflect relative costs (user-pays). 2. A high price-cost margin indicates market power. 3. A price below marginal cost indicates predation. 4. An increase in competition necessarily results in a more efficient structure of prices. 5. An increase in competition necessarily results in a more balanced price structure. 6. In mature markets (or networks), price structures that do not reflect costs are no longer justified. 7. Where one side of a two-sided market receives services below marginal cost, it must be receiving a cross-subsidy from users on the other side. 8. Regulating prices set by a platform in a two-sided market is competitively neutral.
  • 27. 27 considerations alone and such a solely cost based interchange fee is unlikely to improve social welfare (id. p.102). In my selection I have mainly focussed on contributions with formal models that allow for a welfare analysis, either total surplus or user surplus1. I describe these models in non- technical terms. 3.2 Models of interchange fees and policy implications In Baxter’s (1983) model, cardholders use and merchants accept a card payment if the per- transaction price charged to each of them is less than the per-transaction benefit they derive from it. Each of their marginal valuation of the payment transaction depends on the other party accepting, respectively using the card (otherwise the valuation is zero). Cardholders are assumed to be more price sensitive than merchants and so their costs, that is the price they pay for the transaction to the issuer, need to be lowered to reach equilibrium. For this, a transfer – the IF – needs to be made from the acquirers to the issuers who both are assumed to behave (perfectly) competitively. Therefore the IF does not affect the overall price level (it could be at any level), only the price structure. In absence of bargaining power, the price the issuers charge the cardholders is not based on their marginal costs, nor is the price the acquirers charge the merchants based on their marginal costs (which is what ‘one-sided logic’ would suggest); instead, in equilibrium aggregate joint demand for card payment services of merchants and cardholders equals the total combined issuers and acquirers costs of providing them. In this perfectly competitive world, an assumption Baxter acknowledges may not hold in reality, without e.g. fixed costs or membership fees, the equilibrium is efficient (the model is not equipped for a welfare analysis). Baxter concludes his article stating that the characteristics of the payments market, i.e. joint costs and interdependent demand, are not well understood and the controversy that troubled the US banking industry for more than five decades around ‘clearance at par’ of cheques2 is likely to repeat in the context of debit and credit cards (id. p.586). Those words proved prophetic. In light of this anticipated controversy, Baxter warns that governmental intervention should be resisted. See Box 2 for a sidestep to Baxter’s account of payment instruments and substitutes. 1 Total welfare, or total surplus, is the sum of consumer surplus and producer surplus. In the context of payment markets, the producer is (are) the banks and other providers in the chain and the consumer is (are) both the payer/cardholder and the payee/merchant, both are “users” of the card. Consumer surplus, or consumer welfare, thus involves simultaneously both users. 2 Baxter uses the US spelling: checks.
  • 28. 28 Schmalensee (2001) allows for imperfect competition among issuers and among acquirers. He focuses on credit cards and on a four-party, cooperative (not-for-profit) scheme that sets a MIF. Assuming the scheme is facing a multiplicative demand function and conducting a welfare analysis, first with a monopolistic issuer and a monopolistic acquirer which he next generalizes to oligopolistic competition, Schmalensee arrives at similar conclusions as Baxter: privately optimal IF is also socially optimal and the IF depends on demand conditions, costs, competition among issuers and acquirers and on externalities between merchants and cardholders. He sees no cause for a regulator to step in. BOX 2. On payment means and substitutes. Baxter’s historical account of US four-party payment systems covering roughly 200 years narrates the growing popularity of cheques (initially: drafts) and currency (initially: bank notes) as markets developed from very local to increasingly larger geographies, while inferior country bank notes were driving sounder city bank notes out of circulation and transporting bank notes was increasingly costly and risky. This explains the heavy usage of cheques for which later the credit card became the prime substitute. By contrast, cheques have never gained comparable popularity in other countries such as Sweden, Spain, Denmark, and the Netherlands (revisit chart 1). Baxter speculates that the ‘clearance at par’ of cheques, before Federal regulation put an end to this heterogeneous practise by imposing an ‘IF’ of zero, was a result of a shift in relative demands of purchasers (~cardholders) and merchants for cheque services and a shift in relative costs in providing them. It is also interesting to note that Baxter treats cheques and credit cards as main substitutes (thus cheques’ costs are the most appropriate benchmark for credit cards’ costs, including IFs) whereas in Europe, economists and the EC focus more on cash versus debit cards. Though these may perhaps be better comparable substitutes , the comparison is not without difficulties as cash usage is very difficult to measure reliably, comes with “interchange fees at par” and nowadays involves a monopolist supplier (i.e. the central bank, see §2.5). Note 1: Although Baxter occasionally mentions debit cards simultaneously with credit cards, the focus of his article is largely on the latter. Indeed, the debit card is a “poor man’s card” as “…any cardholder entitled to use a credit card, will always use it rather than a debit card” (p.585), because of the float benefits attached to the credit card. For this reason, and because of a lower risk of default, Baxter predicts that debit card transactions will be substantially cheaper than credit card transactions and with different IFs. This prediction is confirmed in Weiner & Wright (2005) who report credit card IFs typically between 1% to 2% of transaction value and debit card IFs typically between 0% and 1% for a number of regions and countries (id. tables 1 – 3). Note 2: Payments practitioners frequently use a rule of thumb categorisation of payment instruments from a payer’s point of view: Pay Before, Pay Now and Pay Later. Cheques and credit cards are examples of Pay Later instruments, while cash and debit cards are Pay Now instruments. Prepaid cards like a gift card or special purpose card, are often considered Pay Before instruments as it involves a transfer of funds (from the cardholder’s account to the card or to the administrator of the card) while the goods or service will be delivered somewhere in the future. Because the card usually cannot be used everywhere and for all purposes, like cash, consumers tend to view their funds have ‘changed currency’. Examples are the Dutch Public Transport card (OV-chipkaart) and, according to most of its (former) users, the Chipknip.
  • 29. 29 Rochet and Tirole (2000, published in 2002) introduce in this seminal paper a formal framework upon which most of the later scholarly contributions build. Like Baxter, they take into account the (direct) benefits of using and accepting a card payment. However, they also model merchants and cardholders as strategic players: merchants use the acceptance of cards to generate higher sales revenue by winning consumers (cardholders) from competitors and consumers may decide to visit stores based on stores’ card acceptance policy. Thus merchant resistance to increases in IF is likely to be lower than in Baxter’s model. Their framework also allows for consumers to hold cards of more than one scheme1, making merchants’ opportunity costs of card acceptance endogenous. Acknowledging that the schemes compete for cardholders, merchant resistance to increases in IF may in this respect be higher than in Baxter’s model. Furthermore, acquirers are assumed to behave competitively while issuers may enjoy some market power2. Cardholders are assumed to have a fixed volume of transactions3, which (technically) implies that there is no difference between an annual fixed fee or a variable per-transaction fee, at least not from the issuer- cardholder relationship perspective4. Cardholders are modelled to have structural preferences for using cash or cards. The model makes no specific distinction between debit and credit cards as it focusses on the payment activity rather than the instrument. Merchants face no fixed costs for accepting payments. Given this model setup where by assumption all profits fall at the issuing side of the market, Rochet and Tirole show by applying a total welfare analysis that the socially optimal IF, which is the one where the total cost of the marginal transaction equals its total benefit, coincides with the privately set IF (as set by the scheme, i.e. the issuers) if that IF exceeds the level at which merchants accept the card. This requires a low cardholder fee. Or the privately set IF exceeds the socially optimal IF, in which case consumer fees are set too low leading to an overconsumption (overusage) of cards5. No cost-based regulatory intervention can prevent this. Both equilibria apply under the no- surcharge rule, a scheme rule that prohibits the merchant to price-discriminate between payment means (e.g. demand an extra fee from the customer for a card payment). Lifting this rule creates however ambiguous welfare effects6. 1 This is sometimes also referred to as multi-homing (vs. single-homing). 2 As Rochet and Tirole note on p.5 fn.13, this is closer to reality. As an indication thereof, they refer to the voting rights of the banks in the US in VISA and MasterCard (before their respective IPOs) which are more sensitive to issuing than to acquiring volume, suggesting some bargaining power is on that side. 3 In their model this is normalized to one transaction for each customer. 4 For their main model, cardholders are assumed to be charged a fixed annual fee. 5 This situation can be found in countries where regulation prevents banks from charging customers for the use of cheques, thus leading to an overprovisioning of cheques (Rochet & Tirole, 2000, p.17 fn.23). 6 Rochet & Tirole, 2000, p. 18-20.
  • 30. 30 The main result, including its corollary that there is no equilibrium where a privately set IF is lower than the socially optimal IF, critically depends on an assumed merchants homogeneity. Relaxing this assumption can lead to an underprovisioning of cards, depending on how well informed the cardholder ex ante is of a merchant’s card acceptance. Interestingly, the way cardholders are charged for card usage does matter as it influences merchant resistance: if the cardholder would be charged a perfect fixed and variable fee, with marginal cost pricing for the variable fee, then this would reduce merchant resistance if (and only if) the IF exceeds the issuer cost. Indeed, the main mechanism at work in this model is merchant resistance to accept cards. A higher resistance is likely to bring an equilibrium where private and socially optimal IF coincide; a lower resistance is likely to create an overconsumption of cards with a higher than socially optimal IF. The question whether the IF is ‘too high’ is left an empirical one. Gans and King (2001b, published in 2003) show in a general way that if merchants can costlessly surcharge, then interchange fees will have no real effects irrespective of the level of competition at either the banks or the merchants. Following closely Rochet and Tirole’s main model, Wright (2004b) relaxes the assumption of identical merchants and introduces heterogeneity on both sides of the market, applying a standard Hotelling model of competition (of ‘linear cities’). This accounts for the fact that in some sectors accepting card payments may be more beneficial to merchants than in others. Cardholders are assumed to pay a per-transaction fee as do merchants and neither pays a fixed fee or faces fixed costs. Cardholders are supposed to be fully informed about merchant’s card acceptance policy before they frequent the store, which maximises the merchant’s incentive to accept cards. Cardholders discover their preference for a cash or card payment at the moment of purchase of the good1. The results show that the privately set IF may or may not be equal to the socially optimal IF and either one can be higher than the other. Thus there may be too many or too few card transactions from a socially optimal point of view. No cost- based regulation would be able to restore balance in case of a troublesome diversion as it depends on differences in price elasticities on both sides of the market, competition (among issuers, among acquirers, among merchants and among schemes) as well as costs. Again, this leaves the question of a possible excessive IF an empirical one. 1 This makes merchants’ card acceptance decisions independent rather than strategic complements as in Rochet & Tirole (2000)
  • 31. 31 By this time the literature had rapidly build into several directions focussing on specific topics such as platform competition1, multiple card membership (multi-homing) and usage, market two-sidedness and the influence of merchants and cardholders entering into a Coasean negotiation as to set their own fee directly. Rochet and Tirole (2005) integrate in particular the findings on multiple membership and multiple usage (in the broader context of two-sided markets, not only card payments industry) and reinterpret some of the previously obtained results. In summary: 1) pricing in two-sided market obeys standard Lerner principles2 with a reinterpretation of marginal costs as ‘opportunity costs’3; 2) a market is two-sided if the price structure matters (and much less so the price level), measured by the ability to affect the volume of transactions; however, in the absence of membership externalities, the market can turn one-sided in the presence of asymmetric information between the card-users (merchants and cardholders) if the transaction between them involves a bilaterally negotiated price or monopoly price. A market turns two-sided when there are transaction costs to the bilateral price negotiation or constraints on this kind of price-setting (like an imposed no-surcharge rule) or when there are membership fixed fees (fixed costs). This latter finding is one found in McAndrews and Wang (2008) who develop a formal model different from the ones described above in that they 1) ignore benefits consumers derive from using a payment instrument (the instrument imposes a frictional cost to the purchase of a good), 2) assume a contestable market for merchants, which simplifies the welfare analysis, and 3) take merchants and consumers as non-strategic players (as Baxter). They do take both fixed and variable costs into account. Starting with cash as a benchmark, they analyse the choices of merchants and consumers when offered a payment instrument such as a card that comes at a higher fixed cost but offers lower variable costs. Consistent with empirical studies (id.), they find that large merchants adopt payment cards faster than smaller merchants and will set a price that is lower than cash customers would experience at only-cash accepting merchants; smaller merchants on the other hand may accept card payments or not and those who do, set a price higher than the competing only-cash accepting 1 See in particular Rochet & Tirole (2002) and Guthrie & Wright (2007). 2 The price charged to a side of the market is inversely related to that side’s elasticity of demand. 3 Marginal cost c is in the context of a two-sided market with a platform replaced by (c-vj), i.e. platform cost c per transaction minus vj, the ‘other’ side j’s willingness to pay to interact with ‘this’ side. See for an accessible account Tirole’s lecture in accepting the Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2014; [10], 18 January 2015.
  • 32. 32 merchants; finally, the small merchants will not accept cards. The paper adds the insight that if a merchant serves both cash-preferring and card-preferring consumers, the first group is facing a lower selling price than they would if the store only accepted cash. This insight contradicts the findings and intuitions that price coherence1, or the inability to surcharge, leads to higher selling prices. In another milestone contribution, Rochet and Tirole (2008) respond to Vickers’ (2005) “must take card” argument. The then head of UK’s Office of Fair Trading2, Vickers posed that in the UK retail business it had become a market practise to accept at least the two major credit card brands (VISA and MasterCard) and non-compliance with this market standard could jeopardise the retailer’s business as he would risk losing customers to competing retailers who do. It should be noted that the use of credit cards (and cheques) is very high in the UK compared to other countries in Europe. See also chart 1 and as a further exhibit: “Indeed, the United Kingdom has accounted for more than 75 percent of credit card spending in western Europe in recent years. By contrast, France and Germany each account for less than 1.5 percent” (Vickers, 2005, p.232). Rochet and Tirole operationalise and validate the argument under different models of merchant competition building on earlier frameworks, in particular the ones described above. The possibility that a merchant might accept a card payment even though this would increase his net operating cost3, was already identified in their earlier paper (2000). The reason for the merchant’s decision lies however not in competition, i.e. to win over a customer over a competitor: the property holds even when the merchant is a local monopolist. It lies in the merchant’s improved quality of service by offering the customer extra payment means, which translate into slightly higher retail prices4. The paper presents an alternative to the benchmark for excessive IF-setting used by regulators, which is based on issuer’ costs5. Their alternative is based on the merchant’s 1 A term coined by Frankel (1998) and later adopted by others, which he defines as: “the phenomenon in which the price paid by a consumer for a product does not vary with modest differences in the costs imposed on the merchant by the customer’s choice of brands or payment methods” (p.314). 2 The OFT was at that time involved in a several-year-long investigation into MasterCard’s MIFs. 3 The property that has become known as merchant internalization states that merchants accept cards if, and only if, the merchant fee 𝑝 𝑀 is equal to or less than the sum of the merchant benefit of accepting a card payment 𝑏 𝑀 plus the average net cardholder benefit per card payment 𝑣 𝐶(𝑝 𝐶): 𝑝 𝑀 ≤ 𝑏 𝑀 + 𝑣 𝐶(𝑝 𝐶) This can lead to merchants accepting card payments even when it increases their net operating costs: 𝑝 𝑀 ≥ 𝑏 𝑀 Rochet & Tirole show that this property holds under three important models of competition: perfect competition, Hotelling-Lerner-Salop differentiated products competition and monopolistic competition. 4 This result is also derived in Bedre-Defolie and Calvano (2009). 5 See e.g. the EC COMP/34.579 in 2007 (as mentioned in par. 1.1), the Reserve Bank of Australia in 2003 (Wright 2004a) and the US Federal Reserve by the Durbin Amendment (section 1075, adding section 920 to the Electronic Funds Transfer Act) to the Dodd-Frank Act as of July 2011.
  • 33. 33 avoided-cost: would he have an incentive to refuse a card payment if an incidental customer, like a tourist, wants to pay and is also able to pay in cash? (hence the nickname the “tourist- test”). Since accepting this card transaction would trigger a fee to his acquirer1 - a significant portion thereof constitutes the IF - the test looks for the merchant’s point of indifference between the two means of payment. The merchant fee passes the avoided-cost test if, and only if, the card payment does not increase his net operating cost compared to accepting a payment in cash. If it does not pass the test, then this should be interpreted as an indication of an excessive IF. Rochet and Tirole show that the test yields unbiased results if the objective is short-term total user surplus2, provided that issuers’ margin is constant. However, the IF that would be optimal from a social point of view lies higher in that case and the same holds in the case of variable issuers’ margin. The test yields false positives if the objective is total welfare. Zenger (2011) shows the avoided-cost test and the model of Rochet and Tirole (2000) under perfect surcharging yield the same outcomes. Discussion of the avoided-cost test Before moving on to the last piece of review of literature in this section, I briefly discuss the test and its implications in some more depth. The avoided-cost test seems reasonable at first sight but on second thought appears problematic. The test makes no distinction between a debit and credit card transaction, while the proposed benchmark is cash. As noted before, from a payer’s point of view, cash and debit cards are more or less substitutable payment instruments, but credit cards and cash are much less of substitutes3. More importantly, as the test looks for the merchant’s point of indifference between the means of payment, it thus removes incentives to choose the most efficient one from a social point of view4. If the aim of 1 The fee paid by the merchant to the acquirer is usually called the Merchant Service Charge in the credit card business (MasterCard and others). Rochet & Tirole have consistently used the term “merchant discount”, indicating that the merchant does not receive the full amount, i.e. the sales price for the goods sold, from the cardholder (through the issuer and acquirer) but that the acquirer subtracts (“discounts”) a fee from this amount. From the merchant’ point of view however, he does not get a “discount” in the meaning of common language. To avoid confusion, I use the more neutral term “merchant fee” throughout this paper. The practise to actually subtract a fee from the full amount was abolished by the PSD, as of then the merchant can still be charged such a fee but it has to be invoiced and paid separately. 2 As indicated before, in a two-sided payment market, the ‘one-sided’ consumer surplus includes both the cardholder’s and merchants’ surplus. I will use the term “total user surplus” from this point onwards, as do Rochet and Tirole. The term “consumer surplus” is then reserved for the card and cash using consumers (their surplus), i.e. ex. merchant surplus. In their (2008) model, all consumers hold cards, so there is no distinction between a consumer paying in cash and a consumer paying with a card in the context of welfare analysis. The “short term” reflects a situation without market entry. 3 A similar remark is made by Rochet & Wright (2009) who define the avoided-cost test with respect to “store credit” (credit supplied directly by the merchant to the consumer) instead of cash. 4 A point Rochet and Tirole are aware of (id. p.8).
  • 34. 34 a regulator is to correct a market that is failing in producing economic efficient outcomes, then adopting this test will at best result in maintaining the status quo if the benchmark, cash, is the correct one. But empirical studies show that cash payments are relatively expensive to ‘produce’ compared to electronic payments, debit card payments in particular, and the price to use cash does often not reflect the full (social) costs; see Schmiedel c.s. (2012), Brits and Winder (2005), Leinonen (2011) and Humphrey, Willesson, Lindblom and Bergendahl (2003). Indeed, putting the avoided-cost test to an empirical test, Bolt, Jonker and Plooij (2013) show that it creates the wrong incentives when a country’ social costs of cash payments are rising and those of debit card payments are declining; as is the case for e.g. the Netherlands1. These findings are particularly relevant since the EC accepted the tourist-test as a “reasonable benchmark for assessing a MIF level that generates benefits to merchants and final consumers”2 and the caps proposed in the IFR are the result of MasterCard’s calculations using the test and three empirical cost studies3 which MasterCard implemented with the EC’s consent as of July 2009. A final remark in this sidestep concerns the welfare objective. Most economists, and I agree, argue that competition authorities should adopt a total welfare standard instead of a consumer welfare standard. See e.g. Motta (2004, par. 1.3) and Rochet & Tirole (2003, p.77). Whether courts and regulators favour one over the other is difficult to say, wording such as in art. 101(3) TFEU4 merely seem to induce its explicit inclusion in court’s rulings, not prioritising consumer welfare over total welfare. However, as evidenced by the adoption of the avoided-cost test and the quoted memo, in the case of the IFR the EC seems to adopt a strict consumer welfare standard5. Such a standard is unlikely to solve a market failure, more likely the opposite (see also Rochet and Wright, 2009). In particular, capping the IFs may initially lead to lower prices but in the longer run are likely to deprive the ‘producers’ of 1 Bolt cs. calculate that the tourist-test may allow the IF level to increase from 0.2% to 0.5% of the transaction amount of an average debit card payment. In case of a full pass-through this would increase the notional merchant fee by 233%. For other debit card favouring countries similar results should be expected, see also Leinonen (2011) and Danmark Nationalbank (2011). 2 Press release by the EC, Commissioner Kroes, 1 April 2009, memo 09/143. 3 These concern the cost studies of the central banks of Belgium, Sweden and the Netherlands. The latter is the study published by Brits and Winder (2005), with data from 2002. The Belgian and Swedish study contain data from 2003 and 2002 respectively. See also Bolt c.s. (2013). 4 Certain restrictions on intra-EU competition may be allowed if the agreement / decision / practice “… contributes to improving the production or distribution of goods or to promoting technical or economic progress, while allowing consumers a fair share of the resulting benefit...” [underline font added]. 5 Another indication can be found in the EC’s Impact Assessment: “Rochet and Tirole (2002) find that the privately set interchange fee either is socially optimal (but only total welfare has been analysed)…” (p.102) [underline font added]. Also Rochet and Wright (2009) have noticed and seem to disapprove: “Thus, if regulators only care about (short-run) consumer surplus, our theory can provide a rationalization for placing a cap on interchange fees.” (id. p.6).
  • 35. 35 incentives to innovate. Note that the issuer and acquirer function can be fulfilled by others than banks and the incentives concern both incumbents and potential new entrants, so this may negatively affect banks, other PSPs, merchants as well as consumers. Thus, at the very least the consumer welfare objective should be framed in dynamic terms. Evans (2011) analyses the relationship of IF levels and innovation and argues that drastic reductions of IFs may invert the skewness of the price structure on the two sides of the market, resulting in a reduction in the overall level of innovation in the industry and a discouragement of new entrants1. Circumstantial evidence of this point can be found in Australia for eftpos- transactions, a PIN based debit card scheme and as far as I am aware, the only country where IFs used to go from the issuer to the acquirer. In 2012 however, the IF direction reversed “…to encourage investment in innovation and enhanced functionality for eftpos, so that it can continue to compete effectively in a fast changing payments landscape”2. Whereas Rochet and Tirole confirm some validity of Vickers’ “must take card” argument but conclude that this need not be harmful to social welfare as there is no systematic bias in the price structure, Wright (2012) now arrives at the opposite conclusion. Using basically the same model as Rochet and Tirole (2000, 2008) and Wright (2004b), he finds that a profit maximizing card scheme sets a MIF or price structure that will lead to an overprovisioning (overconsumption) of cards with merchants paying too high a fee. This possibility is present in the earlier work and arises from the fact that a monopolist scheme only focuses on marginal users and not on average users, thus ignoring the effect of IFs on the surplus of the average merchant or consumer. As long as this happens on both sides of the market, there is no particular bias either way. But with the property of merchants internalizing the consumer benefits (meaning they offer their customers an improved quality of service with extra payment means) for which the extra costs are uniformly included in the prices of the goods sold (and in the absence of surcharging, i.e. price coherence), this introduces a systematic bias in favour of cardholders. Therefore, Wright advises a regulatory intervention, not based on any antitrust considerations since the bias is not the result of any shortcomings in competitive market behaviour, including the argument that an IF puts a floor in what competing acquirers can charge merchants. Such an intervention is to be dealt with cautiously, he warns: the main mechanism at work in the model is merchant internalization 1 As Evans (2011) summarises it: “getting innovation right is likely to be far more important than getting prices right” (p.2). 2 [11], 5 September 2014.
  • 36. 36 but to which extent this holds in a particular case, is an unanswered empirical question. As to the proper level of IF, he stresses that economists have reached near unanimity against a fee based on issuer’ costs and instead proposes a direct cap on merchant fees (which should include three-party schemes) or to apply the merchant’ avoided-cost test. This I already discussed. As far as I know, Korsgaard (2014) is the first to discuss interchange fees in a setting that resembles more that of a debit card payments market. As I use this model, more details are given in the next section. Using basically the same model as Rochet and Tirole (2000, 2008) and Wright (2004b, 2012), he shows that the level of the IF only influences merchant acceptance of cards (the higher the former, the lower the latter) and banks will set an IF that exceeds or equals the socially optimal IF. Two critical assumptions underlie this result: 1) all consumers already possess a payment card and do not incur fees per transaction (cash nor card), 2) merchants nor banks face fixed costs. In essence, the two-sided market then turns one-sided; this echoes a finding earlier stated, see Rochet and Tirole (2005). As the only condition under which the banks set an IF equal to the socially optimal IF, is when merchants are assumed to be homogeneous (else the banks set an IF that is higher), Korsgaard advocates a cost-based regulation. His model features consumers, merchants and banks; the latter in their capacity of producers of payment services1. Issuing and acquiring banks are not distinguished individually, which is motivated by the assumption that acquiring banks are assumed to behave perfectly competitively and so the merchant fee consists only of the IF plus the acquiring bank’s marginal cost of producing the card payment2. Banks are assumed to maximise their joint profit, which depends on the fraction of consumers, respectively merchants using and accepting cards. Both consumers and merchants are heterogeneous and they may face fixed costs for accepting payments. A striking result of this model is that the socially optimal IF can turn out to depend solely on costs. More precise, if merchant fixed costs are assumed to be zero, consumers face no adoption costs and in the absence of surcharging, then the socially optimal IF equals the difference in bank’s marginal costs of producing card payments and cash payments. Under surcharging however, the welfare outcomes are ambiguous and the optimal IF no longer depends on costs alone (same is true when fixed costs are introduced). 1 From a modelling perspective this is akin to a proprietary platform, i.e. a three-party scheme. 2 The merchant fee and the IF are therefore assumed to have an one-to-one relationship.