Ebook Two-sided network effects, bank interchange fees, and the allocation of fixed costs
Payment systems, such as credit-card networks, have features that make economic analysis particularly challenging, for regulators as well as for the theorist. In order to process a transaction between customers of different banks, the banks need to cooperate extensively. Sometimes, this means accessing each others’ facilities or customers. Cooperation may also be necessary in order to reap economies of scale. For example, central parts of the payment system will often provide services to many banks and may be jointly owned by its customers. In addition, in order to achieve optimal network effects, a fee structure that effectively taxes one type of final customers (e.g., merchants that accept card payments) in order to subsidize another category of customers (e.g., card-holders) may have to be implemented. The latter, in turn, may necessitate a multilateral pricing agreement between the banks.
In general, regulatory authorities have well-founded reasons for being suspicious of close cooperation between competitors, in particular concerning pricing decisions. Competition between independent firms will foster efficiency and tends to bring prices down to costs. In the provision of payment services, however, completely independent competition will not be a feasible alternative. The best available option may be competition at the retail level in combination with cooperation at the upstream (system) level. On the other hand, close cooperation at one level may give the banks the opportunity to cleverly design system fees and multilateral fees in such a way that downstream collusion is induced. This could, for example, be achieved by raising the appropriate marginal costs, so that incentives are created for the banks to raise final-customer prices and so that excess profits are generated elsewhere in the system. Considerations of this type has drawn the attention of regulators, resulting in a substantial amount of regulatory activity. The EU Commission, for example, forced Visa and (indirectly) Mastercard to reduce their multilateral interchange fees (see below for a discussion of interchange fees; see Bergman, 2003 and Chakravorti, 2003 for references to cases).
For much the same reasons, payment systems have spawned a considerable economics literature in recent years. In a wider context, these contributions can be seen as a part of the growing literature on two-way network effects (Rochet and Tirole, 2003; Armstrong, 2004). A key insight of this literature is that network effects between two different types of consumers have to be analysed in a system-wide context.
A specific example of a market with two-sided network effects is the payment card market (the market for debit and credit cards). Such markets have four types of participants, as shown in Figure 1: cardholders (or customers), issuers, acquirers and merchants. Cardholders use cards to pay for goods and services provided by merchants. The cards are provided by issuing banks, while acquiring banks provide services to merchants. When a card payment is registered with a merchant, the acquiring bank charges the issuing bank, which in turn charges the cardholder’s account. For its services, the acquiring bank subtracts a fraction, the so-called merchant fee, from the payment to the merchant’s account. Typically, the acquiring bank will have to pay a fee to the issuing bank, the so-called interchange fee, which is therefore subtracted from the payment from the issuer to the acquirer. The merchant fee and the interchange fee can be either a percentage of the transaction value, or a fixed fee. In most cases, the cardholder pays an annual fee, while in some cases, he or she will (also) have to pay per-transaction fees. Credit cards may be free of charge for the cardholder, in particular in the USA, as long as the accumulated debt is paid monthly.
Relative to cash payments, card payments potentially generate benefits for customers, as well as for merchants. For cardholders, these benefits will increase with the number of merchants that accept cards, while merchants will be more willing to accept a card that has a larger base of cardholders. This interdependence creates two-way network effects, which will be discussed more extensively in Section 2. In the presence of network effects, the competitive equilibrium may be inefficient.
The interchange fee, which is set multilaterally by the banks, provides an instrument for improving efficiency. A higher interchange fee tends to reduce cardholders’ fees and to increase the merchant fee. The first-generation literature on two-way network effects in payment markets (Baxter, 1983, Schmalensee, 2002) analysed these issues under the assumption that cardholders’ and merchants’ demand for card-payment services reflect the intrinsic benefits such as convenience and safety that they derive from card payments (relative to cash payments). A typical result is that there will be under provision of card services, because private agents do not internalize positive network effects that accrue to other agents.
An important insight, however, was that merchants with market power (with a positive price-cost margin) may have strategic reasons to accept cards. By accepting cards, they will attract customers away from other merchants that do not accept cards. Conversely, if they do not accept cards, they may loose customers to other merchants that do. These strategic motives do not correspond to social gains. It follows that there may potentially be over provision of card services (Katz, 2001). In the second-generation literature (e.g., Rochet and Tirole, 2002), these strategic motives are explicitly accounted for by incorporating merchant competition in the models (see further Section 10).
This paper is based on the first-generation modelling assumptions. Although possibly important, merchant market power greatly complicates the analysis: there may still be insights to be drawn from a simpler modelling approach. In addition, merchant market power may not always be significant or, perhaps more plausibly, consumers may be uninformed as to whether the merchant accepts card payments or not when they choose which store to patronize. A specific reason to go back to a simpler model is to make it easier to introduce fixed system costs in the analysis. Typically, the literature assumes that there are only marginal costs associated with payment systems, while in practice there will be substantial fixed costs. For example, there will be fixed development costs and some of the costs for setting up equipment for communication, central processing and data storage will be fixed. From a policy point-of-view, the allocation of fixed costs is an important issue, possibly with substantial implications for the barriers-to-entry into the industry.
The outline of the paper is as follows. Sections 2 discusses network effects and Section 3 sets up the model. In Sections 4 through 8 the basic model is applied to different market structures: perfect competition and welfare maximum (Section 4), second-best regulation of the interchange fee (Section 5), bilateral monopoly (Section 6), a proprietary system (Section 7) and a one-sided monopoly (Section 8). Section 9 makes welfare comparisons between these market configurations. In section 10, some aspects of oligopoly interaction are introduced and further references to the literature are provided. In section 11, central system fees are introduced. After that, the effect of system fees are analysed, under the additional assumption of free entry (Section 12) and no-entry Cournot competition (Section 13). Finally, section 14 concludes.
Download
PDF Ebook Two-sided network effects, bank interchange fees, and the allocation of fixed costs
Posted in :
