نقش هزینه های مبادله در شبکه های ATM

We develop a model to study the deployment of shared ATMs when an interchange system compensates banks for processing non-customers' withdrawals. The interchange fee is chosen cooperatively by banks. We show that a high interchange fee softens competition on the market for deposits but increases competition to process withdrawals. As the former effect dominates the latter, profits are increasing with the interchange fee up to some level. This confirms the presumption that the interchange system can be used as a collusive device by banks. We show that the cooperatively chosen interchange fee does not decrease when the number of banks gets higher or when deployment costs decrease. These results are consistent with the empirical evidence concerning interchange fees. The model also predicts that the size of the shared ATM network exceeds the socially optimal level when the number of banks is large enough.

In many countries, banks share their automated teller machines (ATMs) in regional or national networks: a cardholder of a bank can use an ATM of another bank of the network and make a “foreign withdrawal”. The cardholder's bank pays an interchange fee to the ATM-owning bank to compensate it for the costs of deploying the ATM and providing the service. In general, this fee is set collectively by the network members. A network switch, typically owned by banks, acts as a central device and permits clearing. Several observations have been made about interchange fees.1 First, evidence suggests that they involve a substantial mark-up over the average cost of providing ATM services. In Australia, Great Britain and the USA, the mark-up is close to 100%. Second, interchange fees have remained largely unchanged over the last two decades despite the fall of deployment and communication costs and the entry of new banks into the ATM market. Economists and antitrust authorities consider the joint determination of prices with suspicion. They argue that the cooperative setting of interchange fees explains why they remain sticky above the average cost of providing ATM services (see Balto, 2000). On the other hand, other economists suggest that interchange fees are necessary to guarantee the universal access to ATMs and are mainly a transfer payment to equilibrate the costs and benefits between the cardholders' banks and the ATM owners (see Cruickshank report, 2000). This paper studies the role of interchange fees in ATM networks and addresses the following questions: Why are interchange fees so much higher than the average cost of ATM services? What is the effect of the cooperative choice of interchange fees on banking pricing? Is there over- or under-provision of ATMs compared to the social optimum? To answer these questions we set up a model in which banks set the interchange fee cooperatively but choose the ATM deployment and the prices non-cooperatively. Consumers pay a unique account fee to their bank to obtain both basic banking services and the unlimited access to the shared ATM network. Banks compete à la Bertrand on the deposit market. We show that banks collude on high interchange fees in order to reduce competition for deposits. The consequence is that the size of the shared network is above the socially optimal level if there are sufficiently many banks. The properties of the interchange fee jointly set by banks fit the empirical evidence: the fee exceeds the marginal cost of processing a withdrawal and it does not decrease when the number of banks grows or when deployment costs fall. Let us get an intuition for the results. As there is no ATM usage fees, the shared network is a public good for cardholders. Banks do not deploy ATMs to attract deposits, but rather to process withdrawals and receive interchange fees. When the interchange fee exceeds the marginal cost of processing a withdrawal, each bank is willing to process the withdrawals of competitors' cardholders and to reduce its own depositors' foreign withdrawals. Hence, there must be a positive mark-up of the interchange fee over the processing marginal cost to give banks incentives to deploy ATMs. Nevertheless, the interchange fee is not only a transfer payment, but also a collusive device. To understand why, observe that there are two effects when banks raise the interchange fee. On the one hand, the competition for processing withdrawals is strengthened, more ATMs are opened and banks' deployment costs increase. On the other hand, each bank is less willing to accept depositors because the foreign withdrawals they make induce high interchange outflows. Consequently the competition for deposits is relaxed, account fees increase and banks' revenues rise with the interchange fee. We show that as long as the account fee is below depositors' reservation price, the effect on revenues dominates the effect on costs so that banks' profits are increasing with the interchange fee. Nevertheless, they are not indefinitely increasing. Depositors' reservation price gets higher with the size of the ATM network, but less and less as the network is expanding. When the interchange fee becomes very high, the account fee is bounded from above by the reservation price and does not rise enough to cover the additional costs of ATM deployment: profits begin to decrease with the interchange fee. At the cooperative stage, banks choose the interchange fee to maximize joint profits. We show that the network size is over-optimal if there are many banks or consumers' reservation prices are high. When the first condition is met, the over-provision of ATMs comes from the fierce competition between banks to process withdrawals and receive interchange fees. When the second condition is met, banks choose a high interchange fee to extract consumers' surplus through high account fees and many ATMs are deployed. When the two previous conditions are not satisfied, the ATM deployment may be optimal. In our model, customers pay a fixed account fee to their bank, but there is no ATM usage fee. This pricing scheme is observed in many European countries.2 In other countries like the USA, banks impose surcharges on non-customers for using their ATMs and foreign fees on their depositors for making foreign withdrawals. Massoud and Bernhardt, 2002a and Massoud and Bernhardt, 2002b and Croft and Spencer (2004) consider models more suited to the American ATM market. The American pricing scheme leads to different behaviors of banks and customers. First, the shared network is no longer a public good for cardholders: when several ATMs are located at the same distance from a cardholder, the cardholder prefers her bank's machine. Second, surcharging affects the choice of where to establish deposits: consumers prefer banks with large ATM networks to avoid being surcharged. Third, surcharges enhance the ATM deployment: banks deploy ATMs even if the interchange fee is nil because they receive surcharges from non-depositors. Massoud and Bernhardt, 2002a and Massoud and Bernhardt, 2002b consider a framework without interchange fees in which banks impose surcharges on non-depositors. In Massoud and Bernhardt (2000b), they show that surcharging leads to an over-provision of ATMs. Banks propose a two-part tariff to their customers and a linear surcharge to non-customers. Consequently banks can extract their own customers' surplus more efficiently and they compete for a larger customer base. As customers prefer to choose a bank owning many ATMs, banks over-provide ATMs and raise their surcharges to make competitors' banking services less attractive. In an empirical work, Massoud et al. (in press) show that large banks with many ATMs increase their deposit market shares at the expense of smaller banks that have fewer ATMs: surcharges consolidate the market power of large banks. In our model, all ATMs are created equal from the perspective of consumers and banks cannot attract depositors with their own ATMs. In this case, the over-provision of ATMs is a consequence of the fierce competition to process withdrawals and receive interchange fees. Croft and Spencer (2004) consider a framework with interchange fees, foreign fees and surcharges. They analyze the issue of surcharging or not when two banks have exogenous equal fleets of ATMs. The authors show that foreign fees serve to maximize joint profits while the interchange fee is set at the marginal cost of an ATM transaction and surcharges are banned. In our model, such a result cannot occur: if the interchange fee is equal to the processing marginal cost, banks do not deploy any ATM. Our model is related to the one developed by Matutes and Padilla (1994). They study the incentives of competing banks to share their ATM networks. They show that introducing interchange fees makes sharing more attractive. We start from this result by assuming a single shared network. We extend their model by endogenizing the choice of the interchange fee and the ATM deployment. The paper is organized as follows. In Section 2, we set up the model. In Section 3, we solve the equilibrium in deployment and price for a given interchange fee. Section 4 deals with the cooperative choice of the interchange fee by banks and its welfare properties. Section 5 concludes.

We have analyzed how the choice of the interchange fee by banks affects the ATM deployment and the pricing of banking services. Our work shows that the deposit market and the withdrawal market must be studied simultaneously: a high interchange fee strengthens the competition for processing withdrawals but softens the competition for deposits because banks are less willing to accept customers. The latter effect dominates the former. As a result, banks can use the interchange fee as a collusive device. Interestingly the collusive effect appears although the accounting outflows and inflows of interchange fees are balanced at equilibrium. The model shows that when there is no surcharge, the interchange fee must exceed the marginal cost of processing a withdrawal to provide banks with incentives to deploy ATMs. Hence the existence of a mark-up of the interchange fee over the marginal processing cost is justified to some extent. However the collusive role of the interchange fee explains that this mark-up is generally excessive: the interchange fee and the size of the shared network exceed their socially optimal levels.