ATM fees have been a source of controversy in many countries. These fees include (a) ‘interchange fees’ that banks charge each other when one bank's customer uses another's ATM, (b) ‘foreign fees’ that banks charge their own customers for using other banks' ATM's, and (c) ‘surcharges’ that banks charge other banks' customers for using their ATM's. In some countries, a single ATM transaction potentially involves all of these: when a customer of bank A withdraws money using bank B's ATM, the customer could pay a foreign fee to bank A and a surcharge to bank B, and bank A could pay an interchange fee to bank B. A central issue surrounding these fees is whether they decrease consumer surplus by allowing imperfectly-competitive banks to extract rents, or increase it by promoting competition for depositors through account fees or ATM deployment.

Banks' setting of ATM fees varies across countries, in part because of differences in regulatory policy. In some countries, including the U.K. and France, banks generally neither charge foreign fees nor surcharges. In others, such as Australia, banks generally charge foreign fees but not surcharges. In still others, including the U.S., banks generally charge both. In addition, regulatory policy has changed over time. Before 1996, U.S. banks were prohibited from imposing surcharges. Recently, regulators have been re-examining their policy decisions; an example is the Australian government's inquiry into whether allowing banks to impose surcharges would reduce the potential for collusion through banks' setting of interchange fees.

Starting in 2007, the JIE received several high quality papers on this topic. In this issue, we are publishing them together as a symposium on the competitive and welfare impacts of ATM fees. (These papers went through the normal refereeing and editorial process, but on an accelerated basis in which deadlines were shorter.) These papers include one empirical piece, ‘How Does Incompatibility Affect Prices? Evidence from ATM's,’ by Christopher R. Knittel and Victor Stango, and two theoretical papers, ‘Paying for ATM Usage: Good for Consumers, Bad for Banks?’ by Jocelyn Donze and Isabelle Dubec and ‘ATM Surcharges: Effects on Deployment and Welfare,’ by Ioana Chioveanu, Ramon Fauli-Oller, Joel Sandonis, and Juana Santamaria.1

The two theoretical papers build on an applied theory literature on ATM pricing and deployment, in particular on work by Massoud and Bernhardt [2002] that investigates surcharging. They advance the existing literature in several ways, most importantly by allowing ATM deployment to be endogenous. Allowing deployment to be endogenous is important for welfare analysis because the main thrust of the argument for why banks should be able to charge fees for ATM usage is that it benefits consumers by encouraging ATM deployment.

To understand the economics of banks' pricing and ATM deployment decisions, as illuminated by Donze and Dubec's and Chioveanu, et al.'s models, it helps to start with a situation in which banks are prohibited from charging customers ATM usage fees, but do charge each other interchange fees. Consider a situation in which all customers are depositors at some bank.2

First consider banks' incentive to attract depositors from competitors. Donze and Dubec show that marginal cost of a customer has three components (see equation (10)). One is the cost of providing the customer with basic banking services. Another is the interchange fees the bank pays on the customer's behalf when the customer uses other banks' ATM's. The third is the interchange fees that the bank foregoes because the customer is a ‘home’ rather than a ‘foreign’ customer. This is an opportunity cost. In equilibrium, account fees increase with each of these marginal costs.

Next consider banks' incentive with respect to ATM deployment. This too reflects three components. One is the extent to which ATM's attract new depository customers. In a situation where ATM usage fees are banned, the term disappears because customers of one bank can access all banks' ATM's at no additional cost. The second component reflects other banks' customers' utilization of the home bank's ATM's – in a situation where ATM usage fees are banned, firms have an incentive to deploy ATM's to increase revenues from interchange fees. The third is through the banks' own customers' utilization of ATM's – a larger ATM network means that these customers will utilize other banks' ATM's less. The bank thus has an incentive to deploy more ATM's to avoid paying interchange fees on behalf of its customers.

There are several things to note about these incentives in a world where banks are banned from charging customers ATM usage fees.

First, banks do not deploy ATM's to attract new depositors. This has good and bad aspects from a welfare perspective. It discourages deployment incentives, holding constant all prices, and this has a welfare-reducing effect. But it also means that firms do not have a strategic incentive to differentiate their product along this dimension in order to weaken price competition.

Second, banks' deployment incentives reflect market distortions, particularly in that interchange fees are set collusively and that customers' choice of ATM's does not reflect their home banks' costs. Banks' equilibrium deployment of ATM's and account fees are distorted as a consequence. Donze and Dubec show that in equilibrium, firms are able to exploit these distortions and profit through high account fees. These high account fees reflect that attracting a depository customer is costly to them – these customers impose costs on the bank ex post through their use of competitors' ATM's. When ATM usage fees are banned but interchange fees exist, it creates a situation where there is very weak price competition in this market in the form of account fees and none in the form of ATM fees.

Now consider a situation in which banks can charge their customers a foreign fee for using another bank's ATM. In both papers' models, in equilibrium this fee equals the interchange fee; a bank's customers bear directly the bank's cost associated with the customers' use of other banks' ATM's, removing an important distortion that exists when all ATM fees are banned. This, in turn, reduces banks' incentives to deploy ATM's, but also encourages them to compete for accounts – customers are not as costly to them as they are in a regime where all ATM usage fees are banned. Interchange fees are still set collusively, but the collusive price is lower because in equilibrium it is tied directly to customers' choice of ATM's. Allowing firms to charge foreign fees is welfare-enhancing, despite the fact that it may lead to lower deployment, because it leads to more price competition especially in the form of lower account fees.

Finally consider a situation in which banks can charge other banks' customers directly for using their ATM's – a situation also examined by Chioveanu, et al. Both papers show how allowing surcharging eliminates an important distortion – interchange fees no longer play a role in limiting price competition. But it enhances banks' ability to differentiate their offerings from those of their competitors: by raising the fee that other banks' customers pay for access to the firm's ATM's, this makes the size of a bank's ATM network a more important differentiator.

In Donze and Dubec's model, the combined effect of allowing surcharges leads equilibrium outcomes to involve more ATMs but higher account fees than in contexts where surcharges are banned. Customers are better off if transportation costs are high, and worse off if they are low. Banks are worse off regardless. Chioveanu, et al. instead investigate how the welfare effects of allowing surcharges vary with deployment costs. They find that surcharges are welfare-increasing when either deployment costs are low or high; competition mitigates potentially distortionary aspects of surcharges when deployment costs are low, and surcharges support the existence of ATMs when deployment costs are high. However, surcharges are welfare-decreasing when deployment costs are in an intermediate range in which banks' strategic incentives to deploy ATMs drive disproportionately drive deployment incentives.

Taken together, these models illuminate how the net benefits of allowing surcharges will vary with the structure of ATM demand and with the cost of ATM deployment. They imply that customers in the U.K. and France would be better off if banks charged ATM usage fees. The implications with respect to surcharging, in contrast, are more subtle: surcharging's benefits depend on the details of product space and deployment costs; they are more likely to benefit customers when travel costs are high, and in situations where the fixed cost of ATM deployment are very low (or very high).

Knittel and Stango's empirical paper investigates firms' strategic incentives with respect to surcharging by examining how banks' depositor fees relate to the size of their and their competitors' ATM network, and how this relationship varies with the extent of surcharging. Their analysis primarily exploits the change in U.S. regulation described above: how do these relationships differ for U.S. banks before and after the prohibition on surcharging was lifted?

Knittel and Stango find that banks' depositor fees become more closely related to the size of their own ATM network, and less closely related to the size of their competitors' ATM networks, as surcharging increases. Furthermore, fees tend to be higher when surcharging is greater. Taken together, this evidence is consistent with the hypothesis that banks use surcharging to differentiate their products from those of their competitors – the strategic effect that is emphasized in Donze and Dubec's and Chioveanu, et al.'s analysis.

Although their evidence is from banks, Knittel and Stango emphasize that the strategic effect that they analyze is a more general phenomenon. By surcharging, banks decrease the degree to which their ATM network is compatible with other banks' ATM networks. The authors emphasize that firms have the opportunity to control the compatibility of their (real or virtual) networks with competitors' networks. For example, similar situations exist in other ‘network’ industries such as telephones, computers, music players, and countless others. Knittel and Stango's analysis thus bears not only on understanding banks' product and pricing decisions, but also firms' product and pricing decisions in network industries more generally, and represents important early work on this topic.

The JIE editorial board hopes that readers find this symposium valuable, and looks forward to publishing similar symposia on different topics in the future.


  1. 1 The Knittel and Stango paper was accepted in late 2007; the other two papers were accepted in early 2009. The JIE thanks Christopher Knittel and Victor Stango for agreeing to delay publication so that it could include their paper in this symposium.

  2. 2 This part of the discussion draws from Donze and Dubec's paper.