The Efficiency Of Multilateral Interchange Fees
The benefit and need for MIFs is topical in many payment areas, especially where card payments are widely used. As MIFs are collectively decided by the industry, it is in the authorities’ interest to ensure that the use of MIFs is beneficial to the general public. This article1 provides a descriptive framework for establishing the upper and lower limit for an efficient MIF window for a beneficial implementation and points towards payment pricing conventions, which would promote efficient use of retail payment instruments.
Cash is still widely used today and different kinds of subsidisations and hidden pricing mechanisms for both cash and cards hinder the correct cost/price reaching the end user. Even small cost savings per transaction and efficient choice of payment instruments will amount to large total savings due to large retail payment volumes. The right economic incentives would promote and speed up these changes towards more efficient payment habits.
A multilateral interchange fee is a commonly agreed interbank fee among all or a large group of service providers for interbank transfers of a given payment instrument (a bilateral interchange is agreed between two service providers). MIFs are used predominantly in the card industry to transfer revenues/charges from the acquiring (merchant’s) bank2 or institution to the issuing (cardholder’s) bank or institution. Sometimes the MIF flow goes in the opposite direction, which can be interpreted as a negative MIF. MIFs have also been used with direct debit and cheque transactions in some countries.
Because the decision on MIF pricing is made collectively between service providers in competition within the same market, this is a horizontal pricing decision among competitors. According to general competition legislation basics, this implied that service providers should prove that these kinds of cooperative decisions are to the benefit of the general public. A beneficial MIF should therefore improve the general efficiency of the payment market and thereby support the use of more efficient payment instruments at the expense of less efficient ones and bring down the overall costs of payments. A MIF would not be needed with completely transparent payment pricing, as the customers would see the full costs of all payment instruments and thereby be in a good position to select the most efficient ones. The use of an MIF can only be efficient when it is used to correct misleading price signals in a market with heavily cross-subsidised payment services, e.g. free or almost free cash services.
The aim of this article is to provide a descriptive framework for analysing issues concerning MIFs and to find out under which circumstances might be of benefit to the general public and how authorities could respond to MIF usage and proposals. The article will mainly focus on card payments versus cash payments, but can also be applied to closely related direct debit MIFs. This requires analysis of cost structures, pricing conventions and subsidisation levels.
Payments are essentially fund transfer services, i.e. funds sent from the sender to the receiver. In an economic transaction, the seller delivers the goods to the buyer and, in exchange, the buyer delivers funds in some monetary form to the seller (when barter of goods/services is excluded). The buyer/payer and payee/seller mutually agree which of the available and commonly accepted transporting mechanisms they will use, both for the goods and the funds. There can be a wide range of supported transportation alternatives, e.g. simply handing over the goods, mailing, home delivery etc. and different kinds of accepted payment instruments (i.e. cash, cheques, cards, pay later with credit transfer, etc). Cash as legal tender is the default payment instrument, if no other is agreeable by both parties. One of the main grounds for establishing legal tender was to create a generally acceptable and reliable means of payment.
There can be three or four party constructions on both sides/legs of the transportation system. In the three-party constructions, the engaged customers use a common service provider and in the four-party case, the customers use their own service providers and these are connected with each other via an inter-service provider network in order to provide an end-to-end reach for the customers, e.g. two local telephone companies connected by the long-haul network. Typically in payments, credit transfer and debit card services operate using a four-party model, while some credit card services employ a three-party model. Figure 1 describes the alternatives, which are independent from each other for the two types of legs.
In a competitive transportation system, the default option would be the four-party solution, in which the payers and payees can bank with any bank and the fund transfer system connects the two service providers in the same way as inter-transport systems connect surface and electronic mail service providers for sending goods physically or electronically. The transportation route is sometimes shortcut in the four-party solutions, when the customers are by chance using the same service provider. The larger the market share of the service provider in a four-party scheme, the larger the opportunity for shortcutting.
However, although some of the credit card services, typically Amex and Diners, which at first glance look like three-party models, they are in fact based on something that could be called a six-party model working in three different phases. In the simple payment transportation of debit card payments among banks, there is a direct fund transfer between the payer and payee’s account. The three phases in the so-called three-party credit card models, with an independent and separated credit card company are:
In these systems, the payments are completely effected only after these three phases, using one detailed level booking and two aggregated payment bookings via the banking system. The credit card schemes can show unitary structures such as Diners and Amex, but also multi-party structures like within some national Visa and MasterCard systems, in which the acquiring and issuing companies use a clearing and settlement service among them. Due to the pay-in and payout phases, the fund transportation associated with credit cards is more complex than those of debit cards.
Over time, cash circulation has changed due to the introduction of ATMs as the most efficient cash dispensing mechanism. Today, it is close to the four-party model of card payments because the majority of consumers withdraw cash from ATMs using different kinds of cards and after payment at point of sale, the cash is deposited directly by the merchants into their bank accounts. Cash circulation has become a closed one-direction loop and has merely an intermediary role in the necessary transportation of funds. Cash circulation requires several manual processes, while in the circulation of the more modern payment instruments, such as cards; many of these processes have been automated via electronification. Electronic payment instruments are considered to be more efficient. Increased payment efficiency will thereby result in decreasing number of cash payments. However, in order for customers to respond positively to the increased efficiency, they should be able to observe the differences concretely as price/cost differences they face, but the wide cross-subsidisation in particular distorts the price signals for cash service costs.
All parties in the transfer of funds chain incur costs for handling the payment transactions. These costs depend mainly on the technology used, the degree of automation, the speed of execution and the security levels. Different service types and payment instruments have different total costs. The distribution of the total costs varies between parties for the different payment instruments. The costs can also vary from customer to customer due to their circumstances, for example an electronic payment is relatively less costly for a customer who has all necessary ITC resources in place, compared to a case where the customer has to make special investments. The basic cost structure is described in figure 2 below.
All payment instruments show this basic and very general cost structure, although the items and cost levels vary somewhat from instrument to instrument. The end result is always the same, in order words, the payer’s account is debited and the payee’s account is credited although the technical transaction flow in for example card payments and direct debits go in most cases from the payee to the payer within the banking system.
The payer incurs costs for initiating the payment and controlling the debit bookings while the payee incurs costs for receiving the payments and controlling the credit bookings. Depending on the terms of the payment instrument, payers face some risk of fraud and risks of lost instruments and may also have to invest in some added security features, such as chip card readers. The payer will probably forego some interest because funds have to be kept in liquid form compared to more favourable investments and interest will also be foregone for the processing period. Modern instruments require different types of ICT resources for telecommunication and computing power, such as broadband communication or telephone lines. Electronic payments can often reduce payers’ physical transportation needs and manual processes. In this model, there are different kinds of buyer benefits, such as less time used for withdrawing cash from ATMs are converted to cost factors.
The payee incurs costs for receiving the payment and modern payments require ICT resources in different forms such as terminals and telecommunication. The payee has to ensure that expected credits are correctly booked and just like the payer; they will also face fraud risks and forgone interests. The longer the float time, the more capital the payee will need to finance the outstanding payments.
The banks incur costs for providing, marketing and supporting the different payment services which they will recover via pricing. All payment instruments contain fraud and error risks, which are shared between the different parties according to the service agreements. Typically, there are service costs for initiating payments in payers’ banks and costs for receiving payments in payees’ banks. Paper-based instruments require more manual resources and other paper-related costs than electronic payments, which require more ITC resources, which are generally lower in cost per item, when volumes grow. Generally, the total service provider costs decrease through increased automation, especially as ICT related costs shrink rapidly while the costs of manual resources increase. The fund-transfer system between banks has cost implications and the banks have agreed amongst themselves how these costs are shared such as automated clearing house (ACH) costs and central bank settlement transfer costs.
It is probably worth pointing out at this point, that the costs of any credit bundled together with a payment service, should be excluded from the payment costs. It is irrelevant for the processing of the funds transfer itself and the associated bookings, whether the payer account is a pure deposit account, a deposit account with overdraft facilities or a credit account of some type. The payer account will always be debited. It has also been stated that the card credits would bring more purchasing power to the consumers, but that will only happen when customers raise their average indebtedness level. A bundled payment and credit service is of interest only, when they together bring synergies that would decrease the costs/charges compared to acquiring them separately. Just for comparison, a 10% per annum interest rate for consumer credit would translate, without interest on interest, to a 45 day interest rate of 1.25%. In order to focus the analysis on the payment function, the credit function of credit cards is unbundled and both credit and debit cards are viewed as general payment card services.
For all service providers in the payment chain described, including the seller, capital costs are the last item on the list of incurred costs. This is assumed to include a set return on own capital i.e. a reasonable profit margin. These costs are assumed to be long-term average costs, which can be used to determine the long-term payment instrument cost-benefits for the economy. At full-cost pricing, the charges to the payer would amount to the sum of all service provider costs (the direct charges of the issuing bank and the acquiring banks’ merchant fees plus the merchants’ own costs included in the merchant surcharges to the payer). When there is sufficient competition in the market and transparent pricing is applied, the customer payment charges would be close to this total.
Although, these cost items seem easy to define, there are major difficulties in establishing their true and concrete amount. The exact costs will depend on the level of technology, straight-through-processing support, efficient standards, automated interfaces, volumes to cover fixed costs and processing structures/models. Costs will also vary across banks and customers due to their individual circumstances. The major part of these costs is independent of the value of the transfer meaning that the number of bits in the transaction value field does not affect the processing costs. Really high-value payments will require improved security, but the security requirements for standard consumer cash and card payments are basically the same over the whole range, which translates into concrete security investments servicing all payment sizes such as the EMV-chip on new cards.
This article will analyse the costs of this general level under the assumption that card payments are more efficient for the society and in average to payers than cash payments at least above a given threshold, because going to an ATM machine and withdraw cash for one large payment will be more costly than just using the same payment card directly at the point of sale at the merchant. Although, there may be the case under this threshold, in which the same ATM withdrawal can be used for a large number of small payments and the costs of making this bunch of small cash payments would be more efficient than making the same bunch of individual small card payments. This view is supported by recent payment cost studies made by central banks, although it should be acknowledged that current biased cash and card price signals result in biased volumes leading to non-optimal average costs. (The central banks of Belgium, Netherlands, Norway and Sweden have made recent empirical studies on actual payment costs in their respective countries.) This threshold is probably volume dependent and increased card usage would probably lower this threshold, as the fixed costs would be distributed on a large number of payments, while the average cash costs would increase of the same reason.
Payment services show a complex charging structure in which charges to cover fund transporting costs are transmitted in several steps via several service providers before they reach the end-customer (payer/buyer), who is ultimately paying for the funds transfers (see figure 3). There is also a tradition of hidden costs or cross-subsidisation, which limits the price/cost transparency for the end-customer to select the cost efficient payment instrument for a particular payment.
The fund transfer systems mostly function as industry utilities based on cost sharing among participating banks. Some card branding infrastructures (e.g. Mastercard and Visa international) have recently turned or are in the process of turning into listed companies, which may change their pricing policies over time away from pure cost sharing. In most cases the costs are probably shared neutrally among different payment instrument, when the same interbank system processes several different instruments in parallel (for example ACHs providing in parallel clearing of card, credit transfer and direct debit transactions). There are three general organisational models for cash centres; public utilities (mostly central banks), private utilities and private competing companies, which have different pricing (and subsidisation) policies for their cash handling costs.
Between the payer and payee’s bank can a so-called ‘share’ convention be applied, which implies no interbank charges, but both banks cover their costs via their own charges from their customer (payer or payee). This convention is applied to SEPA credit transfers and in many debit card schemes. For credit card payments in particular and some direct debits, banks have agreed on a MIF, which transfers a part of the payee’s bank revenues to the payer’s bank, in order to cover the payer’s bank’s payment instrument costs. There are different MIF rules in use such as pure ad valorem, pure fixed transaction fee or a mix of these and with different parameters. (The variations in MIF levels are substantial in the card industry according to the EU Commission Report on the retail banking sector inquiry.) MIF charges limit competition, as customers are unable to bargain for this part of the service charges, which is set by competitors in cooperation. MIF charges are not applicable to interbank cash transfers. ATM networks generally charge the account keeping bank a service charge for using the ATM facilities and this may sometimes be labelled as an interchange fee when banks share their private ATM networks, but it is essentially a service fee for using ‘outsourced’ ATM services for serving one customer.
The payers’ banks use different forms of transparent payment service charges for different instruments for example yearly or monthly fixed fees, transaction based fees or package fees for several instruments. Payer’s banks’ also use hidden pricing mechanisms among which float (defined as the non-interesting bearing time due to processing time or extra value days to the bank) is the most frequently used. Float increases the interest benefits of the bank, while it reduces the interest benefits of the payer. The payers’ bank charges can for some payment instruments be negative as they are paying net rewards to the payers. The relation between payers’ banks’ costs and revenues (including MIF revenues) result in a cross-subsidisation or profit level. When the charges are lower than the cost, the bank cross-subsidise the service from other revenues and in the opposite case profit margin of the bank increases. As a general observation, without supplying any strong evidence, it may be stated that banks have historically provided cash services using wide cross-subsidisation, while card payments have been charged for in different ways such as using MIFs and merchant fees and often with high margins especially for credit cards (see for example, EU Commission Report on the retail banking sector inquiry).
The payee’s bank charges the payee for card payments a fee, which is generally the merchant fee. This fee can constitute a mix of fixed, transaction and/or ad valorem fee (and also here the EU commission report shows large variations). As the payment handling costs for card payments are mainly transaction related, it would speak for transaction-based pricing instead of value-based pricing. Payees face also hidden float charges, which can be considerably large due to lengthy crediting times, which in some countries can extend up to 7-14 days for credit card transactions, but is generally only a few days for debit card transactions. When there is an agreed MIF among the banks, the payees’ bank have an interest in transferring it onwards to the merchants in order to reduce the cross-subsidisation effect, as the MIF is a cost item for the payee’s bank. The payee’s bank will, in the same way as the payer’s bank, either cross-subsidise or make a profit of the payment services (according to the EU Commission report the acquiring banks are operating closer to cost, as these institutions have not been making considerable profits, which can be due to merchants’ refusal to accept card instruments with too high a mark-up on the MIF). Traditionally payees were granted free cash deposit services. There are no Europe-wide studies on merchants’ cash charges. However, these are probably also quite varying and have probably increased during the recent years resulting in decreased cross-subsidisation level on the merchant side as banks have started an informal ‘war on cash’ project.
Lastly, the payee/seller charges the payer for the payee’s payment costs (both the internal payee costs and the charges levied by the payee’s bank). This can be done transparently by surcharging the payment costs on the prices of goods/services or by embedding the costs non-transparently in the prices of goods/services. As all goods/services will be paid in the end, the seller will never cross-subsidise payments from other services, but will only cross-subsidise among different payment instruments, when the average payment costs are embedded in the prices of the services/goods (see below for details).
In this general model, it is assumed that a long-term average charge and subsidisation level can be employed for calculating payment instrument benefits to the whole economy.
The basic principle in a consumer-to-merchant and customer-to-supplier relationship is that the consumers/customers pay eventually all the merchant transportation costs (both for the deliveries and payment funds) either separately priced via surcharges or embedded in the product prices of the merchant.
Take for example a supermarket offering disposable carrier bags. They come in different sizes and materials and the merchant can provide them free of charge or price them separately, according to size and type. This depends generally on the retail practice in the area and the type of retail unit. When these are priced separately i.e. surcharged, customers will have an incentive to economise on disposable bags in different ways and the opposite is true when bags are provided free of charge, i.e. embedded in the prices of goods. It would of course be in the interest of the bag industry and an individual bag factory to be able to enforce a no-surcharge rule on their bags in order to promote the use of disposable bags. If this were the case, it would be seen as abuse of market position; the merchants should, according to general rules of competition, be free to state their own prices and terms of delivery for the goods and services they provide, without interference or requirements either by manufacturers or wholesalers.
The traditional situation in the payment industry has been that the costs of cash payments have been embedded in the merchant prices. The acquirers of card payments often have a no-surcharge rule in their agreements, forbidding merchants from making a surcharge of the costs of the card. This results in merchants mostly embedding the payment costs in their prices of the goods/services. This will result in cross-subsidisation among payment instruments, when the costs of paying are different depending on the instrument choice (see figure 4). The authorities have recently forced in some countries the acquirers to abandon no-surcharge rules from the merchant agreements.
In our simplified example, we have two payment instruments with costs of, on average, 20 cents versus 50 cents per transaction for making transactions to a value of €100. In order to cover costs, the merchant has to select the price of his goods/services along the straight line depending on the market share of the payment instruments. The higher the share of the more costly instrument, the higher the mark-up on goods must be. If the merchant has the feeling that he will make profits by having larger demand at lower costs, he can refuse the more expensive instrument and thereby sell at the price level €100.2 for goods selling at €100 without the payment mark-up. When the merchant has made a payment instrument limiting choice then the payer does not have a choice, but can only use one payment instrument at that merchant.
If merchants perceive in general that the more efficient instrument (the one with lower total costs) as the more costly for them (sum of merchant fees and internal costs) and therefore limit its acceptability, then the current pricing conventions supports inefficiency. The volumes of inefficient instruments will be too high and the volumes of more efficient instruments too low.
The difference of 30 cents, for example, might not at first sight sound important. However, there are approximately 60 billion non-cash payments made yearly in the EU15 area. There are no exact figures for the number of cash payments, but they are on average at least double the volume of non-cash payments. Based on the experiences of low cash usage countries (for example the Nordic countries), the number of cash payments can easily be halved with the right economic incentives. Savings in the range of 10 cents per transaction for 60 billion cash payments turned into more efficient instruments would result in total savings of €6bn per year in the EU area. Unfortunately, these levels of saving potential get much less public attention than the robberies of cash-in-transit deliveries amounting to only to some millions in lost values. It would be important to get more attention to this potential in order to break the tradition of perceived free payment services, which maintain inefficient payment habits.
On the basis of the cost structure presented in figure 2 and the pricing structure presented in figure 3, different probable cost and pricing scenarios can be constructed. The actual costs, visible charges and total bank charges for different pairs of payment instruments are analysed. The analysis is constructed as a comparison between two instruments, which can be interpreted as cash and cards, but it can be generalised to apply to other pairs of instruments or even to several instruments. Each figure in the following sub chapters shows bars representing the size of average per transaction costs and revenues assuming that the infrastructure is in place for both instruments. The difference in actual costs shows the benefits/cost savings to the economy, the difference in visible charges shows the incentive for payer choice and the difference in total charges shows the payment industry revenue differential. The situations in the figures are exaggerated for visibility, but the analysis will be general and will not depend on the actual size of the differences. The total costs for instruments 1 and 2 are the same in all scenarios; it is only the pricing conventions and the internal cost distributions that changes. Instrument 2 should be the one selected, when total costs are to be minimised.
The first introductory scenario is a straightforward transparent cost-based pricing situation. The seller/merchant surcharges the buyer his own payment costs plus the cost charged by his bank and the buyer’s bank charge its costs directly to the buyer.
In this scenario, the buyer will see all costs including his own and can make a decision based on the true costs and will clearly see, which instrument is the most beneficial. This example would point towards all customers starting to use instrument 2, but with real examples there will be different kinds of situations where the buyer’s costs and seller’s costs are different resulting in opposite choices. However, this will result in an optimal mix of card and cash payments based on their visible costs. When costs are completely transparent the buyer will have the economic incentive to select the most efficient instrument for all situations. All service providers will gain a reasonable profit (as capital costs were included in the listed cost items in figure 2).
In scenario two, banks subsidise heavily both cash and cards. The cost distribution in scenario 2 is such that all stakeholders will have a win-win situation by selecting and supporting instrument 2b. A win-win situation is defined as a case in which all stakeholders will experience lower costs for instrument 2. There is no MIF applied, but the SHARE option is used, which implies that the payer pays the payer bank’s fees, if any, and the merchant pays the payee bank’s merchant fee, if any.
When banks subsidise and merchants embed their payment costs, the buyer will have a very limited view of the total costs. Banks subsidising costs implies that other customers than banks’ payment customers will cover the payment costs. Embedded merchant pricing implies that merchant costs are averaged among payment instruments (as described in figure 4). In order for the buyer to select the more efficient instrument the relationship between the visible charges should remain in favour of the instrument with lower total cost. In scenario 2 the banks only charge the seller/merchant for the cost of instrument 1, which is often the case for cash. For instrument 2b they charge both a seller fee and a (small) payer fee in our example. In order for the more efficient instrument to be selected, the following conditions should be in place:
This scenario is probably close to the current cash versus debit card situation in those countries where banks have priced debit cards carefully in comparison to cash so that