Chris Stewart, Iris Chan, Crystal Ossolinski, David Halperin and Paul Ryan

RDP 2014-14

The Discussion Paper series is intended to make the results of the current economic research

within the Reserve Bank available to other economists. Its aim is to present preliminary results of

research so as to encourage discussion and comment. Views expressed in this paper are those

of the authors and not necessarily those of the Reserve Bank. Use of any results from this paper

should clearly attribute the work to the authors and not to the Reserve Bank of Australia.

The contents of this publication shall not be reproduced, sold or distributed without the prior

consent of the Reserve Bank of Australia.

ISSN 1320-7229 (Print)

ISSN 1448-5109 (Online)

Chris Stewart, Iris Chan, Crystal Ossolinski, David Halperin and Paul Ryan

Research Discussion Paper 2014-14

December 2014

Payments Policy Department Reserve Bank of Australia

We would like to thank the entities that participated in this study. We would also like to thank David Emery, Darren Flood, Tony Richards, Stephanie Bolt, Tim West, Clare Noone, Fumiko Hayashi, Michele Bullock, John Simon and Carl Schwartz for useful comments and suggestions. The views expressed in this paper are our own and do not necessarily reflect those of the Reserve Bank of Australia. Any errors are our own.

Authors: ossolinskic, halperind and ryanpe at domain rba.gov.au

Media Office: rbainfo@rba.gov.au

This paper examines the costs borne by financial institutions, merchants, and consumers in making, facilitating and accepting consumer-to-business payments. It examines the resource costs incurred by these sectors, how these have changed since 2006, and how fees and other transfers determine which sectors ultimately bear these costs. It also examines how resource costs vary at different transaction sizes and, for merchants, how costs differ between small and large entities.

The results suggest that the resource costs of the payments system have fallen as a per cent of GDP since 2006. On a per transaction basis, direct debit remains the lowest-cost payment instrument while cheques remain the most expensive. At the point of sale, payments using cash, eftpos and contactless MasterCard & Visa debit cards have broadly similar costs for transactions under about $20; above $20, eftpos is the lowest-cost payment method. The results indicate that the relationship between resource and private costs varies significantly across instruments. The greater share of the overall cost is borne by merchants. The consumer undertaking a transaction typically pays a small proportion of its cost; consumers face a similar cost for credit card payments as for debit card payments despite the higher cost of credit cards to the economy. Finally, the results suggest that small businesses incur higher costs than large merchants.

JEL Classification Numbers: E4, G2, L2

Keywords: banks, consumers, financial institutions, merchants, retail payments, surcharging

i

1. Introduction 1

2. Related Literature and Scope 4

2.1 Measurement of Costs 5

2.2 Resource Costs, Private Costs and Transfers 6

2.3 Fixed and Variable Costs 7

2.4 Institutional Coverage and Data Collection 7

2.5 Payment Instruments 8

3. Methodology 10

3.1 Data Collection and Sample 10

3.2 Caveats 12

4. Resource Costs of Payments 14

4.1 Overall Resource Costs 14

4.2 Account Maintenance Costs 21

4.3 Cash 22

4.4 Personal Cheques 26

4.5 Credit and Debit Cards 28

4.6 Direct Debit and BPAY 32

5. Private Net Costs 33

5.1 MasterCard & Visa Credit Cards 36

5.2 Debit Cards 37

5.3 Cash 38

6. Influence of Payment Size 38

7. Small Businesses Costs 41

7.1 Acceptance 41

ii

7.2 Costs 42

7.3 Surcharging and Discounting of Payment Methods 44

8. Conclusion 45

Appendix A: Additional Detailed Results 48

Appendix B: Payment Activity in the Sample 58

Appendix C: Costs for Consumers in Making Payments 61

Appendix D: Description of Survey of SMEs 64

References 66

iii

This study provides comprehensive estimates of the costs borne by merchants, financial institutions and individuals in the use of different retail payment methods. The absolute and relative costs involved in making and receiving payments are important as they influence the decisions of these sectors and, therefore, the mix of payments in the economy. They are also important considerations for policymakers when trying to understand the efficiency of the payments system. These costs are typically not transparent to policymakers or end users of payment systems.

Given the significant changes in technology, payment functionality, issuing arrangements, pricing and payment use patterns that have occurred in recent years it is timely to refresh this work with new cost data. This study extends the earlier work in a number of directions. In particular, it:

• captures the payments costs of a wider range of merchants, including small businesses

2

• collects information on a wider range of payment instruments, for example so-called ‘companion credit cards’ (American Express credit cards issued by the major banks to customers alongside a MasterCard or Visa credit card)

• separately identifies the costs of contactless card transactions given their rapid growth and implications for tender times and other costs.

As with the earlier Reserve Bank work, the study has benefited from close cooperation with a wide range of financial institutions and merchants. These entities helped determine what information could be reliably obtained, provided a large amount of data to the Reserve Bank for analysis, and worked with the Reserve Bank to improve the accuracy and consistency of these data.

The key findings of the study include:

• The aggregate resource cost incurred by merchants and financial institutions in receiving payments from consumers is estimated to be about $8.4 billion in 2013, or about 0.54 per cent of GDP. Financial institutions incur the majority of these costs. Around one-third of costs are incurred by merchants, with tender time (the time taken at the till to process the payment) being the most significant component.

• The aggregate costs associated with consumer-to-business payments are estimated to have changed little in nominal terms since the 2007 study, and to have fallen as a per cent of GDP. The fall primarily reflects that per transaction costs have fallen across most instruments. Marginally offsetting the fall, the shift towards greater use of more expensive instruments has worked to raise the cost of the payments system.

• To conduct a comparison of the cost across instruments, the cost of maintaining accounts (which may facilitate payments made with multiple instruments) is excluded and only the direct cost of making a payment using that method is considered. On this basis, cash, eftpos and contactless MasterCard & Visa debit transactions have broadly similar resource costs for transactions of under about $20. Above $20, eftpos is the lowest-cost payment method. At the average transaction size for each instrument, MasterCard & Visa debit card payments are

3

more resource intensive than eftpos, while credit card transactions are the most resource-intensive card payment method even when excluding the costs of credit and rewards (Figure 1). Of all methods considered in the study, direct debit remains the lowest cost, while cheques remain the most expensive.

Per average-sized transaction for each payment method

Note: Payment function only

Source: Authors’ calculations based on survey data

• The aggregate and relative costs associated with card payments are changing considerably with the advent of contactless payments. Contactless card payments are estimated to incur 10 to 20 per cent lower resource costs than a comparable contact-based card transaction.

• Once fees and other transfers between sectors are included, the burden of private costs across sectors differs from that of resource costs and varies significantly across instruments. The majority of private costs are incurred by merchants and consumers who typically pay a net transfer to the financial sector to use payment services, although merchants may pass these costs on to consumers in general via the prices on their goods and services. Across instruments, the private cost to

4

consumers is relatively similar despite large differences in resource costs. On average, the private cost to consumers of using a credit card is similar to that for a debit card despite the higher resource costs incurred for credit card transactions. Although consumers pay fees to hold credit cards, they also receive significant incentives to use them to make purchases, due to rewards points and the interest-free period.

• Results from a survey of small and medium enterprises (SMEs) suggest that while the ranking of the private costs of instruments is similar to that for large businesses, the private costs faced by SMEs are higher. In part, this is because SMEs do not benefit from the economies of scale that can be achieved by large merchants due to their larger payment volumes. In addition, merchant service fees are higher for small businesses.

The rest of this paper proceeds as follows. Section 2 discusses the literature on payment costs. Section 3 outlines the scope of this study and details the estimation methodology. Section 4 presents resource costs of payments in Australia, both in aggregate and by instrument. This is supplemented in Section 5 by a discussion of how these costs are borne by different sectors. Section 6 examines how resource costs vary at different transaction sizes and Section 7 focuses on the private (gross) costs to SMEs of accepting payments. Section 8 concludes.

5

2 For more discussion on the benefits and limitations of cost studies for central banks, see Hayashi and Keeton (2012).

3 Arango and Taylor (2008) is one exception, looking at the marginal costs of different payment methods for merchants in Canada, along with merchant perceptions of costs, reliability and risk. Another is the paper by Garcia-Swartz, Hahn and Layne-Farrar (2006), which looks at benefits and costs.

In general, studies undertaken since the Bank’s 2007 cost study are similar in scope and methodology. In all cases, estimating the benefits of payments has been beyond the scope of these studies given the difficulties of defining and measuring these benefits. Any differences in scope or coverage have generally reflected national circumstances. For example, most countries participating in the ECB study did not include costs associated with cheques given they are generally not used extensively for retail payments in most of Europe.

6

One respect in which the measurement of costs differs across studies relates to how prescriptive each study has been around the allocation of costs that might be common to payment instruments and other business functions. Allocation is required because much of the infrastructure that supports payment transactions also facilitates other functions for financial institutions and merchants, such as managing statements and invoicing. To address this, some studies have taken a prescriptive approach as to the types of costs to allocate to payments in order to improve consistency across respondents, while others – including this study – have left these allocations to responding institutions to better account for differences in the structure of each.

While resource costs have been the primary focus of international studies, analysis of institutions’ private costs has also been considered in a number of studies (e.g. Brits and Winder 2005; National Bank of Belgium 2006). Some studies have also combined this with an analysis of fixed and variable costs, thereby allowing for a consideration of how private incentives may change at different payment values and how these compare to socially optimal outcomes (Danmarks Nationalbank 2012; Segendorf and Jansson 2012).

7

Payment costs can be categorised into fixed and variable components. Fixed costs are generally infrastructure-type costs that would be incurred regardless of the number of payments made, while variable costs are those that depend on the number or value of transactions undertaken. The ability to distinguish between fixed and variable costs permits comparison of the cost of particular payment types as the transaction value varies. It can also inform the extent to which different payment instruments benefit from economies of scale.

The categorisation of costs as fixed or variable will differ to some extent between different institutions. This study asked financial institutions and merchants to indicate whether different cost items are fixed or vary with transaction volumes and values. Different cost items were then allocated as fixed or variable using this information.

Institutional coverage is broadly similar across different studies of payment costs. Studies conducted by central banks tend to rely on direct surveys of the costs of financial institutions and merchants. Some studies collect additional data directly from companies that provide services to these entities, such as cash-in-transit companies, to examine more detailed aspects of resource and private costs (Segendorf and Jansson 2012). This paper follows the same approach as the Bank’s 2007 study by directly surveying financial institutions and merchants, and proxying the resource costs of their service providers by using the fees paid to them, paying careful attention to avoid double counting between resource costs and transfers.

8

(for example, see comments by Gresvik and Haare (2009)). This study includes 17 large merchants, and has used a survey of small merchants (with around 260 respondents) to better understand any differences in costs faced by these businesses.

The particular payment instruments included in each cost study have been determined by their relative importance to that country’s payments system. For example, cheques tend to be excluded from studies conducted by countries in which cheque use is low. Additionally, some studies focus solely on point-of-sale payments to the exclusion of billing and remote payments. Due to the wide variety of payment instruments used in Australia, this study’s coverage is in line with the most inclusive overseas studies. In particular, this study considers the costs associated with cash, debit and credit cards, cheques, direct debit and BPAY

9

4 The study also collected information on payments made through agency arrangements. Agencies handle payments on behalf of merchants. This can be electronically, such as PayPal, or physically, such as at Australia Post branches. Merchants directly incur low resource costs of a few cents per transaction when employing an agent. However, the agent can charge relatively high fees – often of up to 2 per cent of the transaction value – for providing this service. Market concentration prevents the disclosure of further results.

Reflecting recent developments in the Australian payments landscape, the study also examines the costs of new products and new methods of authorisation and/or authenticating retail payments. Recent innovations have included:

• The transition from using the magnetic stripe on cards to using chips for the storage of card details and transmission of these details to the terminal.

• The introduction of contactless payments functionality, which allows the wireless transmission of card details.

• The move from signature authentication to PINs for American Express, Diners Club, MasterCard and Visa cards (eftpos has always been PIN-only). In addition, lower-value payments often no longer require PIN authentication.

• The increased issuance of dual-network debit cards by the major banks.

• An increase in the number of institutions offering companion American Express cards.

These innovations will have directly altered the costs of accepting and making payments. For instance, some of these developments will have directly affected terminal and tender time costs for merchants, while others will have altered card production and fraud costs for financial institutions. These innovations may also have indirectly altered costs by affecting both the economies of scale for system participants and potentially the nature of competition.

10

This study estimates the resource cost of consumer payments in the Australian economy in a manner similar to the Bank’s 2007 study and international studies. Financial institution and merchant costs were directly surveyed by the Bank and consumer costs are estimated based on the cost of a consumer’s time to make payments. This study focuses on average costs, which take into account the cost of infrastructure supporting payment instruments. The results shown are for weighted-average costs across entities, although the use of median costs leads to the same conclusions about the relative cost of instruments.

Consistent with previous studies, the current study focuses on payments by individuals to merchants rather than on business-to-business payments and/or person-to-person payments. The payment instruments covered by the study are estimated to account for nearly all of the number and more than 95 per cent of the value of consumer payments to businesses (Ossolinski, Lam and Emery 2014).

The majority of costs are measured directly by surveying financial institutions, merchants and, in the case of the costs of currency production, the Reserve Bank and the Royal Australian Mint. In a few cases, costs are estimated indirectly based on publicly available information or the fees that survey participants pay (non-surveyed) third parties for payment-related services. For example, the resource costs of transporting cash was not directly collected from cash-in-transit companies but instead proxied by the fees paid for this service. While estimating some costs based on fees will overstate the cost of these payment services given that the profit margins of these third parties will also be captured, the size of these indirectly estimated costs is generally small in both absolute size and relative to other payment costs.

11

5 Around half of financial institutions and merchants that were asked to participate in the study accepted. Those that declined cited competing demands on their time or an inability to provide sufficiently detailed payment cost information.

• Financial institutions ranged in size from large banks to specialist service organisations. In aggregate, they reported nearly 30 million transaction accounts and 10 million personal credit card accounts, capturing the vast majority of all such accounts in Australia.

• Merchants were selected to cover a wide range of consumer expenditure categories. Ten large merchants were retailers predominantly collecting payments at the point of sale, whether in a supermarket, department store or general retail environment. Over the twelve-month sample period, these retailers reported total sales of $109 billion, about a third of the value of retail sales in Australia over 2013. The other seven merchants were billers that predominantly receive payments remotely – with little point-of-sale activity – for insurance, telecommunications and utilities. These billers reported total sales of $46 billion during the sample period, which represents a significant share of total household consumption on insurance, telecommunications and utilities of about $120 billion.

• Small merchants were recruited through a number of merchant associations, with more details provided in Section 7 and Appendix D.

Survey forms were distributed in April 2014. To reduce reporting burden, respondents were given flexibility in selecting the twelve-month period for which they reported costs. Financial institutions typically provided figures for the year to September 2013, while merchants provided somewhat more recent figures. The data were subjected to a number of validation checks following submission, including internal consistency checks, querying responses with participants, benchmarking against responses from other participants and comparison with other

12

sources such as the Reserve Bank’s Retail Payments Statistics and responses to the 2007 study.

A number of the caveats are worth bearing in mind. Like all such studies, the focus is on measuring resource costs (and some financial flows between parties). The study does not measure the benefits associated with different payment instruments nor whether the structure of the market promotes innovation. Both these factors need to be considered when drawing policy implications from these numbers; increased use of the lowest-cost payment system or less use of the higher-cost systems does not necessarily imply better outcomes.

A second issue is that while a comprehensive approach to identifying costs and the number of transactions has been used, coverage is not exhaustive. The study does not attempt to measure, for example, any costs arising from: the ‘cash’, ‘informal’ or ‘black’ economies; arguments that costs that might arise if cash is less hygienic

13

6 For example, there is no reliable way to estimate the number of informal or black economy transactions that would cease if different payment options, particularly cash, were not available. In addition, the tax revenue forgone because of undocumented income and sales, which Chakravorti and Mazzotta (2013) identify as a large cost associated with the informal economy, could generally be considered – in the framework of cost studies as – an (illegal) transfer.

A third issue is the ability of financial institutions and merchants to separately identify costs and transactions across payment instruments. A number of financial institutions reported difficulty identifying subcomponents of costs, particularly overhead costs and the amortised share of previous investment expenditure. Further, some financial institutions reported that it was difficult to allocate costs across the different types of card payments. Similarly, some merchants were unable to distinguish between the costs of different types of card payments in the information provided to them by their acquirers. However, in most cases respondents were able to provide an estimate of total costs which were then apportioned across different cost items or instruments using the Reserve Bank’s understanding of the Australian payments industry and the costs incurred by similar institutions in the sample.

The final issue involves the choice of costs that are included when measuring the resource costs used in making credit card payments. Credit cards generally have three aspects – a mechanism by which consumers can pay a merchant, an interest-free loan for up to 60 days, and reward points tied to the value of the purchase. While each of these aspects is measured, the main results of this study focus on the payment aspects, and do not include the costs associated with the interest-free period or rewards. Including the costs of providing the rewards or the interest-free period would increase the cost of credit cards, particularly to financial institutions, although they remain less expensive than cheques.

14

It is worth noting that these estimates are likely to be somewhat sensitive to sample coverage. While, as noted above, the coverage of financial institutions is very good, the coverage of merchants is narrower, with detailed information focused on larger merchants that are likely to experience lower costs per transaction for payment services. Augmenting these results with indicative cost numbers from the small business survey (see Section 7) would increase the total costs to the economy to around $10 billion or 0.64 per cent of GDP.9 Such an augmentation is, however, imprecise.10 Regardless of whether the base or augmented measure of aggregate cost is considered, these aggregate estimates suggest that the cost of ‘producing’ a

Our results suggest that the costs involved in providing payment services to households have fallen from 0.80 per cent of GDP in 2006 to 0.54 per cent of GDP in 2013. Based on the most recent estimate, it appears that Australia now has a relatively low cost payments system by international standards. With greater choice in payment methods, including wider access to, and acceptance of, electronic payment methods, today’s payments system also serves its users better than it did at the time of the Wallis Inquiry.

The decline in costs between 2006 and 2013 has been due to cost savings across most payment instruments. In particular, overhead cost savings have been realised through greater use of new technology, and tender times have been reduced through the adoption of PIN and contactless technology. Economies of scale are also likely to have reduced the per transaction cost of electronic payment methods that rely on networks with large fixed costs.

Measured on a per transaction basis and at the average transaction size for each payment instrument, the ranking across instruments by resource costs is mostly unchanged from 2006. BPAY and direct debit, which are electronic payment methods typically used for remote bill payments, remain the least resource intensive. Debit cards remain cheaper than credit cards, and cheques remain more expensive to society than any other payment method. However, cash has become slightly more expensive as its use has declined, while eftpos has become less expensive; these two methods are estimated to be the least costly of the methods available at the point of sale, with very similar per transaction costs. While cash has traditionally been the least costly payment method available for small payment values, developments since the previous study indicate that electronic payments are increasingly able to offer a low-cost alternative to cash. The point at which cash is no longer lower cost than eftpos has fallen from about $60 in 2006 to about $20 in 2013. Contactless debit transactions are also relatively low cost at very low payment values.

Decision-making by consumers and merchants, however, is not driven by the costs to society, but rather the private costs and benefits that these groups face.

Regulation recommended in the Wallis Report and put in place since 2001 by the Reserve Bank has worked to reduce the incentives for consumers to use high-cost options such as credit cards (RBA 2008). However, consumers continue to receive incentives in the form of rewards and interest-free periods to use credit cards, so that the private cost to consumers of credit cards is no different from debit card payments despite their higher social cost. Merchants continue to bear higher costs for credit card payments than debit card payments, and small merchants face significantly higher costs than large merchants.

The coming years will see further innovation – boosted by the current initiative for real-time payments – and ongoing change in consumer preferences. In such a climate, it may be worth considering whether relatively frequent updates of payment cost estimates might be useful to policymakers and the payments industry. Future studies will also need to consider carefully the scope of their investigation; cost estimates may be required for an even broader set of payment instruments or for other segments of the payments system, such as person-to-person payments.