When it comes to paying for goods and services, few of us think of the cost involved in running the systems that allow us to meet our obligation towards the vendor by handing over some bank notes and coins, by presenting a plastic card, or – increasingly rarely – by signing a cheque. Yet, these costs are far from negligible, and vary systematically across different ways of paying. Some studies suggest that the cost of making payments may account for more than two percent of the value of the purchases. Other studies indicate that we are not necessarily paying with the cheapest payment instrument, and that in particular cash is over-used. In a recent report, we have tried to bring together research into the cost of paying that has been undertaken by various central banks, academics and other stakeholders over the last decade, and to identify the common themes from this research.
Big, but how big?
First of all it is surprisingly difficult to measure the cost of making payments, and even more difficult to do so in a way that would allow one to compare the cost of paying with different payment instruments (cash, card, cheques, direct debit etc.). Some of the costs cannot be easily quantified. For example, the cost of the time spent making and accepting a payment depends on the appropriate measure of the opportunity cost of time. The cost of cashing up, depositing money with the bank, or the shoe leather cost of making withdrawals are equally difficult to quantify. The cost to consumers of having to bear the risk of loss or theft in relation to cash are potentially substantial, but measuring them is far from straightforward.
What price do consumers put on the risk of holding cash?
The costs associated with the risk of holding cash may be difficult to measure – but they are substantial. One study shows that, owing to the perceived risk associated with holding cash balances, the true opportunity cost of holding cash can be more than four times the interest lost on the consumers’ cash holdings.
The starting point for the estimation of these costs provided by Bergman et al. is the fact that Swedish customers generally do not pay for the withdrawal of cash (and only bear the fixed fee they pay for the card they use for making ATM withdrawals), but incur implicit costs in the form of shoe-leather costs incurred in visiting ATMs and the time spent withdrawing cash, and the lost interest on cash holdings. These two cost components are related: making smaller, more frequent withdrawals reduces the opportunity cost of holding cash, but increases shoe-leather costs, and vice versa. Assuming that customers choose the number of withdrawals and the amounts withdrawn so as to minimise the sum of lost interest and shoe-leather costs, information on the average amount withdrawn, the number of withdrawals made, and the interest rate that measures the direct opportunity cost of holding cash, can be used in a so-called inventory-theoretic approach to calculate the implied shoe-leather costs of making withdrawals.
However, this approach does not take into account that the full opportunity cost of holding cash may go beyond the lost interest and may include the risk of loss or theft associated with holding cash balances. Using the actual number of withdrawals and withdrawn amounts, in combination with assumptions about the average time taken to make a withdrawal and a measure of the opportunity cost of time (an appropriate after-tax wage rate), Bergman et al. (2007) explicitly calculate the shoe-leather costs incurred in making withdrawals.
The explicitly calculated figure for shoe-leather costs can then be plugged back into in the inventory-theoretic framework in order to establish at what interest rate the observed pattern of withdrawals would be optimal (i.e. minimise the overall cost for consumers). This implied interest rate reflects the total opportunity cost of holding cash – that is the sum of lost interest and the perceived risk of holding cash balances. Using the highest estimate for the explicitly calculated shoe-leather costs, Bergman et al. obtain a perceived risk premium of 17.5 per cent on top of the interest rate of 5.15 per cent that measures lost interest on cash holdings. This means the opportunity cost associated with the risk of loss or theft is more than three times the interest forgone on cash holdings.
Likewise many costs are common across different payment instruments (and potentially other banking services), making it very difficult to allocate them to one or the other. The issuance of debit cards, for example, supports both direct debit card payment and ATM withdrawals of money for cash payments, and cannot be attributed exclusively to card or cash payments. For example bank branches do not only deal with deposits and withdrawals, but also provide a wide range of services, and therefore many of the costs associated with a branch network are common across cash payments and other banking activities.
Last but not least, much of the information that would be required in order to make meaningful comparisons between the cost of different payment instruments is not generally available. There is a remarkable degree of uncertainty over the number and value of cash transactions for example, which means that unit cost estimates are potentially subject to a large margin of error.
The elusive value of cash payments
Whilst there is generally very good and reliable information about the value and number of non-cash payments, the anonymity of cash payments means that the number and value of cash payments is normally not known, and needs to be estimated on the basis of other data. Sometimes, there are large discrepancies between official statistics on cash usage even in what are considered to be highly transparent economies such as Norway, which might indicate not just methodological problems with statistics on cash usage, but the presence of unreported or under-reported transactions – i.e. shadow economy activity.
Minsch et al. (2007, p 20ff.), for example, estimate the number of cash transactions based on statistical information about value added tax receipts, using assumptions about sector-specific average transaction values and data on card usage. Gresvik and Haare (2009a, p 17ff.) discuss alternative options for obtaining estimates of the number and value of cash transactions which can lead to substantially different results. Using information from a survey of households, they arrive at an estimate of 285 million cash transactions with a value of NOK 62.1 billion in 2007. This contrasts sharply with a total value of cash withdrawals in Norway of NOK 141.5 billion for the same year. Based on official statistics for household consumption, the value of cash payments at the point of sale would be NOK 227.7 billion, which at the same average cash transaction value would correspond to more than 1 billion cash transactions. As Gresvik and Hare (2009b, p 24) note, ‘The anomalies and contradictory data to our estimate of cash payments at point of sale raise the question whether the calculation based on the household survey is correct. In our opinion, the use of cash payments and the use of cash for hoarding and illegal activities should be investigated further.’
Different researchers have addressed these difficulties in different ways – some have simply ignored some of the costs that are more difficult to measure (especially the largely non-monetary cost to customers), whilst others have tried to include estimates. Some studies look only at average cost figures, whilst others attempt to provide a more detailed analysis of the underlying structure of costs, and in particular the extent to which costs vary with the number or the value of transactions. This poses a challenge for anyone wishing to consolidate the various pieces of research, as comparing and contrasting is far from straightforward.
Despite the substantial differences in approach, and despite the differences across countries, there is some robust consensus emerging that cash is not, as many may believe, the cheapest payment method.
Measured as proportion of the value of purchases made – cash is more expensive than cards (in particular debit cards) for all stakeholders. For example, the cost for society as a whole of cash payments as a proportion of purchase value is almost three times the cost of card payments. This is largely because the costs of paying by cash vary to a large extent with the value of transactions, whereas the costs of card payments are largely fixed with regard to transaction value (though they vary with the number of payments). This means that cash may be the best payment method for low-value transactions, but that substantial cost savings could be achieved if cards were used instead for larger value payments. For a country such as Hungary, where cash is used for almost all payments and cards are the exception, these cost savings could amount to up to 0.4% of GDP, according to estimates by the Hungarian National Bank.1 It should therefore not be surprising that the cost of running the payment system, expressed as a proportion of GDP, is positively linked to the share of payments that are made in cash, as the diagram above shows.
The purchase value at which cash becomes more expensive than cards is surprisingly low: debit card payments are cheaper than cash for purchases above an average of approximately €10, and, for values above €70 using a credit card (which provides not just a simple payment functionality but added benefits to the customer) is cheaper than cash. Yet, average cash transaction values are often above these values, suggesting that cash is being used too often for large purchases even though cards would provide the cheaper way of paying.
Overall, this suggests that we could save money by changing the way in which we pay.2 Replacing costly cash payments with lower cost non-cash payments should generate appreciable cost savings, not least because economies of scale are most likely much stronger in relation to non-cash payment systems.3 Any shift in payment usage would result in the cost of cash payments falling by more than the cost of non-cash payments would increase.
This of course raises the question why we are not always using the cheapest way of paying for our purchases. Following on from that, what might be done in order to improve payment method choices?
Making better payment choices
Although it is ultimately the customer opening his wallet who decides whether to take out cash or a card, merchants and banks also play an important role in determining how we pay. Banks can promote or discourage the use of particular payment methods through their pricing, and merchants directly determine their customers’ choice through their decision whether to accept particular payment methods (and whether to steer their customers’ choice through measures such as surcharges or discounts, minimum purchase values etc). The reasons why the cheapest payment method is not always selected are that first, what is cheapest for society as a whole is not necessarily cheapest for individual stakeholders, whose decisions ultimately affect how purchases are paid for, and second, some of the costs are hidden and therefore easily ignored.
Particularly in the case of card payments a larger share of costs is re-distributed through fees from banks to merchants. By contrast, most of the costs of cash payments appear not to be recovered from merchants and consumers but are borne by banks and recovered across a range of banking activities. This means that from the perspective of merchants and customers, cash may look deceptively cheap relative to card payments. Taken together with the fact that many of the costs associated with cash payments (risk of loss and theft, time spent cashing up, risk of counterfeit notes etc.) are not as transparent and obvious as many of the fees associated with card payments, this often creates the erroneous belief that cash is essentially ‘free’.
Given that the current structure of charges does not necessarily provide the right price signals, the question is how one might improve the decisions about how to pay.
An obvious answer is that private costs and incentives need to be aligned with social costs. This could involve changes to the pricing regime such that banks reduce the extent to which they recover the cost of distributing and collecting cash from other activities. But other measures have been suggested in order to reduce the reliance of cash, in particular for high value payments, such as the withdrawing high value notes from circulation, limiting the amount of cash that can be withdrawn in a single transaction, or increasing the density of POS terminals.4 The most appropriate mix of measures will of course depend on specific market conditions and established payment habits – but being aware of the fact that making payments costs money, and that cash is far from free is certainly a good start towards making better informed payment choices.
- See: Magyar Nemzeti Bank (2011) ‘Cost of retail payments and payment habits of the public sector’, Presentation by Dr. Aniko Turjan, 1 April 2011. [↩]
- This ignores the wider benefits that might be obtained from reducing the size of the shadow economy by shifting away from anonymous cash payments towards card payments. The relationship between cash usage and the shadow economy is complex, and a detailed discussion is beyond the scope of this article. Interested readers might want to look at AT Kearney and F Schneider (2010) ‘The Shadow Economy in Europe 2010 – Using payment systems to combat the shadow economy’. [↩]
- Humphrey et al. (2003) argue that the fall in the ratio of operating cost to assets in the banking sector across 12 European countries between 1987 and 1999 can be explained by the replacement of check and paper-based giro transactions with electronic giro transactions and card payments. [↩]
- L. Van Hove (2009) ‘Could ‘nudges’ steer us towards a ‘less-cash society’?’, mimeo, Free University of Brussels, V2.0. [↩]