RDP 2015-03: The Value of Payment Instruments: Estimating Willingness to Pay and Consumer Surplus 3. Data

We use data from the Reserve Bank of Australia's 2013 Survey of Consumers' Use of Payment Methods (the Survey). The dataset consists of 1,167 individuals representative of the Australian adult population and includes information on respondents' demographic characteristics, their ownership of debit and credit cards, their use of various payments methods as logged by a week-long diary, and information about their preferences as to payment attributes. The Survey was conducted by Colmar Brunton during November 2013. Full details of the survey methodology and results of the study are available in Ossolinski, Lam and Emery (2014).

In addition, we use data on the features of the credit card held by the respondents. As part of the Survey, each respondent identified their most commonly used credit card and debit card. The detailed features of the credit cards were obtained from information on card issuers' websites.

In the Australian market, institutions issuing credit cards typically offer a number of card product categories that involve different mixes of card features.[3] The main credit card features that we are interested in are the rewards program and the interest rate. We quantify the generosity of the rewards program on cards by the rewards rebate – which we calculate to be $100 divided by the spending required to obtain a $100 major store gift card (i.e. the effective rebate on a $1 of spending).[4] Around 40 per cent of respondents held cards without rewards programs (Table 1). Around 5 per cent of respondents belong to premium rewards programs that provide a rebate of over 100 basis points.

Table 1: Characteristics of Credit Cards Held by Respondents
Number of respondents Per cent of credit card holders
Debit card ownership 938  
Credit card ownership 605 100
Credit card features    
Rewards rebate (bps)(a)    
No rewards rebate 234 39
20–39 57 9
40–59 151 25
60–79 128 21
80–99 5 1
≥ 100 30 5
Interest rate (per cent)    
No interest (charge cards) 37 6
10–13.99 95 16
14–17.99 58 10
18–21.99 408 67
≥ 22 7 1
Premium services 218 36

Note: (a) The rewards rebate is calculated as $100 divided by the spending required to obtain a $100 major store gift card

Sources: Authors' calculations based on survey data; RBA

The interest rate charged on credit card purchases after the expiry of the interest-free period (typically 30–55 days) can vary from card to card. Posted interest rates on card products vary from 10 per cent per annum to over 22 per cent, although the majority of respondents held cards that have an interest rate of 18–22 per cent.[5] Charge cards require their users to pay the balance in full each month and thus do not charge interest on their balances. Credit card holders can be divided into those that pay their balance in full prior to the interest-free period expiring (transactors) and those who revolve their credit card balances (revolvers).

3.1 The Modified Discrete Choice Experiment

The modified DCE itself was contained in a questionnaire answered by respondents after they had completed the week-long payment diary. Responses were collected online from 1,069 participants and via a paper questionnaire from another 98 participants who did not have access to the internet. The modified DCE was shortened for the paper participants as there were concerns that participants would not follow the full sequence of questions in a paper format. This paper draws only on the responses of the online participants who faced the full sequence of modified DCE questions and, for credit card holders, provided the necessary details to identify their primary credit card. The sample of 938 participants is representative of the Australian population aged over 18 years (see Table A1).

The DCE asked respondents to consider a hypothetical situation in which they are making a $50 transaction at a store. Respondents were told that they had their typical amount of cash on them and their most commonly used debit card. They were required to choose whether to pay using a debit card and pay a 1 per cent surcharge, or to make the purchase using cash with no surcharge. Contingent on whether they chose to proceed with their debit card or proceed with cash, two follow-up questions were posed where the level of the card surcharge was increased or decreased to refine the respondent's range of willingness to pay (as per the logic described in Table 2).[6] From the series of questions, the willingness to pay of each respondent can be identified as falling within one of eight willingness-to-pay ranges listed in Table 2.

Table 2: DCE Questions and Resulting Respondent Willingness-to-pay Range
Decision to proceed with card payment; per cent of card surcharge on $50 transaction
1st question 2nd question 3rd question Resulting willingness-to-pay range
1% if Yes 3% if Yes 4% if Yes ≥ 4%
if No 3–4%
if No 2% if Yes 2–3%
if No 1–2%
if No 0.5% if Yes 0.75% if Yes 0.75–1%
if No 0.5–0.75%
if No 0.1% if Yes 0.1–0.5%
if No < 0.1%

Source: RBA

The sequence of three questions was repeated for credit card holders with the variation that the respondent had to choose between using their most commonly used credit card and paying the surcharge, or using cash (which did not attract a surcharge) to make the purchase. (All respondents held debit cards and thus answered the debit card questions.) The results of the two sequences of questions is that we observe each respondent's willingness-to-pay range for the use of credits cards (provided they own a credit card) and separately, their willingness-to-pay range for the use of debit cards, both relative to the cash alternative. The questions were designed to allow comparison, assuming transitivity, of the willingness to pay for the use of debit cards and credit cards against a common baseline (the use of cash).

The DCE was designed to limit certain biases that can be introduced when asking for stated preference data.[7] The scenario was designed to be realistic so that respondents could provide an answer that is representative of their actions in the real world. First, surcharging is a known practice in Australia and the range of surcharge values included in the hypothetical scenario was also realistic, which should have helped to reduce hypothetical bias. Further, the $50 price point was specifically chosen to increase the likelihood of measuring the value of the payment function of the card. Debit cards and credit cards are both commonly used for $50 transactions (Ossolinski et al 2014). At values of $100 or higher, credit cards are more frequently used than debit cards, presumably because liquidity constraints factor increasingly in the consumer's decision of which payment method to use as the value of the payment increases. Finally, data from the Survey indicated that 55 per cent of respondents at the beginning of the Survey carried enough cash to make this payment without needing to go to an ATM.[8]

The realism and simplicity of the scenario may, however, have unintended effects. First, it is possible that some consumers may refuse – both in reality and in our scenario – to pay a surcharge due to strong negative feelings to additional charges applied at the point of sale rather than due to an objective evaluation of the benefits of different instruments; this behaviour is necessarily outside our model. Second, some consumer advocacy groups have advanced 1 per cent as a reasonable surcharge on credit cards, which may have influenced respondents' own valuation of the benefit of cards. Third, all respondents faced the same surcharge level of 1 per cent in the first question of the DCE (randomisation was not possible), which may have anchored their responses to subsequent questions.

A final issue is that $50 is a whole number also equal to the value of one of the Australian banknotes; this is likely to lower the perceived cost of using cash and increases the probability of using cash relative to payments in the nearby price range.[9] Our results, therefore, are going to be most applicable to point-of-sale payments of around $50. Payments data from the same sample of consumers suggest that payments of $50 account for around 2 to 3 per cent of the number and value of all point-of-sale payments, whilst payments of between $40 and $60 make up around 13 per cent of the number and value of all point-of-sale payments. Despite the small shares, these are non-trivial amounts considering that over $300 billion is spent at the point of sale each year.

Importantly, however, any biases and caveats are likely to affect the willingness to pay for the use of debit cards and credit cards equally. This provides us with more confidence in comparing the value of the payment function of debit cards relative to credit cards, a key measurement of interest for this paper.


Credit cards are often classified into standard, low-rate and premium cards, which all attract different annual fees. Standard cards provide a moderate level of rewards and attract a moderate annual fee. Low-rate cards specialise in offering comparatively low rates of interest, but may attract higher annual fees than standard cards, and typically offer no rewards. Lastly, a premium category of cards exist that have high rewards, provide extra services (e.g. travel insurance, price protection, or concierge services), but are offset by a high annual fee. [3]

Basing our measure on gift cards from major store chains provides us with a comparable measure of generosity across different rewards programs. There may nevertheless be some imprecision with our measure; for example, some rewards programs are designed around the redemption of air travel benefits, the value of which is difficult to measure given the differing and fluctuating prices of air fares. [4]

Most financial institutions offer zero interest balance transfers for individuals switching credit card providers. This may mean the posted interest rate is not reflective of the interest costs some credit card holders may actually face. [5]

In more-typical implementations of DCE, price levels faced by respondents are randomised to create additional variation and mitigate the effects of anchoring (where respondents' answers are influenced by a value shown earlier in the question or by the answer structure of the DCE). Randomisation was not possible in the Survey. [6]

For a critique of the use of willingness-to-pay estimation techniques see Kling et al (2012). [7]

The Survey data only provides information on the amount of cash respondents held at specific points in the Survey rather than their ‘typical’ amount. The DCE uses the ‘typical’ amount of cash as a way of framing the situation under realistic conditions, and the amount of cash is not intended to serve as a constraint for the respondents' answers. The DCE responses suggest this was not interpreted as a constraint, as respondents who held less than $50 at the start of the Survey were distributed relatively evenly across each willingness-to-pay range and were not over-represented in the group of respondents willing to pay the 4 per cent surcharge. [8]

Payments data from the same sample of consumers indicate that the share of payments made using cash rises from around 35 per cent for all point-of-sale payments between $40 and $60 to around 50 per cent for payments of exactly $50. This may reflect that the speed of the transaction is lower when change is involved, or that individuals prefer not to receive change. [9]