Abstract
The two-sided market theory holds that consumer adoption and merchant acceptance of payment cards are interdependent. However, empirical evidence on such network externalities is scarce, especially for the merchant side. This paper addresses this issue by examining merchant card acceptance in France. We exploit shopping diary data to construct a novel and fine-grained measure of French consumers’ payment preferences and match these with data from a nation-wide merchant survey. Controlling for (among other factors) cost, degree of competition, and customer characteristics, we find that the higher the probability that the average basket of a merchant is paid for by card in shops in the same sector and region, the higher the probability that the merchant will accept cards. In other words, we find that consumer preferences drive merchant card acceptance, which underpins the existence of network externalities on the merchant side of the payment card market.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
This number excludes hypermarkets; see Sect. 4.1 for details. In Germany, card acceptance would seem to be even lower: According to payment diary data from 2011 collected by the Bundesbank, in 40 % of all payment situations consumers could only pay cash (Eschelbach and Schmidt 2015, p. 2). This is not to say that the situation is all-pervasive: Recent figures for the Netherlands put acceptance of debit cards at 94 % of merchants (DNB 2015).
Direct (or intra-group) network externalities exist when the value of a good for a user depends directly on the number of other users, as in communications networks.
Using descriptive statistics from a Canadian merchant survey, Arango and Taylor (2008) provide only indirect evidence that merchants accommodate consumer preferences. Carbó-Valverde et al. (2012), for their part, use quarterly bank-level information—i.e., aggregated data—to examine the feedback loop effects between cardholders and merchants. The Carbó-Valverde et al. paper is clearly differently inspired, as compared to ours.
Source: Cartes Bancaires. The corresponding figures for 2008 (the year of our merchant survey) are: 1.2 million merchants, 58 million cardholders, and 6.2 billion payments (retrieved from www.cartes-bancaires.com; last visit: 12/06/2015).
Merchant service charges can differ from one acquirer to the next—see below—but any given acquirer will charge identical fees for all types of CB cards. Also, the deferred payment is an option that is proposed to cardholders by issuers; it does not affect the timing with which merchants’ bank accounts are credited.
Namely: equipment and personal services, transport, restaurants/bars/hotels, food and beverages, leisure and culture, home furniture, legal affairs, newspapers/tobacco/gambling, health, and other.
We could have had a ‘truncated sample’ problem here, in that merchants that did not answer might be those that received a favorable deal. However, upon inspection, characteristics such as sales revenue and number of transactions do not differ significantly between merchants who answered and those who did not. Calculations can be provided upon request.
This includes online, telephone order, and mail order transactions. However, online transactions accounted for only 0.5 % of the total in 2005 and for 2 % in 2011. We therefore do not treat them separately. Telephone and mail order transactions are even more marginal at, respectively, 0.45 and 1.2 % in 2011.
Excluding the overseas territories, France is divided into 96 “départements”. Departments are administrative divisions that correspond to the NUTS3 level. One level higher, on the NUTS2 level, the departments form 22 regions. NUTS1, for its part, divides France into only 9 regions.
Card ownership overstates the externality that is generated by cardholders if there are cards in circulation that are inactive or seldom used; see Van Hove (2000).
232 merchants refused to give this information, so we lose observations here.
Also, as documented by Bouhdaoui and Bounie (2012), in France changes in payment behaviour over 2005–2011 were gradual, not abrupt. (As a sensitivity analysis, we have repeated our estimations with the 2005 data only.) However, a strict implementation of the proposed procedure would have resulted in a substantial loss of observations, in particular for the more disaggregated spatial dimensions. The reason is that not every combination of basket size + sector + region/department that appears in the merchant survey is covered in both consumer surveys. With NUTS3 departments, for example, we would have lost no less than 2380 of the 4601 observations (or 52 %). In order to preserve as many observations as possible, we have therefore also used observations for which we found a match in only either the 2005 or the 2011 survey. As long as we work with NUTS1 regions, this affects a mere 118 observations of the 4369 (or 3 %). With NUTS2 and NUTS3 regions, the impact is progressively bigger: the observations for which we use a single-year value rather than an average increase to, respectively, 520 out of 4299 (12 %) and 1454 out of 3443 (42 %). In a robustness check, we introduced dummy variables that flagged this single-year usage; but this did not affect the results.
In the diaries, respondents were given the possibility to check a box that said (translated) “forced choice of payment instrument”. One of the examples of such a situation that was given in the user guide for the 2011 survey was “you have paid by cheque even though you wanted to pay using your bank card but the shop refused the bank card”. Another illustration was: “you wanted to pay by bank card but the amount is below the threshold at which the shop accepts card payments”. So, provided that the respondents filled out the diary correctly, if we eliminate all transactions for which they indicated that the choice of payment instrument was forced (and, reassuringly, the patterns in the discarded transactions make sense), we in effect are left with transactions where consumers could use the payment instrument they wanted.
Let us also stress that we do not have on one side of our regressions the share of card payments (in a certain region and sector, and for a certain transaction range) and on the other side the share of card-accepting merchants. Rather, our card preferences variable is based on the experience of multiple consumers at multiple merchants, whereas the dependent variable is acceptance (y/n) of an individual merchant. This difference in level is another reason why endogeneity would not seem to be an issue. Nevertheless, in a robustness check, we also ran a regression with consumer card preferences for the year 2005 alone (rather than an average of 2005 and 2011), so that there is also a difference in timing with the dependent variable (which relates to 2008). The results proved stable.
We eliminated all transactions that are also excluded in the construction of the CCP variable.
In France most payment terminals nowadays rely on the Internet to connect to payment scheme servers, and merchants typically pay a subscription for the service. Telecommunications costs can thus be considered fixed.
The ranges of the tertiles are as follows: ‘less than 0.747 eurocents’ for ‘low’ (36.5 % of the merchants); ‘between 0.747 and 0.75 eurocents’ for ‘intermediate’ (32.8 %); and ‘more than 0.75 eurocents’ for ‘high’ (30.7 %). The minimum and maximum values are 0.00857 and 0.96 eurocents. (The tertiles do not account for exactly 33.3 % each because the HMDFs are discrete and sometimes a great many merchants share the same fee.)
Since we needed to make quite a few assumptions in order to be able to compute the HMDF and since the survey responses that we use in our calculations may well be imprecise, we thought it unwise to try to predict, in step 3, exact fees for merchants for whom we do not have the information. By predicting only a rough level, we reduce our dependence on both our assumptions and the quality of the answers.
As explained in Sect. 3, in France pricing strategies differ widely between banks.
We opted for the NUTS1 level rather than NUTS3 (departments) so as to be sure to have a sufficiently high number of merchants per territorial division. As explained in Sect. 4.1, the geographical stratum of the merchant survey was on the NUTS1 level. For certain departments we have less than 10 merchants.
We also used a standard probit model to estimate the fee category and found very similar results.
For a discussion of the pros and cons of this type of measure, see Jonker (2011, p. 14).
We have also tried the ratio, per department, between the number of CB-affiliated merchants and the total number. However, this variable proved insignificant. Compared to the first, the alternative variable is at the same time less fine-grained (it is not on the level of individual merchants) and narrower (competition between merchants involves more than just card acceptance). This probably goes same way to explain its insignificance.
Because the answer categories were non-exclusive, a merchant can, for example, receive a 1 for both ‘less than 30 years’ and ‘between 30 and 60 years’, and a 0 for ‘more than 60 years’.
We also control for the sector of the merchant (10 categories), the length of the bank relationship (5 categories), and the identity of the bank (14 categories). The idea underlies the inclusion of the length of the bank relationship is to capture the phenomenon that in recent years (compared to 2008, when we collected the data) French banks started promoting packages of services that included card acquiring services, which was not the case in earlier years. Hence, the shorter the banking relationship, the higher the probability of card acceptance. The variable might also to some extent capture the age of the merchant (and thus technology aversion) or even overall ‘conservativeness’.
With a cut-off value of 0.5. Also, the standardised residuals appear to be normally distributed and the exclusion of the 18 observations with a Pregibon influence statistic higher than 0.2 did not alter the outcomes.
Due to the large number of variables, we do not report the correlation matrix. It is available upon request.
We also tried the volume (rather than the value) of card payments in the department, and this for both the year 2008 and the period 2000–2008. Neither variable turned out to affect card acceptance.
In a robustness check, we added the average transaction size of the merchant (categorical variable) to the regression. This did not affect our results.
We discuss the effects of binary variables in terms of odds ratios.
Note that an interaction term between the degree of competition and consumer preferences proved insignificant.
We also tried a variant with the five original income categories (rather than the three that we created by grouping the first and final two together). This did not change the results.
One could argue that the taxation regime is also related to the size of the merchant. However, our correlation matrix did not reveal a strong correlation with either sales revenue or the daily number of transactions.
References
Arango, C., Huynh, K. P., Fung, B., & Stuber, G. (2012). The changing landscape for retail payments in Canada and the implications for the demand for cash. Bank of Canada Review, Autumn, 31–40.
Arango, C., & Taylor, V. (2008). Merchant acceptance, costs, and perceptions of retail payments: A Canadian survey. Bank of Canada Discussion Paper 2008-12.
Bagnall, J., Bounie, D., Huynh, K. P., Kosse, A., Schmidt, T., Schuh, S., & Stix, H. (2016). Consumer cash usage: A cross-country comparison with diary survey data. International Journal of Central Banking (forthcoming).
Bagnall, J., & Flood, D. (2011). Cash use in Australia: New survey evidence (pp. 55–62). Bulletin, September: Reserve Bank of Australia.
Bouhdaoui, Y., & Bounie, D. (2012). Modeling the share of cash payments in the economy: An application to France. International Journal of Central Banking, 8(4), 175–195.
Cameron, A. C., & Trivedi, P. K. (2005). Microeconometrics: Methods and applications. Cambridge: Cambridge University Press.
Carbó-Valverde, S., Liñares-Zegarra, J. M., & Rodríguez-Fernández, F. (2012). Feedback loop effects in payment card markets: Empirical evidence. Review of Network Economics, 11(2), 1–24.
Chakravorti, S., & To, T. (2007). A theory of credit cards. International Journal of Industrial Organization, 25, 583–595.
De Nederlandsche Bank (DNB). (2015). Cash–retailers’ behaviour and perception. http://www.dnb.nl/en/binaries/712585_factsheet_retailersonderzoek_EN_tcm47-321626.pdf.
Eschelbach, M., & Schmidt, T. (2015). Precautionary motives in short-term cash management—evidence from German POS transactions. Paper presented at the Joint European Central Bank/Suomen Pankki conference on ‘Getting the balance right: innovation, trust and regulation in retail payments’, June 4–5, Helsinki, Finland.
Fédération Bancaire Française. (2008). Banque de détail. Des progrès pour un marché européen.
Hayashi, F., & Klee, E. (2003). Technology adoption and consumer payments: Evidence from survey data. Review of Network Economics, 2(2), 175–190.
Jonker, N. (2011). Card acceptance and surcharging: The role of costs and competition. Review of Network Economics, 10(2), 1–35.
Jonker, N., Kosse, A., Hernandez, L. (2012). Cash usage in the Netherlands: How much, where, when, who and whenever one wants? DNB Occasional Studies, 10(2). https://www.dnb.nl/binaries/DNB_OS_1002_tcm46-271572.pdf
Koulayev, S., Rysman, M., Schuh, S., & Stavins, J. (2016). Explaining adoption and use of payment instruments by U.S. consumers. RAND Journal of Economics, 47(2), 293–325.
Loke, Y. J. (2007). Determinants of merchant participation in credit card payment schemes. Review of Network Economics, 6(4), 1–21.
McAndrews, J., & Wang, Z. (2012). The economics of two-sided payment card markets: Pricing, adoption and usage. Working Paper Series 12-06, Federal Reserve Bank of Richmond.
Rochet, J.-C., & Tirole, J. (2002). Cooperation among competitors: Some economics of credit card associations. RAND Journal of Economics, 33, 549–570.
Rochet, J.-C., & Tirole, J. (2011). Must take cards: Merchant discounts and avoided costs. Journal of the European Economic Association, 9, 462–490.
Rysman, M. (2007). An empirical analysis of payment card usage. Journal of Industrial Economics, 55, 1–36.
Rysman, M. (2009). The economics of two-sided markets. Journal of Economic Perspectives, 23(3), 125–143.
Van Hove, L. (2000). The New York City smart card trial in perspective: A research note. International Journal of Electronic Commerce, 5(2), 119–131.
Whitesell, W. C. (1989). The demand for currency versus debitable accounts: A note. Journal of Money, Credit and Banking, 21(2), 246–257.
Wright, J. (2011). Why do merchants accept payment cards? Review of Network Economics, 9(3), 1–8.
Acknowledgments
We thank the Editor Lawrence J. White, two anonymous referees, Nicole Jonker, Sibel Aydogan, Yassine Bouhdaoui, Cédric Sarasin, and Ludovic Francesconi as well as participants at the 2015 Bank of Canada Annual Conference for helpful comments on earlier versions of this paper.
Author information
Authors and Affiliations
Corresponding author
Additional information
An earlier version of this paper circulated under the title “Merchant acceptance of payment cards in France: le client est roi!”.
Rights and permissions
About this article
Cite this article
Bounie, D., François, A. & Van Hove, L. Consumer Payment Preferences, Network Externalities, and Merchant Card Acceptance: An Empirical Investigation. Rev Ind Organ 51, 257–290 (2017). https://doi.org/10.1007/s11151-016-9543-y
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11151-016-9543-y