“WHAT DO PAYMENT CARDS ADD? DISENTANGLING CUSTOMERS’ VALUATION FROM MARKET POWER”

1. “WHAT DO PAYMENT CARDS ADD? DISENTANGLING CUSTOMERS’ VALUATION FROM MARKET POWER” Santiago Carbó-Valverde (UGR & Funcas) IDEI, Toluose – Panel 1 ‘Competition Policy in Two-Sided Markets’ 29 June 2006

2. 1. The value of cards in a two-sided card market • Payment cards have become one of the main examples of two-sided markets. • Unlike other platform industries, however, the analysis of cards have been frequently framed within a standard vertically organized concept of market, and antitrust agencies have undertaken resolutions based upon this perception. • An intense debate between the industry and antitrust agencies.

3. Card payment instruments are becoming increasingly used because of their lower cost, relative to paper-based payments [Humphrey and Berger (1990); Humphrey et al. (1997)] . • It has also been observed that credit cards help shifting illiquid customer consumption forward in time (Chakravorti and To, 2000; Brito and Hurtley, 1995). • Banks providing cards and ATMs may extend their deposit services outside their branch network enabling depositors to withdraw cash at more convenient times and places (McAndrews, 2003)

4. As a consequence of the large range of interactions among acquirers and issuers, banks tend to coordinate these activities building open payment networks. • Card markets also present adoption and usage externalities (Stango, 2002; Knittel and Stango, 2006). • Cards increase consumers’ willingness to pay and, as a result, merchants that accept cards increase their sales and earn more profit than merchants that do not accept them (Wright, 2005).

5. 2. Pricing schemes and antitrust policies in card markets • As a consequence of the complexity of this pricing scheme, there are different and, at least partially, contradictory findings regarding the link between network effects and pricing decisions. • McAndrews (1996) finds that both demand-side network effects and the economies of scale effect affect these pricing decisions. • Hannan et al. (2003) find that the probability of surcharging is an increasing function of size and of market share (of ATMs) and that depositor’s affiliation incentives imply higher surcharges.

6. Interchange fees: setting the fee at the network association level eliminates costs related to bargaining between individual card issuers and acquirers and uncertainty about the actual costs of a card transaction (Baxter, 1983; Small and Wright, 2001). • Other studies -such as Carlton and Frankel, (1995), Frankel (1998), Chang and Evans (2000) or Balto (2002)- identified potential anticompetitive effects of the collective determination of interchange fees. • Recent contributions have maintained that collective determination should not be banned. It is not obvious whether negotiations between issuers and acquirers would lead to lower or higher interchange fees [Gans and King (2003); Guerin-Calvert and Ordover (2005)].

7. Interchange fees are not set as an ordinary market price but as a balancing device for increasing the value of a payment system by shifting costs between issuers and acquirers and thus shifting charges between consumers and merchants. (Schmalensee, 2002; Rochet and Tirole (2002). • Wright (2004) shows that when merchants compete and consumers are fully informed as to whether merchants accept cards, the profit and welfare maximizing fee coincide for a non-trivial set of cases.

8. Many prices should be consider along with interchange fees (annual fees, merchant fees, …). • IMPORTANTLY: Banks typically bundle the charges for various products, offering their customers a choice of ‘packages’ (Stavins, 1999).

9. 3. An empirical test to disentangle customers’ valuation from market power Main aims: • (i) To quantify customers’ valuation of cards services provided jointly with other bank services as an output mix. Valuation is indirectly quantified by estimating revenue scope economies from a composite multiouput specification. • (ii) To disentangle the ‘valuation’ effect from any source of market power derived from the development of card services. This empirical goal involves the simultaneous estimation of a system of demand and supply relationships in order to undertake a ‘mark-up test’ that permits to quantify the percentage deviation of prices, quantities (and, therefore, revenues) from perfect competition levels.

10. (iii) In a third stage, we rely on a hedonic approach to check the robustness of our measures to various sources of convenience and indirect network effects. • We employ a unique database that contains quarterly bank-level information on card transactions and revenue sources at ATMs and EFTPOS terminals as well as other bank information. The sample consists of all savings banks operating in Spain from 1997:1 to 2003:3 summing up to 1.242 panel observations.

11. Spain is the second largest ATM and EFTPOS industry of the world (55,399 ATMs and 1,055,103 EFTPOS at the end of 2004). The number of cards has almost doubled from 1996 (33,189,000) to 2004 (63,027,000). • The number of transactions per EFTPOS terminal has increased from 511 in 1996 to 1,204 in 2004. The number of EFTPOS terminals has also augmented in that period from 575,325 to 1,055,103 (a net increase of 83.39%). • The ‘no-surcharge’ and the ‘honor-all-card’ rule apply in Spain and, as in many other countries there has been an intense debate about the fees charged on cardholders and merchants transactions .

12. Under certain conditions, revenue scope economies will illustrate synergies in the joint consumption of financial services: • Three-output vector: loans, deposits and other earning assets. • Five-output vector: loans, deposits, other earning assets, value card of transactions at ATM and value of card transactions at POS. • Differences between revenue scope economies from the five-output vector composite function and revenue scope economies from the three-output vector composite function will show the net contribution of cards to the joint valuation of bank services.

13. The basic structure of our revenue function model may introduce some bias in our estimations of customers’ valuation due to market power, we aim to disentangle both effects employing the so-called ‘mark-up’ model proposed by Bresnahan (1982) and Lau (1982): The ‘mark-up’ test estimates the extent to which the average firm's perceived marginal revenue deviates from the demand. • The model consists of the joint estimation of a demand and a supply relationship (computing marginal costs from the same composite function employed to estimate revenue complementarities).

14. With the “mark-up” model we estimate quantity and price percentage deviations from competitive equilibrium models. • Finally, a series of hedonic regressions of various output prices on a series of characteristics of the bank own and bank competitors’ services are also estimated to show the marginal effect of these characteristics on the willingness to pay of customers. They also act as a robustness check for the valuation outcomes from scope economies when market power influences are appropriately controlled.

15. 4. Main results • Main results of the valuation tests shown in Table 5 (as % of revenues): • Positive valuation of card services (2.63%). • Indirect network effects also positive: additional revenues complementarities considering transactions of non-depositors (0.35%). • Revenue deviation from competitive equilibrium (2.45%). • Net valuation (0.52%).

17. Hedonic regressions. Main results in Table 8: • Consumers appear to value positively and significantly the development of EFTPOS terminals, no matter if they are own bank or competitors’ terminals. Hence, no-surcharging rules imply a valuation of EFTPOS no matter which is the acquirer bank. • The hedonic regressions also reveal the existence of certain substitution relationships between ATMs and EFTPOS. This reflect a current concern of the industry in order to promote card payments at the POS.

19. 5. Some policy implications • The empirical results reveal that there is a net positive valuation of card services provided jointly with other bank products even when market power influences are controlled. • Furthermore, indirect network effects appear to be a significant component of this joint valuation. • The absence of perfect competition is overcome by these valuation effects. Intervention may artificially reduce prices although it may also reduce the incentives to invest in card payment services.