File No.: CT-2005-006 COMPETITION TRIBUNAL IN THE MATTER OF the Competition Act, R.S.C. 1985, c. C-34, as amended; IN THE MATTER OF an application by B-Filer Inc., B-Filer Inc. doing business as GPAY GuaranteedPayment and Npay Inc. for an order pursuant to section 103.1 granting leave to make . application under sections 75 and 77 of the Competition Act;
AND IN THE MATTER OF an application by B-Filer Inc., B-Filer Inc. doing business as GPAY GuaranteedPayment and Npay Inc. for an interim order pursuant to section 104 of the Competition Act.
BETWEEN: B-FILER INC., B-FILER INC. doing business as GPA Y GUARANTEEDPA YMENT and NP A Y INC.
-and-
THE BANK OF NOVA SCOTIA
Affidavit of Lawrence P. Schwartz I, Lawrence P. Schwartz, of the City of Toronto, in the Province of Ontario, AFFIRM: p R 0 o,, u: I T
Applicants
Respondent
2 1. I attach my report setting out my op1mon m response to the report of Professor Mathewson as exhibit "A" to this affidavit.
AFFIRMED BEFORE ME at the Cit 2006.
BETWEEN: B-FILER INC., B-FILER INC. doing business as GPAY GUARANTEEDPAYMENT and NPAY
THE BANK OF NOVA SCOTIA Respondent
LAWRENCEP. SCHWARTZ
(Affirmed August 14, 2006) AFFLECK GREENE ORR LLP Barristers & Solicitors One First Canadian Place Suite 840, P.O. Box 489 Toronto, Ontario M5x le5
Michael Osborne T 416-360-5919 F 416-360-5960
E mosborne@agolaw.com
Jennifer Cantwell T 416-360-F 416-360-5960
E jcantwell@agolaw.com
EDY, DALTON 800-1015 4 ST. S.W. CALGARY, AB, T2R IJ4
Sharon J. Dalton T 403-263-3200 EXT:105 F 403-263-3202
E jdalton@edydalton.com
Solicitors for the Applicants
Court File No. CT-2005-006 COMPETITION TRIBUNAL INC. Applicants -and-
AFFIDAVIT OF
CONFIDENTIAL LEVEL A Expert Rebuttal and Reply Affidavit of Lawrence P. Schwartz August 14, 2006
Page 1 of 25
CONFIDENTIAL LEVEL A [1] I have been asked by counsel to B-Filer Inc. to review and comment on the Report of Professor Frank Mathewson dated July 31, 2006 (the “Mathewson Report”). My response includes comments in rebuttal as well as in reply to his comments on my Affidavit dated July 10, 2006 (“Affidavit”). [2] Professor Mathewson opines that 1. Scotiabank biller services and EMT business deposit accounts are not properly defined product markets for the purposes of this case 2. Personal deposit accounts and perhaps other products are in the same product market with EMT business deposit accounts 3. Scotiabank biller services is not a product market; rather, the market that includes Biller Services also includes EMT deposit accounts (both personal and business) 4. Scotiabank’s termination of its relationship with GPAY will have a minimal effect on competition 5. Existing Scotiabank consumers who wish to use GPAY have alternatives to meet their needs, including credit cards 6. Scotiabank terminated its banking relationship with GPAY because, inter alia, GPAY’s manner of conducting business requires banking customers to disclose their online banking password and because Scotiabank wished to minimize the legal, reputational, and regulatory risks associated with doing business with GPAY 7. Scotiabank’s termination of GPAY was not driven by a strategic attempt to disadvantage a competitor or to dampen competition for current or future merchants or consumers. [3] In evaluating Professor Mathewson’s conclusions, I will address his approach to market definition and his assessment of the adverse effect on competition of Scotiabank’s termination. I will make only limited comments on his acceptance of Scotiabank’s business justification because this subject was not part of my mandate and is an area in which I have only limited expertise.
Page 2 of 25
CONFIDENTIAL LEVEL A [4] I will refrain from commenting on Professor Mathewson’s understand of the technical descriptions of the product and bank services found at ¶10-¶20 of the Mathewson Report as these are matters that will be established in evidence by the various fact witnesses. [5] For ease of reference, I will follow Professor Mathewson’s general reference to “GPAY”, although my Affidavit refers to “B-Filer”.
Product Market Definition Generally [6] The principal difference between Professor Mathewson and me lies in the approach to product market definition. In the analysis in my Affidavit, I relied on the hypothetical monopolist approach to market definition contained in the merger guidelines of the enforcement agencies in Canada and the United States and accepted increasingly, as I noted, in Europe. I discussed the issues involved in applying this approach to monopolization/abuse cases and refusal-to-deal cases and I concluded that the approach was appropriate for delineating the product market in the matter at hand. [7] I identified three relevant product markets: 1. Online debit payment service for Scotiabank depositors, which market includes the GPAY Service and Interac Online 2. The market for merchants, in which B-Filer competes with Interac Online transaction acquirers 3. Regarding the means of providing online debit payment to Scotiabank depositors, biller status at Scotiabank but excluding business accounts that accept payment by EMT. [8] The essence of the hypothetical monopolist approach is substitutability of other products for the product at hand when viewed from the prospective of the buyer. Thus, I evaluated whether Scotiabank depositors who use the GPAY Service would switch to alternatives to the GPAY Service in the event of a small but significant increase in the price (“SSNIP”, typically 5% maintained for at least one year) thereof. I also evaluated whether online merchants currently using the GPAY Service would switch in the event of a SSNIP imposed by B-Filer and whether B-Filer would switch to EMT’s in the event of a SSNIP imposed by Scotiabank on its biller status.
Page 3 of 25
CONFIDENTIAL LEVEL A [9] The hypothetical monopolist approach delineates the product market by asking what price increase a monopolist in the same situation would impose. If, for a given product, it would impose a SSNIP, then substitutability is sufficiently low that that product or set of products constitutes the relevant product market. If it would not impose a SSNIP, then substitutability is sufficiently high that the provisionally-adopted market must be expanded. At all times, the question is what buyers would, or would likely, do. In this approach, the question posed is not what buyers could do, or might do, or what alternatives may exist for them. [10] The guidelines of the Competition Bureau (“Enforcement Guidelines”) adopt this approach to market definition in examining conduct involving monopolization and abuse of dominant position. The analysis
“focusses on whether there are close substitutes for the product(s) in question, such that buyers would turn to these substitutes in the event that the product price was raised above competitive levels by a significant amount for a non-transitory period of time. In general, a 5 percent real price increase above competitive levels lasting one year is considered a significant and non-transitory amount.” 1 Thus, the approach in the guidelines is to examine, prospectively, what buyers would do in response to a significant price increase. Those guidelines indicate that price is not the only indicator for defining product markets and that other, qualitative factors will be examined including:
“The views, strategies, behaviour and identity of buyers. Whether buyers have substituted between products in the past and whether they plan on doing so in the future can indicate whether a price increase is sustainable.” 2 1 Competition Bureau. Enforcement Guidelines on the Abuse of Dominance Provisions, July 2001 at ¶3.2.1(a) (available at http://strategis.ic.gc.ca/pics/ct/aod.pdf). 2 Ibid. Page 4 of 25
CONFIDENTIAL LEVEL A The enforcement guidelines thus emphasize the prospective nature of the market definition inquiry. In this regard, evidence of substitution prior to the conduct under review and plans for the future can be helpful in delineating the product market. [11] The approach adopted by Professor Mathewson is very different. He states: “Accordingly, when assessing the substitutability of other products for ‘Scotiabank biller services’ to determine whether these other products belong in the same product market, the operative principle is that other products are substitutes if the purchaser’s business is not substantially affected by switching to these other services. In other words, if GPAY, or a supplier of similar services, can switch from Scotiabank biller services to an alternative, say Product A, without its business being substantially affected because of this switch, the product A belongs in the same product market with Scotiabank biller services.” (Mathewson Report at ¶23) [12] There are several things to note about this approach. First, it applies to business buyers. Second, it delineates the market based on what a business buyer could do, rather than what a business buyer would, or would likely, do. Third, the criterion is not the conventional SSNIP test but rather a “substantial effect” on the buyer’s business that is not well defined. Fourth, it raises the possibility of over-broad definition of the market.
1. By focusing on the business buyer and the presence or absence of a substantial effect of post-termination switching on that buyer’s business, this approach does not address whether Scotiabank depositors who are not businesses and currently use the GPAY Service would switch to other means of payment in the event of a SSNIP thereon by Scotiabank. I concluded that they would not switch in sufficiently large numbers and I stated that “it is quite clear that bank accounts at other financial institutions that accept B-Filer as a biller are not part of the relevant market for Scotiabank depositors who wish to use the GPAY Service.” (Affidavit, at ¶54). 2. It appears that Professor Mathewson has not challenged my conclusion in this regard. Rather, he asks whether GPAY, as a business customer of Scotiabank for biller services, could switch to EMTs without substantial effect. Even if
Page 5 of 25
CONFIDENTIAL LEVEL A his approach has merit in regard to business buyers in GPAY’s position, Scotiabank biller services remains a relevant product market when evaluated from the Scotiabank non-business depositor perspective. 3. Returning to Professor Mathewson’s focus on the business buyer, asking what a business buyer could do without resulting in a substantial effect on its business is a very different question than that posed by the hypothetical monopolist approach. The two approaches could be reconciled by posing the question differently: “Does the business buyer have so many good options (i.e. none of which result in a substantial negative effect on it) that a hypothetical monopolist would not impose a SSNIP?” This approach to market definition is quite similar to the hypothetical monopolist approach that I have used, at least in regard to business buyers. However, this is not the approach that Professor Mathewson has taken. 4. A related concern is that the “substantial effect” as used by Professor Mathewson is not well-defined. In contrast to the hypothetical monopolist test that defines a SSNIP as a “small but significant price increase” of 5% for at least one year, the Mathewson Report offers no definition of “substantial effect”. As discussed further below, this approach leads to a narrow conception of effects on GPAY’s business. 5. The fourth difficulty is that Professor Mathewson’s approach raises the possibility of over-broad definition of the product market. It is clear that, post-termination, Scotiabank depositors, merchants, and GPAY itself, have alternatives that were not chosen pre-termination. In effect, alternatives that were not good substitutes prior to the termination became, in his view, good substitutes after the termination. That alternatives become attractive only after termination is not surprising. Indeed, this situation is quite similar to the well-known “Cellophane trap” in which the presence of good substitutes after the monopolizing conduct appeared to indicate the absence of market power when, in fact, the broader market was a direct result of that conduct. 3 3 The Enforcement Guidelines discuss the Cellophane fallacy and its implications for market definition at ¶3.2.1(c).
Page 6 of 25
CONFIDENTIAL LEVEL A Thus, there are significant differences between my approach to product market definition and Professor Mathewson’s. [13] Market definition according to the hypothetical monopolist approach is, as Professor Mathewson notes, a hypothetical test. It is prospective and is a preliminary step in the assessment of the likely market power consequences of alleged misconduct. Professor Mathewson questions why relevant markets should be defined in this way when the economic agents’ reaction to alleged misconduct can be observed directly (Mathewson Report, at ¶78). [14] His question highlights another difference between Professor Mathewson and me. He seeks to infer the relevant product market from observing the impact of Scotiabank’s termination of GPAY’s banking services including biller status. If there is no or little harm, then he infers that the relevant market must be broader than that which I define based on the hypothetical monopolist approach. I have two comments on his approach. [15] First, I could agree that if there was clear and convincing evidence proving or disproving the lessening of competition from the conduct under review, then the preliminary step of market definition would be unnecessary. For example, if Scotiabank’s termination of GPAY’s banking services had caused GPAY to go out of business, then delineating relevant markets would add little because the market power consequences could be evaluated directly. In this regard, a leading expert in market definition issues notes: “If the natural experiment has been performed, one can measure directly the extent of a defendant’s market power, and there may be nothing gained by delineating the relevant market in which that power is held. The best reason for proceeding with market delineation may be to double check the market power conclusions reached through other means. This may be useful in some cases, but if the market is delineated on the direct evidence of market power, then market delineation is not an independent analytical path and therefore adds very little.” 4 [16] In the current matter, the facts are less clear. GPAY was not forced out of business, but the termination does affect Scotiabank depositors, GPAY’s merchants
4 G. Werden, “Market Delineation under the Merger Guidelines: Monopoly Cases and Alternative Approaches”, Review of Industrial Organization, 16, 2000, 211.
Page 7 of 25
CONFIDENTIAL LEVEL A (current and potential), and GPAY itself. There is little evidence of the reactions of economic agents to Scotiabank’s termination of GPAY’s banking services, apart from the fact that GPAY began to process payments by Scotiabank depositors by way of EMTs. Professor Mathewson does not provide or refer to empirical evidence that, for example, Scotiabank depositors switched to other banks or turned to credit cards following Scotiabank’s termination but instead he enumerates options open to those agents. [17] Absent this type of direct evidence, the market definition approach adopted by the enforcement agencies in the leading jurisdictions has much to recommend it. First, it is based on conventional economic theory; second, it constitutes an independent analytical approach, and third, there is information including independent economic analyses that can be marshaled to determine the relevant product markets under this approach. Nonetheless, I agree with Professor Mathewson that, as the Tribunal stated in Chrysler, the “ultimate test concerns the effect on the business of the person refused supplies” (Mathewson Report at ¶22). In my view, however, product market definition in this case is a necessary preliminary step in addressing this ultimate question, particularly when the harmful impact of the conduct at issue is not clear and convincing. [18] There is a second, broader issue raised by Professor Mathewson’s approach. In competitive markets that the Act seeks, business enterprises routinely come and go as more efficient firms replace less efficient competitors. Business failure, exit, and adjustment in methods of doing business are normal and, by themselves, do not imply that firms have behaved anti-competitively. Indeed, the results of the competitive process are economic efficiency, the equitable opportunity of small and medium-size businesses to participate in economic activity, and the other objectives of competition policy set out in s.1.1 of the Act. [19] However, certain types of conduct can, when adopted, also lead to failure, exit and/or adjustment by a competing business. Such conduct may be viewed as simply competition on the merits leading to economic efficiency, or it may be viewed as harming the competitive process and having an adverse effect. In many cases, it is hard to distinguish between the two interpretations based on observation of the outcome alone. The preliminary step of delineating relevant markets is an important aid in making this distinction.
Page 8 of 25
CONFIDENTIAL LEVEL A [20] With these general observations on Professor Mathewson’s approach, I now review his specific comments. For ease of reference, I adopt his headings and subheadings.
Is access to Scotiabank Biller Services a ‘Product’? [21] I disagree with Professor Mathewson’s claim that EMTs are “the most obvious” alternative because it was the option chosen by GPAY following termination of biller status by Scotiabank and certain other banks (Mathewson Report, ¶31). This behaviour does not, in my view, put Scotiabank biller service and EMTs in the same product market. [22] As I discussed in my Affidavit, the conventional test for inclusion in the product market is the SSNIP test, i.e. whether an alternative would be chosen following a small but significant increase in the price of the current product or service. Thus, both the current product or service and the alternative must be available for choice and the buyer’s behaviour following the SSNIP in the current product or service determines whether the product market is too narrow. However, when the current product or service is withdrawn completely and no longer available for choice, it is not surprising or helpful to market definition to observe that the buyer chose another alternative. [23] As indicated in the Enforcement Guidelines, evidence of switching behaviour in response to a price change prior to Scotiabank’s termination of GPAY or of plans to do so in the future would support the delineation of a broader product market.
EMT Transfer Limits [24] I also disagree with Professor Mathewson’s conclusion that “the limit on EMT transactions does not prevent EMTS from being included in the same product market with Biller Services” (Mathewson Report, at ¶38). First, however, I note that GPAY had biller status at Scotiabank until termination thereof in September, 2005; the Mathewson Report notes similarly (Mathewson Report, at ¶17). However, the Mathewson Report analyses payment data from Scotiabank for April 2005 which was “the last full month
Page 9 of 25
CONFIDENTIAL LEVEL A before Scotiabank terminated GPAY’s banking services” (Mathewson Report, at ¶34). There appears to be an inconsistency in the Mathewson Report 5 . [25] I understand that Scotiabank terminated GPAY’s biller status in September 2005. As I noted in my Affidavit, GPAY processed payments in excess of $680,000 from Scotiabank depositors in that month. In the following month, the amount of such payments fell to approximately $350,000 (Affidavit, at ¶15). I attach the payment data provided to me by counsel to GPAY as Schedule A: Monthly Payment Volumes. [26] In any case, Professor Mathewson finds fifteen transactions over $1,000 in April 2005, which constitutes 0.26% of all payments from Scotiabank depositors and 4.6% of the total value of those transactions. He also points out that GPAY’s own data for the period June 2004 to August 2005 show that payments exceeding $1,000 were 7% of all payments from Scotiabank depositors. I agree with him that the number of transactions exceeding $1,000 was low, but I regard the value of such transactions as high when expressed as a percentage of all payments from Scotiabank depositors in the relevant periods. [27] Professor Mathewson states that a Scotiabank depositor trying to make a payment of $1,250 would be unable to do so if GPAY were processing payments by way of EMT, and he states that the depositor would “retry the transfer at $1,000 and would be successful” (Mathewson Report, at ¶35). I am not sufficiently knowledgeable to know whether this statement is accurate; my understanding is that this situation is just one example of how processing payments by way of EMT is problematic and costly for GPAY to resolve. I further note that if the retailer would allow the $1,250 payment to be broken up into two EMTs then perhaps the payment could be completed in two stages. Note however that this would involve an additional EMT transfer that GPAY would have to process, and an additional EMT fee to the Scotiabank depositor and, as well, any other applicable account-level fees. [28] In addition, due to the daily limit of $1,000 on accounts receiving EMTs, the second part of the payment would have to be made on the following day, thereby making the transaction more complex for both the payer and the merchant. A merchant that
5 Data provided to me by GPAY indicate that total payments processed from Scotiabank depositors in April 2005 were $494,193.36. Professor Mathewson’s figure obtained from Scotiabank is $491,787. I do not regard the difference as significant.
Page 10 of 25
CONFIDENTIAL LEVEL A accepted payment in two stages would be at risk for the next day’s payment. As noted in my Affidavit, such considerations are among the reasons why EMTs are not part of the relevant product market; i.e. when biller status at Scotiabank is available, users thereof such as GPAY would likely bear a SSNIP thereof rather than switch to processing payments by way of EMTs. [29] It is also stated that the individuals whose payments exceeded the $1,000 limit that would have applied had GPAY been processing by way of EMTs in April 2005 had “the option to open (or use existing) accounts at other banks” where GPAY had biller status (Mathewson Report, at ¶35). I agree with this, but for the reasons indicated above and in my Affidavit, the fact that they could do so does not make accounts at other banks part of the relevant market. Indeed, it is likely that Scotiabank depositors using the GPAY Service would bear a SSNIP in the cost of using that Service at Scotiabank rather than open accounts at other banks. [30] It is also stated that there is “no evidence on the record” indicating that EMT limits prevented any merchants from signing up with GPAY (Mathewson Report, at ¶36). Up to December 2003, GPAY had biller status at seven banks (Affidavit, at ¶11), so no merchants approached by GPAY would be concerned about EMT limits up to that date. [31] A possible explanation for the absence of evidence on the record is that GPAY was not marketing its services to merchants of high-priced goods after December 2003 because EMT limits would have prevented large-value payment transactions or may have made processing thereof difficult and costly. If this was the case, then the record would not indicate that the EMT limits deterred any merchants from using GPAY. Again, the record is silent, to my knowledge.
EMT Fees [32] Professor Mathewson points out that the number of transactions processed by GPAY actually increased after Scotiabank terminated its banking business including biller status. Thus, he concludes that Scotiabank depositors were not deterred by the $1.50 EMT fee and hence EMTs and Scotiabank biller status are in the same product market. Several comments are in order.
Page 11 of 25
CONFIDENTIAL LEVEL A [33] Using Professor Mathewson’s own approach to market definition, it cannot be maintained that GPAY’s business with Scotiabank depositors was not substantially affected by Scotiabank’s termination. As noted above, its monthly dollar payment volume from Scotiabank depositors fell dramatically (i.e. by 48%) from September 2005 to October 2005. December 2005 payment volume was approximately $439,000, and the June 2006 payment volume was approximately $552,000, down 19% from September 2005. In other words, as noted in my Affidavit (at ¶15), GPAY’s monthly payment volume from Scotiabank depositors has yet to recover to the level in September 2005, the month when Scotiabank terminated GPAY. [34] By contrast, GPAY’s June 2006 payment volume from deposit customers of each of the other five banks has surpassed their respective levels in September 2005. In addition, the changes in the number of monthly transactions show that payments from Scotiabank have lagged those from the other banks:
September 2005-June 2006 Bank Change in $Payment Volume Change in Number of Transactions Bank of Montreal +79% +69% CIBC +66% +64% Caisse Desjardins +143% +85% Royal Bank +53% +84% Toronto-Dominion +106% +81% Scotiabank -19% +27% It is clear that once Scotiabank biller status was no longer available, GPAY lost significant payment volume from those depositors. It is true that the number of transactions from all banks surpassed their September 2005 levels, but the increase in transactions from Scotiabank is much lower than that at each of the other banks. [35] Based on the data in Schedule A, the aggregate monthly payment volumes from all banks including Scotiabank were below the level of September 2005 until January 2006:
Page 12 of 25
CONFIDENTIAL LEVEL A September 2005: $3.20 million October 2005: $3.14 million November 2005: $2.86 million December 2005: $3.13 million January 2006: $3.90 million Thus, it was not until January 2006 that GPAY’s total payment volume surpassed the level of September 2005. Although the number of monthly transactions generally increased over this period, GPAY’s business from other banks did not offset the losses of payment volume from Scotiabank depositors following termination in September 2005. This suggests to me that the switch to EMTs following Scotiabank’s termination did have a substantial effect on GPAY’s business. [36] Referring to the increase in the number of payments, Professor Mathewson states that the switch to EMT did not have a “major effect on Scotiabank account holders’ behaviour with respect to the GPAY service” (Mathewson Report, at ¶43). However, having regard to the value of payments, it appears that Scotiabank’s termination did have a substantial effect on GPAY and therefore under Professor Mathewson’s approach, EMTs would not be part of the market that includes Scotiabank biller services. [37] Nevertheless, I continue to disagree with Professor Mathewson’s approach to market definition. Scotiabank biller service constitutes the relevant product market because a significant price increase thereon would not likely cause online debit service providers such as GPAY to switch to processing payments by EMTs. [38] I note a mistake in my Affidavit at ¶15 where I stated that monthly payment volume for October 2005 fell at Scotiabank, increased at three banks and declined at two others. In fact, as shown above, payment volume rose in October 2005 at each bank other than Scotiabank.
Effect on GPAY’s Costs [39] I agree with Professor Mathewson that it would be desirable to quantify the costs that GPAY incurs to process payments by way of EMTs (Mathewson Report, at ¶44). These costs include the quantifiable costs such as increased accounting costs, increased bank fees, and increased programming and other direct costs identified by Mr. Grace.
Page 13 of 25
CONFIDENTIAL LEVEL A However, the impact of the switch also includes costs that would be hard to quantify yet are nonetheless real in the sense that they represent use of real resources. For example, the costs to GPAY of hiring two new employees would not be limited to their salaries and benefits, but would also include management supervision, training, and evaluation of performance. These costs of doing business are apparent to all organizations that hire personnel, but organizations (particularly small businesses) do not quantify these costs. [40] It is appropriate to ask whether the additional employees were hired to deal with the increased volume of EMTs following termination by Scotiabank or whether they would have been hired anyway. In this regard, my information, which comes from GPAY management, is that the new staff were hired in order to deal with the problems arising from the additional volumes of EMTs. Both the direct hiring costs and the indirect management-related effects are costly and should be included in any reckoning of the substantial effect on GPAY’s business. Moreover, the resources that GPAY devoted to handling EMTs have alternate uses, for example, new product development, marketing initiatives, or simply greater profits and dividends. Each of these alternatives has a cost (in economic terms, the “opportunity cost”) that must be included in reckoning the substantial effect, even when that reckoning is a qualitative assessment. [41] Thus, I do not feel that “with the available information, it is impossible to determine whether the alleged increase in costs suffered by GPAY … could be said to have substantially affected GPAY’s business” (Mathewson Report, at ¶44). In my view, the real problem is that no precise definition of “substantially affected” has been provided.
Are ‘EMT Business Deposit Accounts’ a Product? [42] In my Affidavit, I considered whether EMTs and the business accounts that accept deposits thereby are in the same product market as Scotiabank biller services. I did not consider the question whether EMT business deposit accounts are a product for the purposes of the current inquiry or whether business deposit accounts and personal deposit accounts are sufficiently similar to each other that GPAY could use the latter interchangeably with the former. Professor Mathewson states that “there appears to be no evidence on the record” that indicates that they are not (Mathewson Report, at ¶47).
Page 14 of 25
CONFIDENTIAL LEVEL A [43] In one sense, the similarity of business and personal deposit accounts is not relevant. If GPAY management conducted GPAY business through personal accounts at Scotiabank while it had biller status at Scotiabank, then those personal accounts must have been used to receive EMTs from depositors at other banks that had terminated GPAY’s biller status (i.e. Toronto-Dominion Bank, Canadian Imperial Bank of Commerce). In other words, when doing so, GPAY was not processing payments from Scotiabank depositors and thus whether Scotiabank biller services is, as I have concluded, a relevant product market in relation to the means of providing online debit payment to Scotiabank depositors (Affidavit, at ¶107) is unaffected by the use of such accounts. [44] It is true, as stated in the Mathewson Report (at ¶46), that after Scotiabank terminated GPAY’s biller status and business accounts and all personal accounts of GPAY managers, GPAY began to process payments from depositors at Scotiabank by way of EMTs, having already begun to use EMTs for payments from depositors from Toronto-Dominion Bank and Canadian Imperial Bank of Commerce following those banks’ terminations. For the reasons given above concerning the methodology of market definition, this does not establish that the deposit accounts (whether business or personal) are in the same product market as Scotiabank biller service. [45] I am not sure whether the characteristics of business deposit accounts are sufficiently similar to personal accounts that users would switch to the latter in the event of a significant price increase in the former, but it is my understanding that banks do not permit holders of personal accounts to use those accounts for business purposes, except perhaps for occasional business transactions. If this is correct, then GPAY’s volume of transactions into personal deposit accounts would likely lead the bank in question to terminate such accounts. These matters are perhaps better addressed by someone with retail and/or business banking experience. [46] Thus, subject to this more detailed inquiry, personal deposit accounts are not in the same market as business deposit accounts, and biller status at Scotiabank remains a relevant product market that excludes business and personal accounts that accept deposit by way of EMTs.
Page 15 of 25
CONFIDENTIAL LEVEL A GPAY’s Performance After Scotiabank’s Termination of its Banking Services [47] Professor Mathewson believes that GPAY turned to EMTs without suffering substantial harm after Scotiabank terminated GPAY’s banking services in September 2005 (Mathewson Report, at ¶48-¶51). As stated above, I disagree with this approach to product market definition; GPAY’s behaviour when Scotiabank biller service was no longer available is not informative in the hypothetical monopolist approach to market definition. [48] As stated above however, even on his approach to market definition, I cannot agree that that GPAY was not substantially affected by the termination. Noting the drop in Scotiabank payment volumes from September to October 2005, Professor Mathewson states in respect of those volumes that i. “volumes recovered through the rest of 2005” ii. “Between October and December 2005, dollar volumes of GPAY transactions increased by almost 25 percent” iii. “This indicates that the reduction in GPAY volumes processed from Scotiabank’s accounts after Scotiabank’s termination was a short-term phenomenon. GPAY’s Scotiabank payments have now fully recovered their pre-termination levels.” (Mathewson Report, at ¶50). [49] My review of monthly payment volumes in Schedule A indicates that payment levels from Scotiabank accounts have not recovered and that those payment levels between October and December 2005 continued below the September level. Payments have “now fully recovered their pre-termination levels” if Professor Mathewson’s comparison is with August 2005, but by June 2006 they had not exceeded the level attained in September 2005, the month in which termination occurred, despite having done so from all other banks. [50] Professor Mathewson excludes the September 2005 payment volumes, from depositors at Scotiabank and apparently at the other banks, from his calculations because GPAY’s volumes in that month “appear to be anomalously high” (Mathewson Report, at ¶50, footnote 40). It is true that the September 2005 payment volume from Scotiabank was the highest monthly level attained to that date. However, the approximately 37% change over the preceding month was not a unique occurrence: February 2004 volume
Page 16 of 25
CONFIDENTIAL LEVEL A increased by 94% over the previous month, and May 2004 volume increased by 47% over the previous month; July 2005 volume increased 22%. [51] Data analysis in business and economics frequently uses quarterly calendar periods (January-March, April-June, July-September, October-December) for measurement and comparison. Had I done so in the matter at hand, I would not exclude information for September 2005. Large month-to-month changes in GPAY’s payment volume are common and new monthly highs are established frequently. In such circumstances, exclusion of a particular month’s datum from quarterly comparisons on the basis that it is anomalous omits useful statistical information. [52] Looking at other banks’ payment volume also shows some large changes. GPAY’s September 2005 payment volume from Bank of Montreal increased approximately 24% over the preceding month; volume from Caisses Desjardins fell by approximately 38%. [53] I noted above that GPAY’s volume from Scotiabank depositors fell by 48% in October 2005. By itself, this fact is not telling. It has significance however when joined with the fact that volumes at all other banks increased in October, that GPAY’s monthly volume from Scotiabank has not yet returned to its September 2005 level despite having done so at all other banks, and that Scotiabank terminated GPAY’s banking business in September 2005. However, I do not dispute that September 2005 was the month with the highest payment volume from Scotiabank attained to that date. [54] It appears that the differences between Professor Mathewson and me are in part due to his use of 3-month comparisons, whereas I have drawn conclusions from monthly data.
EFFECT ON COMPETITION [55] Professor Mathewson states that the fundamental issue in a refusal to deal case is whether the alleged refusal will preserve or enhance the market power of allegedly refusing firm and that in the matter at hand, the question is whether Scotiabank’s termination of GPAY’s banking services increases any hypothetical market power accruing to Scotiabank’s Interac Online (Mathewson Report, at ¶52).
Page 17 of 25
CONFIDENTIAL LEVEL A [56] I prefer to state the issue in terms of s.75: Will Scotiabank’s termination of GPAY’s banking services, including biller status, have an adverse effect on competition in a market? In economic terms, the issue is not simply the impact on GPAY’s business, but whether the effects of the termination, taken as a whole, constitute an adverse effect on competition. I discuss the anti-competitive effects in my Affidavit and I conclude that the overall effect is adverse. [57] The important related question is whether a proven adverse effect has a legitimate business justification, an issue that I did not address in my Affidavit. However, since Professor Mathewson has done so, I will comment briefly on his views. [58] I agree with Professor Mathewson that the GPAY Service and Interac Online are not close “substitutes” currently (although Interac Online’s merchants could switch because GPAY is functionally similar) because of the lack of overlap in their respective merchant bases. There is some ambiguity in whether he means only their current merchant bases or the future merchant bases as they evolve. It is quite possible that the merchant bases of the two services could evolve similarly, in which case they could be closer substitutes in the future. As I discussed in my Affidavit, GPAY has been making efforts in this regard to diversify its merchant base to include more merchants offering large-value goods and services. [59] I think that Professor Mathewson is alluding to the future evolution of the merchant bases as well as the current situation. He notes that Interac Online’s current merchant base currently consists of twenty-eight merchants (Mathewson Report, at ¶58), and I infer that it is his expectation that that base will grow. He also expresses doubt that GPAY’s efforts to diversify its merchant base will succeed (Mathewson Report, at ¶68). Thus, Professor Mathewson is forecasting how the merchant bases of the two services will evolve and, on his forecast, he appears to conclude that the GPAY Service and Interac Online will not be close substitutes in the future. Thus, he concludes that Scotiabank’s termination has no market power consequences now or in the future (Mathewson Report, at ¶54). [60] Any such forecast must take into account the likelihood that a merchant would choose only one online debit payment service since both GPAY and Interac Online offer direct payment from depositors’ accounts. They could compete against each other on the
Page 18 of 25
CONFIDENTIAL LEVEL A basis of merchant fees and/or elements of service such as suspense periods with pro-competitive effect. However, GPAY cannot serve merchants offering large-value goods and services with the current limits on EMTs. Such a merchant choosing between the two systems for processing payments by Scotiabank depositors will select Interac Online for this reason alone. Hence, quite independent of the effect on GPAY’s costs and profitability of processing by way of EMTs is the reduction of competition and of the pro-competitive effects thereof in the market for merchants. [61] Professor Mathewson also feels that the market power consequences of the termination must be evaluated in light of other suppliers of “substitute services, such as credit cards” (Mathewson Report, at ¶54). He also raises other payment options such as prearranged debit, credit cards and EMTs in his discussion of the impact of termination on Scotiabank consumers (Mathewson Report, at ¶63). He states: “There is no evidence presented by the applicants that GPAY will develop into a viable and sustainable payment mechanism as a substitute for instruments such as debit and credit cards and other online payment instruments.” (Mathewson Report, at ¶64) As indicated in my Affidavit, I believe these options are not part of the relevant product market as conventionally defined in competition policy. Once again, I acknowledge that GPAY’s marketing efforts to large-value merchants have not generally been successful to date. [62] However, this does not preclude success in the future absent Scotiabank’s termination. Since forecasting is difficult, it is premature to conclude that GPAY’s efforts will not succeed in competition with Interac Online or with other payment instruments. Indeed, it is the heart of the competitive process that the market determines which competitors succeed and which do not. Accordingly, I do not forecast whether GPAY’s efforts to diversify its merchant base would or would not succeed absent the termination. Thus, I regard the statement that: “… GPAY’s current business portfolio may be the best indicator of its viable future business portfolio, a business that Scotiabank does not wish to serve.” (Mathewson Report, at ¶68) as unwarranted.
Page 19 of 25
CONFIDENTIAL LEVEL A [63] Regarding Professor Mathewson’s comments on Scotiabank’s business rationale for the termination of GPAY’s banking services, I agree that Scotiabank has significant brand name recognition that is sunk capital. He also states: “… Scotiabank terminated its banking relationship with GPAY for a number of reasons, including the fact that GPAY’s manner of conducting business requires banking customers to disclose their online banking password. Scotiabank also had regard to legal, reputational, and regulatory risks associated with continuing to do business with GPAY.” (Mathewson Report, at ¶65) “… Permitting GPAY to have access to individual accounts and customer passwords violates the terms under which Scotiabank provides these banking services. …” (Mathewson Report, at ¶66) [64] I feel that it would be beyond my expertise to comment on whether the “disclosure” to which Professor Mathewson refers is the correct description of GPAY’s manner of conducting business. In addition, I do not know whether GPAY’s access “violates” the terms under which Scotiabank provides banking services. These matters are beyond my expertise. [65] I accept that there could be reputational and regulatory risk. From my general knowledge and my interest in financial-sector policy and regulation, if Scotiabank failed to comply with applicable laws and regulations, then its reputation could indeed suffer if such non-compliance became public. However, it is not within my expertise to comment on whether Scotiabank is at risk of non-compliance.
DISCUSSION OF THE SCHWARTZ REPORT [66] In his discussion of my Affidavit, Professor Mathewson makes certain comments and observations that I have not addressed above. At ¶74, he correctly calls attention to my statement that “B-Filer requires biller status at the financial institutions whose depositors wish to use the GPAY Service” (Affidavit, at ¶11). I have erred in this regard; my statement should have been that B-Filer requires access to depositor accounts at those financial institutions. As I recognized in my Affidavit, processing transactions could be done in other ways, although I do not believe that those ways are part of the relevant market for Scotiabank depositors.
Page 20 of 25
CONFIDENTIAL LEVEL A [67] Professor Mathewson calls attention to my application of the SSNIP test when the bank depositor has not exhausted his/her “free” debit transactions (Mathewson Report, at ¶78). In this circumstance, my conclusion based on the hypothetical monopolist approach accords with common sense: a 5% increase in the account-level fee is in fact a zero increase in the marginal cost of the transaction, but whether one accepts this or not, common sense dictates the Scotiabank depositor would not change banks. Accordingly, for Scotiabank depositors currently using the GPAY Service, the market is limited to that Service at Scotiabank. [68] My Affidavit also points out that the increased account-level fee would apply to all future debit transactions by the Scotiabank depositor and may lead him/her to consider changing banks. Moreover, as I noted, there are other ways of imposing a SSNIP on that depositor that may also lead him/her to consider changing banks. However, for the reasons given in my Affidavit, I conclude that the depositor would bear the price increase rather than switch. Accordingly, the relevant market for Scotiabank depositors who use the GPAY Service excludes accounts at other banks that accept GPAY as a biller. [69] Professor Mathewson states that there is “no hard evidence” that Scotiabank depositors would not switch to other banks in the event of a price increase. He suggests that depositor mobility would be high for those who already have accounts at other banks (Mathewson Report, at ¶80). I was unable to find any research indicating how many bank depositors who use the GPAY Service are in this situation. If such evidence exists, then it should be presented; indeed, it may be that Scotiabank has this information. Absent such information, Professor Mathewson’s suggestions are speculative. [70] Professor Mathewson further states that my discussion of the features and use of credit and debit cards is “qualitative and non-empirical” (Mathewson Report, at ¶81). I dispute this, as it is neither qualitative nor non-empirical to point out that Scotiabank specifically targets its online debit payment product at current debit users who do not hold credit cards and already bank online (Affidavit, at ¶60). This evidence tends to suggest that Scotiabank deposit customers using the GPAY Service would not switch to credit cards under the SNNIP test for market definition. [71] With regard to my conclusion that Scotiabank depositors using the GPAY Service would not switch to credit cards in the event of a SSNIP, Professor Mathewson states that
Page 21 of 25
CONFIDENTIAL LEVEL A evidence on what those depositors actually did following termination would be required (Mathewson Report, at ¶81). It is noteworthy that he presents no evidence about post-termination behaviour of Scotiabank depositors’ use of credit cards. [72] Moreover, such evidence would not be dispositive because, as I have discussed above, markets for the purpose of competition policy are not properly defined in that manner. If there is information on the use of credit cards by Scotiabank depositors who used the GPAY Service prior to the termination, such evidence would be informative. In this regard, such evidence would not be limited to credit cards issued by Scotiabank. However, at the examination for discovery on June 12-13 2006, the Scotiabank representative was unable to say how many Scotiabank depositors held credit cards issued by Scotiabank, let alone how many of such depositors used the GPAY Service. [73] Finally, in my analysis of relevant markets, I concluded that the GPAY Service and Interac Online are in the market for online debit payment service for Scotiabank depositors (Affidavit, at ¶82). I recognize that their respective merchant bases do not overlap currently. I would expect that, even in the future where GPAY had biller status, the merchant bases of the two systems would still not overlap because it is unlikely that any merchant would sign up for both GPAY and Interac Online. However, there could be direct competition between them as they compete for (i) individual merchants and (ii) for representative merchants in various categories. Thus, GPAY might succeed in signing up Computer Retailer A and Interac Online might sign up Computer Retailer B even though neither A nor B would find it worthwhile to sign up for both online debit services (Affidavit, ¶80-¶81). [74] In this environment, if Scotiabank imposed a significant price increase on its depositors using the GPAY Service, it is likely that those depositors would switch to merchants signed up with Interac Online (Affidavit, at ¶76), especially since Scotiabank is not charging a transaction fee for purchases on Interac Online. This substitutability places both the GPAY Service and Interac Online in “the market for online debit payment service for Scotiabank customers” (Affidavit, at ¶107). If one believed that, in this environment, this switch would not occur, then one has a basis for excluding Interac Online from this product market.
Page 22 of 25
CONFIDENTIAL LEVEL A [75] In the matter at hand, the fact that payment instruments operate in “two-sided” markets is a consideration in delineating relevant markets. One reason why Scotiabank depositors would not switch in the environment that I have described is that the merchants were not in the same categories. This is the case now, and it will remain the case as long as Scotiabank’s termination of GPAY’s biller status continues because, relying on EMTs, GPAY is unable to attract the large-value merchants that Interac Online can attract. Taking the two-sided nature of the online debit payment market into account allows the market definition exercise to assist in determining the ultimate question of Scotiabank’s termination and whether it has had an adverse effect on competition. [76] Professor Mathewson seems to indicate that the switch would not occur. However, he is analyzing a different environment, one in which, per his forecast, GPAY does not succeed in attracting “non-gambling merchant business” and therefore is not “viable and sustainable” in that segment (Mathewson Report, at ¶82). [77] I do not assume that GPAY will succeed in this regard and my approach to market definition does not turn on such an assumption. Indeed, for the purpose of market definition, it is not important whether GPAY turns out to be successful or not; competition in the marketplace will decide its future success. The relevant question is whether Scotiabank’s termination has an adverse effect on that competition. In order to answer this question, one must adopt analytical procedures for market definition that do not predetermine the outcome. [78] In this regard, Professor Mathewson correctly states my view “that GPAY is attempting to compete with Interac Online” (Mathewson Report, at ¶82). In my view, this observation is sufficient to motivate the analytical approach I have taken. [79] However, I disagree with Professor Mathewson’s approach and his characterization that my approach “simply assumes that these two services will compete at some unknown future date” (Mathewson Report, at ¶82).
Page 23 of 25
CONFIDENTIAL LEVEL A Schedule A: Monthly Payment Volumes Bank Month BNS BOM CIBC Jan-03 $2.00 $3.00 Feb-03 $102.03 $101.38 $103.04 Mar-03 $21.01 $0.02 Apr-03 $0.02 $0.03 May-03 $1.01 $0.04 Jun-03 $1.04 $0.00 Jul-03 $17.50 $1.00 Aug-03 $36.01 $6.01 $221.50 Sep-03 $753.04 $3,001.56 Oct-03 $8,693.29 $5,984.78 $6,668.06 Nov-03 $34,307.95 $11,003.92 $7,781.37 Dec-03 $56,632.10 $28,791.18 $3,384.35 Jan-04 $51,561.49 $59,656.24 $4,230.91 Feb-04 $100,213.62 $47,104.29 $13,309.93 Mar-04 $117,186.67 $70,342.25 $48,407.98 Apr-04 $120,633.67 $147,902.31 $86,551.20 May-04 $177,110.74 $140,380.59 $121,346.83 Jun-04 $213,604.59 $151,723.86 $120,539.86 Jul-04 $258,071.86 $186,317.30 $136,271.07 Aug-04 $324,926.38 $132,641.35 $128,234.26 Sep-04 $293,275.01 $188,759.81 $162,323.98 Oct-04 $309,067.26 $291,499.46 $219,192.42 Nov-04 $368,382.93 $280,448.33 $235,585.18 Dec-04 $381,154.59 $268,469.76 $250,037.01 Jan-05 $420,168.24 $292,330.10 $263,544.80 Feb-05 $414,578.78 $346,516.01 $285,245.79 Mar-05 $449,985.37 $390,045.79 $339,074.93 Apr-05 $494,193.96 $430,891.20 $363,008.70 May-05 $518,792.03 $418,830.93 $357,785.35 Jun-05 $484,978.34 $422,097.17 $360,833.42 Jul-05 $593,716.00 $389,066.06 $386,626.65 Aug-05 $500,277.40 $382,985.50 $343,124.44 Sep-05 $683,332.30 $473,714.49 $357,826.56 Oct-05 $353,218.42 $563,437.87 $406,581.76 Nov-05 $388,074.46 $575,916.53 $427,728.05 Dec-05 $438,573.44 $624,261.70 $417,076.34 Jan-06 $402,705.98 $766,993.58 $495,580.38 Feb-06 $504,172.33 $818,313.24 $487,720.17 Mar-06 $583,756.45 $899,771.93 $592,784.67 Page 24 of 25
CP RBC TD $2.00 $14.00 $4.00 $103.06 $101.10 $1.01 $41.52 $34.06 $0.00 $0.04 $1.03 $0.02 $5.04 $1.04 $1.00 $222.00 $503.02 $42.00 $15.12 $2.07 $1,512.01 $100.03 $0.01 $117.34 $6,210.86 $9,577.03 $13,731.97 $13,317.27 $21,920.03 $21,994.33 $19,231.66 $40,256.69 $38,461.00 $41,103.04 $44,837.86 $76,959.30 $68,130.26 $725.01 $124,703.55 $119,508.16 $11,892.02 $166,960.45 $189,402.30 $11,606.89 $203,341.93 $216,246.53 $8,174.59 $209,639.98 $242,906.11 $28,489.98 $257,314.96 $272,816.66 $36,421.13 $282,127.42 $298,366.31 $46,359.31 $264,554.75 $337,530.70 $33,890.48 $306,867.20 $350,068.11 $22,214.35 $349,345.43 $372,358.46 $16,482.64 $429,080.22 $393,336.50 $14,290.72 $430,891.64 $422,948.16 $26,239.15 $568,621.22 $513,415.27 $36,850.48 $721,493.45 $496,410.27 $26,442.30 $885,003.63 $535,509.69 $43,144.36 $818,425.42 $571,454.41 $50,778.42 $830,801.67 $614,839.75 $1,144,626.0 $59,683.48 9 $577,874.28 $1,040,254.9 $37,042.59 3 $606,104.31 $1,094,032.0 $44,063.03 3 $676,282.34 $49,738.77 $772,662.81 $649,736.03 $91,055.73 $891,487.22 $665,870.12 $1,160,914.9 $95,485.30 0 $973,867.72 $1,113,992.3 $1,096,667.5 $101,403.71 1 9 $135,634.15 $1,634,785.4 $1,319,659.5
CONFIDENTIAL LEVEL A Apr-06 $564,181.57 $900,917.83 $582,820.44 May-06 $548,357.81 $928,183.77 $629,241.19 Jun-06 $552,021.22 $847,188.40 $592,865.67 Page 25 of 25
4 1 $1,518,511.3 $1,286,897.6 $95,417.20 7 6 $1,600,373.2 $1,411,834.2 $78,230.28 4 6 $1,591,865.9 $1,248,613.8 $90,182.92 8 9