Jump to Content Jump to Main Navigation

Part IV Measurement, 14 Cost-based Approaches

From: Cartel Damages: Principles, Measurement, and Economics

Cento Veljanovski

From: Oxford Competition Law (http://oxcat.ouplaw.com). (c) Oxford University Press, 2021. All Rights Reserved.date: 18 June 2021

Cartels — Damages

(p. 232) 14  Cost-based Approaches

14.01  Cost-based and margin approaches are attractive to courts because they rely on actual data and are easily understood. There are several variants:

  1. 1.  A bottom-up approach using the defendant’s marginal or average costs plus a competitive mark-up or profit margin to calculate the but for prices and overcharges.

  2. 2.  More recently the English courts have adopted pure cost-based approaches which treat cost as a proxy for benefits or to take account of cost inefficiencies and cost savings during the cartel period.

Main Points

  • •  The non-infringement prices can be calculated using a measure of avoidable or marginal cost plus a reasonable or competitive gross profit margin.

  • •  The English High Court in BritNed1 awarded cost-based damages for the defendant’s built-in inefficiencies and common costs savings. These damage measures are suspect as is the general approach adopted by the court.

  • •  In bid-rigging cases the bid-rigging cost ratio approach can be used.

  • •  Other cost-based approaches have been used including the economists’ Merchant Indifference Test (MIT).

A.  Cost-plus Approach

14.02  The simplest version is to collect data on the cartel’s avoidable, marginal, or average costs, and add to this a competitive or pre-cartel margin to determine but for prices. If the cartel’s prices are found to exceed the prices calculated using the firm’s marginal costs plus a competitive margin, the difference is regarded as the overcharge.

(p. 233) 14.03  This approach is attractive because it is based on the defendant’s cost data, the prices charged by the defendant, and evidence of what might be regarded as a reasonable or competitive mark-up on these costs.

14.04  The cost-plus approach can also be used to qualify the lost profit claim provide that there is evidence of, or a method of, determining lost sales.

14.05  The approach has several drawbacks:

  • •  Obtaining data on the appropriate costs is often difficult. The defendant may not have retained detailed data on the historical costs of production especially for cartels that have operated for some time or in industries which produce different types of the products. It is possible to use proxy costs drawn for indices or the prices of the principal raw materials and inputs. The principal cost of bread is flour, so it is not that hard to use the relationship between flour costs and bread production costs as a basis for a cost approach. But for other industries the relevant costs are difficult to determine especially if there is limited data in the hands of the defendant. Moreover, the argument can always be made that the cost proxy is not an adequate proxy for the actual costs, and therefore should not be used (as the court decided in BritNed2).

  • •  Obtaining data on or estimates of firms’ and industry marginal costs is even more difficult especially for extended periods and many firms. Typically, accounting costs, average costs, and fully allocated costs are used which do not conform to the economists’ avoidable costs concept. The use of average accounting costs may lead to an underestimate of the losses. This will be especially so where costs include elements of fixed costs. On the other hand where the firm prices its products in a way which departs from marginal or incremental cost pricing, then it is arguable that these costs should be used to calculate damages. Further, where the cartel has persisted for a considerable period, the firm may have adjusted product techniques or incurred additional fixed and common costs which are in principle also recoverable.3

  • •  Even where cost data is available there will be questions as to what cost items should be included and excluded from the calculation. Ideally only avoidable costs should be used, although this may not be such a big issue if the products are sold in quantity and are homogeneous. In industries where there are large fixed and common costs, identifying the specific unit costs for individual product types produced by the cartel will be difficult, arbitrary, and contentious.

  • •  The identification of the appropriate counterfactual gross margin is also be an area of uncertainty and dispute. If the counterfactual is an oligopolistic industry the margin earned in the absence of the cartel may be larger than the competitive margin. This requires care in identifying what is the appropriate counterfactual. Having said this, as a practical matter there will be evidence of gross margins earned by the firms in the past, by similar industries, and generally which can form the basis for a claim.

14.06  If the cartel has existed for several years’ costs, prices, and margin will change, as will the non-infringement prices and margins. Price will be affected by factors other than the presence of a cartel, and by cyclical factors beyond the control of the cartel. Mark-ups are known to be pro-cyclical, i.e. higher in buoyant periods than in periods of recession or where (p. 234) demand is falling.4 Where this is the case, the counterfactual price and gross margin will differ over the period of the infringement.

B.  Other Cost-based Damages

14.07  In BritNed the English High Court awarded damages based on two other costs factors: baked-in inefficiencies and common cost savings.

1.  Baked-in inefficiencies

14.08  According to the High Court in BritNed there were ‘baked-in’ inefficiencies that inflated ABB’s direct costs. ABB had used thicker more expensive cables than its competitors. This was not pleaded by BritNed but a finding of the court based ‘on a few internal ABB documents and the fact that ABB lost bids in the post-cartel period’.5 The court said that had ABB used a thinner 1,000 MW cable there would have been an assessed 15% cost saving on the copper content which represented the equivalent of a 2.6% gross margin.6

14.09  There are problems with this approach. According to the evidence ABB used thicker copper cables in both cartel and post-cartel periods. This suggests that the use of thicker cables was not due to the cartel at least on a ‘but for’ causation test. The alternative view taken by the court was that the allocation of tenders by the cartel enabled ABB to secure the BritNed project which it would not have otherwise. The evidence for this, said the court, was ABB’s failure to secure comparable projects in the post-cartel period. However, this does not explain why: a) ABB persisted with this assumed cost ‘inefficiency’ after the end of the cartel; and b) BritNed accepted an obvious and expensive inefficiency in the ABB tender. The court simply asserted that this was cartel-induced but it could equally have been a production decision independent of cartel albeit an inefficient one.7 This would explain why BritNed persisted with using thicker more expensive cables in the post-cartel period. The evidence supporting or contradicting either view was simply not available to the court.

14.10  The court also took contradictory positions on ABB’s costs. When dealing with the claimant’s expert the court rejected the proposition that ABB’s costs were inflated by the operation of the cartel, yet it used the claimant’s expert’s observations to then find that costs were inflated and that this was a compensable head of damages. However, there is a problem with the court’s economics and factual logic on this count. It is that a costs inefficiency would only be compensable if it had been passed onto the claimant in a higher contract price, which the court found it had not.

(p. 235) 2.  Common cost savings

14.11  The court in BritNed found that because members of the cartel were allocated projects their overall common costs of tendering could be assumed to be lower. These ‘cost savings’, said the court, should be compensated. Yet there was no documentary or other evidence of these savings. Instead the court estimated these by, first, calculating the difference between the ABB’s average gross margin during the cartel and non-cartel periods. This came to a gross margin difference of 5.65 percentage points. This excess margin was, said the court, made up of higher profits and common cost savings. To separate the cost savings component the court multiplied the excess margin by ABB’s failure rate in securing projects in the post-cartel period.8 This, said the judge ‘suggests that 1.9% of the “overcharge” is attributable to the cartel savings I have identified’.9

14.12  Higher project prices and lower common costs are two different things. The higher contract price, if there was one, exploited the inelastic demand offset by BritNed’s buyer power, and was not cost-induced. On the other hand, common cost savings would not have directly affected the gross margin nor necessarily increased the contract price. Indeed, according to the court it did the opposite: ‘These [common cost] savings were competed away—in the case of the BritNed Interconnector— by ABB: but all that means is that ABB chose to allocate some common costs to other projects. That does not mean that BritNed is not entitled to a share of these cartel savings.’10

14.13  Since the supposed common cost savings were ‘passed-on’ to BritNed in a lower contract price it suffered no loss and should not be awarded damages.

14.14  The court did not follow this logic. It said that because ABB’s other customers had not been given a share of these cost savings, BritNed should be awarded damages. It based its decision on the view that cost-savings arising from the cartel should form part of the relevant counterfactual. This is incorrect. The appropriate counterfactual is the common costs absent the cartel, which by the court’s own logic were higher. The common cost savings cannot be treated as a ‘loss’ to the claimant even if these gave the members of the cartel a higher net margin. At worst they open the grounds for exemption under Article 101(3), which was not pleaded by either party or discussed in the judgment. The court’s confusion here arises from its focus on the costs and gains to the cartel in general rather than whether the cartel caused harm to BritNed.

14.15  The way the Court calculated the common costs savings is also highly problematic. There was no documentary evidence of common cost savings. Instead the judge derived and imputed the reduction in common costs using an arithmetic calculation based on the proportion of ABB’s unsuccessful bids during the post-cartel period as a proxy. There were three problems with the court’s calculations:

  1. 1.  The court found that BritNed was not overcharged and did not suffer any loss due to the reduction of common costs. The court said it had been competed away in the price negotiations between ABB and BritNed.

  2. (p. 236) 2.  There was an error with the court’s arithmetic, although this is not central to the present criticism of the court’s approach.11

  3. 3.  The available evidence used by the court (Table 6 of the judgment) did not show that the average gross margins during the cartel period were ‘excessive’. The court calculated and compared simple average gross margins for the cartel and post-cartel periods. This assumed that the type and mix of projects in both periods were similar. They were not. The average contract price for all post-cartel projects was €1.3 billion (€2.2 billion for lost bids and €993 million for successful bids) compared to only €491 million for ABB’s bids during the cartel period. The projects allocated to ABB during the cartel-period were on average one-tenth the size of comparable ABB projects in the post-cartel period. ABB’s projects during the cartel period also varied markedly, with higher gross margins on small projects (several projects with contract values less than €10m had gross margins of 30% to 40%) which pushed up the cartel-period average gross margin. Given these large differences in project size the more appropriate comparison would have been between weighted averages. Using contract prices as weights, the weighted average gross margin for the 15 cartel-period submarine projects was 20.8% compared to 20.3% for all post-cartel submarine projects. This wipes out the difference in average gross margins relied on by the court to calculate the common cost savings.

14.16  On appeal ABB won its cross-appeal to strike-out cost savings damages. The Court of Appeal found that the trial judge had made ‘an error of law’.12 His approach violated the compensatory principle by looking at the defendant’s alleged gains rather than the claimant’s loss; that his attempt to translate the purported cost savings into an ‘overcharge’ was a mere assertion ‘not open to the judge’, and in any case the judge had expressly found that any cartel savings had been competed away with no effect on the price of the BritNed project. The Court of Appeal decision makes clear that judges must base their decisions on the evidence before them, and that they are to focus exclusively on the loss to the claimant and not the gains to the defendant.

C.  Wider Issues Raised by BritNed

14.17  The use of the gross margin in BritNed raised a number of other issues:

  • •  Project based claim. The BritNed decision differs from those usually associated with secret cartels. It concerned one massive infrastructure project where the specifications, terms, and prices were bespoke, and directly negotiated by the parties. This (p. 237) influenced the court in several ways. First, it could not be argued that the claimant was ignorant of the nature of the costs, project specifications, and reasonableness of the price. The court found that BritNed was able to negotiate reductions in ABB’s price to such an extent that it concluded that there was no price overcharge. Secondly, it enabled the court to undertake a very detailed forensic examination of the documentary and witness evidence on how the bid was put together, the knowledge of those putting together the bid, the knowledge of the claimant, the history of negotiations, and way the contract price was negotiated. As the court said it was concerned with the specific circumstances of and loss to the claimant, and not the loss in general or on average to purchasers of ABB’s submarine cable projects during the cartel period.

  • •  Confused counterfactuals. The court set out a counterfactual which was at odds with the way it quantified damages. The court said that the overcharge was the difference between the project price and price in the absence of the cartel whichever party BritNed would have contracted with in the counterfactual. The counterfactual was not, said the court, a comparison between ABB’s submarine project prices during and after the cartel-period. Yet the counterfactual adopted by the court was precisely this. The only comparison made in the judgment was between ABB’s prices, costs, and margins on different projects. There was no data on the tenders of other members of the cartel or those not in the cartel, and crucially how many competing bids there were for each ABB tender or the dates of the various ABB tenders which might be useful to determine whether market conditions had changed over what was a long period. Arguably the latter failed because they were too expensive, or as the court put it they were ‘were inferior—including inferior as to price—to the winning bids’ and therefore ‘uncompetitive’.13 The reason, as the court remarked, was that there was no data on non-ABB tenders as neither party had requested the relevant data (which would have had to be sought from third parties). Put another way there was no way to implement the correct counterfactual, and so that court compromised with the consequence that it cost-based calculation were inherently flawed.

  • •  Was the defendant’s gross margin comparison valid? As already commented the defendant’s expert’s gross margin approach was inherently flawed (para. 13.33). As the court put it ABB’s failed post-cartel tenders ‘were inferior—including inferior as to price—to the winning bids’ and therefore ‘uncompetitive’.14 The implication being that the gross margins calculated for ABB’s failed tenders were higher than the appropriate competitive counterfactual gross margins, and therefore favourable to showing that ABB had not overcharged BritNed. Their use to derive common cost savings was therefore dubious.

  • •  Costs and experts. Both parties used their economist to make statements, or refuse to make statements, about costs—whether they were legitimate, reliable, efficient, and so on. As the court pointed out the two economists were not experts in submarine cable projects and their costs (echoing similar criticism made in Sainsbury’s v MasterCard15 (p. 238) about the use of economists to make factual statements). ABB, said the court, should have organized evidence on costs from those with competence in the area.

  • •  Lack of evidence. As we have seen the court accepted there was a near complete absence of factual evidence on cost inefficiencies and common costs.16 It nonetheless pushed ahead to ‘bridge’ the ‘gaps’, as it put it, with a ‘broad brush’.17 This consisted of assuming that there were cost inefficiencies and savings, and using a questionable arithmetic calculation as a proxy for costs savings. One can easily predict the court’s reaction had one of the experts advanced a similar ‘broad brush’ approach to damages quantification.

  • •  Price or cost effect. The court’s cost-based approach to damages ran counter to the pleadings of the claimant and to some of its own statements. The claimant sought compensation for a cartel induced price overcharge and not cost factors. The court said: ‘in terms of overcharge, there is no difference between the two’.18 There was and is a major difference as was obvious from the fact that the defendant’s approach using gross margins found no price overcharge while the court’s cost-based approach showed about a 5% overcharge.

D.  Bid-rigging Ratio Approach

14.18  The quantification of damages for collusive tendering will often be based on a comparison of gross margins.19 As already discussed econometric and gross margin, and cost-based approaches can be used to estimate the overcharge as shown in BritNed.

14.19  Another approach which has been used is engineering cost analysis of rigged and unrigged bids. This measures the overcharge as the difference between a bid price for a competitively costed job and the bid price for a rigged tender. This is often called the ‘ratio approach’. It is expected that the winning bid will have a higher price when rigged. If the ratio of winning bid (B) to engineers’ estimate (E) is designated as B/Er for average winning bids and for unrigged bids (B/E)u, then the estimated damages on each rigged bid D is D = [(B/E)r – (B/E)u]Er. Thus if the winning rigged bid is 20% above the engineer’s estimate and average winning bid is 10% below the engineer’s estimate, then the overcharge damages are 30% of the engineering estimate for the contract. Whether or not this is statistically significant can be tested using a t-statistic.

14.20  This approach has several advantages—simplicity, can be shown in graph and bar charts, is straightforward, and easy to verify. On the minus side the approach implicitly assumes that all competitive bids are the same which is unlikely to be the case, and that the overcharge is accurately represented by the B/E ratio whereas other factors may explain a high B/E ratio.

(p. 239) E.  Interchange Fee Approaches

14.21  The European Commission and the courts have adopted cost-based approaches to estimating the exemptible level of interchange fees.20 Although these cases do not involve a secret cartel, they do according to the Commission involve the illegal collective setting of prices. In the Commission decisions and cases two different cost-based methods were used to quantify the overcharge. The first took the costs to issuers of providing the service and taking as the but for costs only those that benefited the merchant. The second approach took the opportunity costs of the providing card services using the Merchant Indifference Test (MIT) which calculates the interchange fee which would make a merchant indifferent between accepting a card and cash payment for a transaction. This is a much more technical approach based on the two-sided nature of card schemes, and on assumptions which may not be accepted by the courts, e.g. see views expressed in Sainsbury’s v MasterCard rejecting the MIT compared with those in Asda21 and by the European Commission.

14.22  The Tribunal in Sainsbury’s v MasterCard based it calculation on the costs MasterCard had used to set its MIFs, adjusted (reduced) to take account of the extent to which merchants benefited from the services generating these costs. This inevitably led to lower bilateral interchange fees (or ‘BIFs’) than the default multilateral interchange fees (or ‘MIFs’). The High Court in Asda rejected this approach because 1) it used costs as a proxy for merchant benefits; 2) failed to address the causation question of whether the merchant benefits were derived from the scheme or the MIFs; and 3) left out costs which were incurred to confer benefits or else these were arbitrarily allocated.22 Further, the High Court said in Asda:

An issuer cost-based approach is not supported in any of the academic literature as having any sound theoretical basis. Rysman & Wright 201523 describe it as ‘not supported by any economic theory’ and Tirole as ‘unfortunately bears little relationship with the theoretically correct level’. It does not capture all costs and costs are not to be equated with value to merchants.24

14.23  The High Court said that ‘the best available approach to quantifying an exempt or exemptible MIF is to use the MIT methodology and the [EC] Commission’s Survey results as the starting point’25 and to adjust these to take account of other relevant merchant benefits. The CAT rejected the MIT as ‘entirely unhelpful for predicting how merchants would behave during bilateral negotiations.’; ‘a conceptual construct that bears little or no relation to what happens in the real world of retail’ ‘far divorced from commercial realities’ and based on an assumption ‘that only needs to be stated to be rejected’.26

(p. 240) 14.24  Several comments can be made on the respective courts’ approaches to quantifying the ‘competitive interchange fee’.

14.25  The Tribunal’s adjusted-cost approach is not necessarily consistent with the bilateral counterfactual. In a bilateral situation the negotiated fee or price will be the outcome of the relative bargaining power of the two parties. It is therefore indeterminate lying between the minimum the issuer would accept and the maximum the acquirer would pay, with the latter derived from the merchants’ willingness to pay. The issue is whether the cost factors used by the Tribunal serve as an adequate proxy for the outcome of bilateral negotiated BIFs.

14.26  The MIT—also known as the ‘tourist test’ and ‘avoidable-cost test’—was developed by economists to identify the efficient—i.e. joint consumer and producer welfare maximizing—interchange fee. The MIF that complies with MIT—what could be called the ‘MIT compliant MIF’ or ‘MIT MIF’—makes a merchant indifferent between accepting a card payment and an alternative form of payment generally assumed to be cash. The European Commission’s Press Release sets out the test and its rationale:

As regards calculation of the (cross-border) MIF, MasterCard has engaged to apply a methodology developed in economic literature to assess efficient interchange fees which is called the ‘avoided-cost test’ or ‘tourist test’. The fee which meets this test, also referred to as the balancing fee, ensures that user benefits are enhanced. The balancing is such that merchants do not pay higher charges than the value of the transactional benefits that card use generates for them. Merchants derive such transactional benefits if card payments reduce their cost relative to cash payments, for instance, because transportation and security expenses for cash are saved or if check-out times at cashier desks are reduced. The implementation of the balancing fee ensures that the merchant is indifferent as to whether card or cash payments are made. To the extent that the fee is passed on to the cardholder, it will ensure that cardholders make efficient choices with respect to payment instruments, being effectively led by the MIF to internalise the cost saving that card usage entails for the merchants. Importantly, this approach prevents the MIF from being set at a level such that banks would take advantage, by collective agreement, of the fact that individual merchants feel compelled to accept a payment card even if it is more expensive than other payment instruments, fearing their customers would otherwise not make purchases at their store (e.g. because other merchants accept the card).27

14.27  The High Court and economists see the MIT as a floor to the efficient interchange fee.28 Recent estimates of MIT compliant interchange fee are close to zero.29 The way the High (p. 241) Court calculated its MIT MIFs resulted in exemptible MIFs exceeding Mastercard’s actual MIFs during the infringement period.

14.28  Several observations can be made on the different approaches of the two courts.

14.29  The first relates to the ‘realism’ of the ‘overcharge’ counterfactuals. The CAT adopted the approach and the costs used by MasterCard to set its MIFs adjusted downwards to take account of the fact that some of these costs did not generate legitimate merchant benefits. It thereby adopted a consistently ‘realistic’ approach to its counterfactuals. The High Court did not. It used an approach and costs which had not been used by either MasterCard or Visa, or as far as the author is aware any card scheme, to set interchange fees. The High Court having adopted a ‘realistic’ counterfactual in dealing with Article 101(1) then used a theoretical method to calculate the exemptible MIF under Article 101(3). The High Court thus found itself rejecting a restriction counterfactual (bilateral negotiations) because it was unlikely to succeed in practice; while positing an exemption counterfactual that had never been or would never have been used in practice. Having said this there is no (legal) reason why a counterfactual cannot be something that has not existed in practice as National Grid held.

14.30  Secondly, the MIF is a regulatory approach or benchmark designed to estimate what economists refer to as the ‘social welfare maximising’ interchange fee. There is no implication that issuers and acquirers would necessarily negotiate a MIT-based interchange fee under the assumed bilateral case or any other realistic counterfactual. Further, since European antitrust does not adopt a social or total welfare approach, the MIT test is biased against consumers, and sets an interchange fee that arguably is too high. Expressed more starkly, the MIT MIF is a regulatory price control which jars with the requirement that it should be a realistic basis for pricing interchange in the counterfactual world and is inconsistent with the consumer welfare goal or objective basis of European antitrust.

14.31  Thirdly, both courts used an aggregative approach, which again was patently ‘non-factual’. They calculated one interchange fee for all retailers for the entire claim period. This despite evidence that credit card interchange fees varied by card type, merchant, and over time; and in the case of the data used by the High Court varied by country. This was purely an administrative convenience. Both courts and the parties saw dealing with the reality of interchange fees as a daunting evidential and computational task which was not possible within the confines of the respective litigations. This led to inconsistencies in the respective judgments. For the CAT this crude estimation approach contrasted sharply with its insistence that in proving merchant pass-on in higher retail prices required demonstrable evidence of higher prices for individual products which was acknowledged as unattainable.30 The High Court’s single MIF contrasted with the reality of myriad MIFs over the alleged infringement period.


1  BritNed Development Ltd v ABB AB and ABB Ltd [2018] EWHC 2616 (Ch) (‘BritNed’).

2  ibid.

3  Karl Aiginger and Michael Pfaffermayr, ‘Looking at the Cost Side of Monopoly’ Journal of Industrial Economics Vol. 45, 1997, pp. 245–67.

4  Clare Macallan, Stephen Millard, and Miles Parker, The Cyclicality of Mark-ups and Profit Margins for the United Kingdom: Some New Evidence, Working Paper No. 351, Bank of England, August 2008.

5  BritNed, para. 448. The documentary evidence is reviewed in ibid., 441.

6  ibid., para. 453.

7  ibid., para. 450.

8  ibid., para. 331, Table 6.

9  ibid., para. 453(9)(d)(iv).

10  ibid., para. 457(e).

11  The average gross margin on the 14 successful ABB cartel submarine projects was an average of 26.7%. However, the 14 projects listed in Table 6 of the Judgment have an average gross margin of 27.4% suggesting an excess margin of 6.3%. If BritNed is included, it is 26.7% as stated in the judgment but this gives 15 not 14 successful projects. This alters the percentage cost saving using the court’s formula in footnote 558 to (5.6%/21) x 15 = 4.0%; 5.6% - 4.0% = 1.6%, or without BritNed 2.1%, not 1.9% as stated in the judgment. Note two features of the court’s arithmetic: (a) the calculation can be simplified by multiplying ABB’s post-cartel failure rate by the gross margin difference; and (b) it is based on the averages of all ABB cartel projects and not the BritNed project only.

12  BritNed v ABB [2019] EWCA Civ 1840, para. 235.

13  BritNed, para. 373.

14  BritNed, para. 373.

15  Sainsbury’s v Mastercard [2016] CAT 11, para. 41. The judge (Marcus Smith J) was also a member of the CAT panel in Sainsbury’s v MasterCard.

16  BritNed, para. 264(1)(e).

17  ‘The material before me inevitably has gaps. These I seek to bridge through careful deployment of the broad brush.’ BritNed, para. 79.

18  BritNed, paras. 467 (8), 17(3), and 17(4).

19  Jeffrey H. Howard and David Kaserman, ‘Proof of Damages in Construction Industry Bid-rigging Cases’ The Antitrust Bulletin, Vol. XXXIV, 1989, pp. 359–93, especially pp. 375–7.

20  Cases COMP/34.579 MasterCard, COMP/36.518 EuroCommerce, and COMP/38.580 Commercial Cards (2009).

21  Asda Stores Ltd v MasterCard [2017] EWHC 93 (Comm). This was a joined action consisting of supermarkets and High Street retailers some of which dropped out as claimants during the course of the proceeding.

22  Asda, para. 354.

23  Marc Rysman and Julian Wright, ‘The Economics of Payment Cards’ Review of Network Economics, Vol. 13, 2014, pp. 303–53. This survey article was ‘supported by a grant from Visa Inc’. But note that this paper is much more equivocal about the optimality of duopolistically determined interchange fees; and the welfare standard which should be adopted. cf. John Vickers, ‘Public Policy and the Invisible Price: Competition Law, Regulation and the Interchange Fee’ Competition Law Journal, Vol. 4, 2005, pp. 5–16.

24  Asda, para. 414.

25  Asda, para. 355.

26  Sainsbury’s v MasterCard, para. 212.

27  Commissioner Kroes notes Mastercard’s decision to cut cross-border Multilateral Interchange Fees (MIFs) and to repeal recent scheme fee increases—frequently asked questions, MEMO/09/143, 1st April 2009. http://europa.eu/rapid/press-release_MEMO-09-143_en.htm.

28  Jean-Charles Rochet and Jean Tirole, Must-Take Cards and the Tourist Test, DNB Working Paper No. 127 January 2007; and Jean-Charles Rochet and Jean Tirole, ‘Must Take Cards: Merchant Discounts and Avoided Costs’ Journal of the European Economic Association, Vol. 9, 2011, pp. 462–95. Jean Tirole, ‘Payment Card Regulation and the Use of Economic Analysis in Antitrust’ Competition Policy International, Vol. 7 (no. 1) 2011, pp. 137–58.

29  Published empirical research computing MIT MIFs have found them to be zero or significantly negative. Søren Korsgaard, ‘Paying for Payments Free Payments and Optimal Interchange Fees’, European Central Bank, Working Paper Series No. 1682, June 2014; European Commission, Survey on Merchants’ Costs of Processing Cash and Card Payments—Final Results, March 2015. Sainsbury’s gave evidence based on its accounting data that the MIT MIF was very close to zero at 0.04%.

30  See Cento Veljanovski, ‘The Law and Economics of Pass-on in Price Fixing Cases’ European Competition Law Review, Vol. 38, 2017, pp. 209–18.