Trial 2A of the Umbrella Interchange Fee Proceedings: Lessons from quantifying merchant pass-on

In February 2026, the Competition Appeal Tribunal (CAT) handed down its Trial 2 judgment in the Umbrella Interchange Fee Proceedings, addressing how unlawful multilateral interchange fees (MIFs) were passed on through the supply chain.

In this article, Florian Mockel examines what the ruling means for the quantification of pass-on in large and complex competition damages claims. Drawing lessons on the use of proxy costs to assess very small overcharges, the extrapolation of results across heterogeneous claimant groups, and the practical limits of supplier pass-on as a defence.

Florian Mockel is a Vice President at Compass Lexecon and was part of the Compass Lexecon team instructed by the Umbrella Interchange Fee Claimants in the Merchant Interchange Fee Umbrella Proceedings. He thanks Stefano Trento for his valuable insights and feedback. The views expressed in this article are the views of the author only and do not necessarily represent the views of Compass Lexecon, its management, its subsidiaries, its affiliates, its employees, or its clients.

Introduction

The Umbrella Interchange Fee Proceedings cover claims brought before the Competition Appeal Tribunal (‘CAT’) by more than 2,000 merchant claimants against Visa and Mastercard (‘the defendants’, together with the merchants, ‘the Parties’), who argued that the multilateral interchange fees (‘MIFs’) set by Visa and Mastercard were unlawful.

Having found in a first trial that the MIFs did indeed infringe Article 101(1) TFEU, the CAT next had to consider whether, and if so, to what extent, the merchant claimants had passed on some or all of the alleged overcharge to their customers. Its recently published ‘Trial 2 judgment’[1] addresses this.

Unlike in previous judgments like Royal Mail or Kent, where pass-on was assessed either only for a small number of firms or treated as somewhat more ancillary to overarching questions of liability and overcharge, it was front and centre in this judgment.

Against this background, the Trial 2 judgment answers several questions that will be critical for any proceedings involving pass-on:

• How to assess pass-on of a very small cost by using proxy costs?
• How to assess pass-on for a large and highly diverse class of claimants?
• Does supplier pass-on offer defendants caught between direct and indirect purchasers a pass-on defence that actually works?

In this article, I first provide some background on the case and then discuss each of these questions from an economic[2] perspective and the implications of the CAT’s judgment for future litigation. The key takeaway for practitioners is that the CAT has yet again confirmed that economic expert evidence helps to inform the Tribunal but only if it is grounded on factual evidence.

Background

Payment flows in Visa's and Mastercard's payment schemes

Visa and Mastercard operate payment schemes generally referred to as ‘open’ or ‘four-party’ payment systems. These are distinct from ‘closed’ or ‘three-party’ systems, such as Amex’s system.

The four parties in a four-party scheme (five, if one includes the scheme operator) are those necessarily involved in the payment flows arising from a transaction. They are:

• the cardholder, who pays for a purchase using his or her payment card;
• the merchant, who sells a good or service to the cardholder;
• the issuer, who issued the cardholder’s card; and
• the acquirer, who provides ‘acquiring’ services to the merchant – i.e., services that allow the merchant to accept payments made with Visa or Mastercard cards.

The scheme owner develops the card brand and sets the various rules and regulations by which members of the schemes (issuers and acquirers) must abide.

A four-party system may be illustrated schematically as follows:

Figure 1: Schematic illustration of payment flows in Visa's and Mastercard's payment systems

^{Source: Payment Systems Regulator, “Market review of UK-EEA consumer cross-border interchange fees“, December 2024, Figure 1.}

When a cardholder makes a nominal payment to a merchant using a payment card, three payments are effectively being made before any money reaches the merchant:

• the cardholder gives the issuer the payment sum in full;[3]
• the issuer gives the acquirer the payment sum less the interchange fee (‘IF’);[4] and
• the acquirer gives the merchant the payment sum less a merchant service charge (‘MSC’).[5]

Case history

The CAT had tried the question of whether the MIFs did indeed violate Article 101(1) TFEU in early 2024 and found that they did.[6][7]

Trial 2 addressed two questions:

• To what extent did the merchant claimants pass on the unlawful MIFs to their customers (‘Trial 2A’, tried in late 2024)?
• To what extent did the acquirers pass on the unlawful MIFs to the merchant claimants (‘Trial 2B’, tried in March 2025)?

Figure 2: Issues covered in Trials 2 A and 2B

^{Source: Compass Lexecon.}

Ultimately, there were thus three types of pass-on pleaded by various parties:

• Acquirer pass-on, from acquirers to merchant claimants,
• Supplier pass-on, from merchant claimants to their suppliers, and
• Merchant pass-on, from merchant claimants to their customers.

The Trial 2 judgment addresses all three. However, the remainder of this article deals with merchant pass-on and supplier pass-on only.

How to assess merchant pass-on of a very small cost by using proxy costs?

Typically, MIFs account for a small share of merchant claimants’ costs; for instance, they amount to 0.2% in the case of consumer debit cards. A “small” overcharge – in the sense that it is minor relative to other factors that affect a firm’s prices – can present challenges in a pass-on case, as it makes it difficult to empirically isolate the impact the overcharge has on prices.[8] In general, it is not uncommon for an overcharge affecting any single product to be small relative to a merchant’s total cost.[9]

Therefore, all experts agreed that, instead of directly analysing the pass-on of the MIFs, they would quantify the pass-on of proxy costs. However, while the experts agreed on using a proxy cost in principle, there were different opinions on how to apply the approach in practice.

Here, I focus on how to determine which costs constitute an appropriate proxy.[10]

The experts broadly agreed that there are two conditions a proxy cost should meet:

• The proxy cost must be large enough to allow for empirical estimation of its pass-on rate.
• The proxy cost must be passed on at (or as close to) the same rate as the MIFs.

Meeting this first condition can have its challenges, but selecting a large-enough candidate proxy is comparatively straightforward.

Assessing the second condition is much more difficult and was critical in this case. In Trial 2, the debate chiefly focussed on whether it was more important that the economic nature of the proxy cost and the MIFs should match, or the factual channels through which each affected prices.

Identifying an appropriate proxy cost by its economic nature

At face value, MIFs are variable (mostly) ad valorem costs. So, under this approach, the proxy cost should have the same nature: a variable cost, and ideally an ad-valorem one. In practice, due to data availability, it was argued that one should look at a merchant’s total cost of goods sold (‘COGS’)[11] or total costs as an appropriate proxy cost for the MIFs.

The rationale put forward was that the goal was to establish pass-on across a group of firms and that a suitable proxy cost should therefore share the same economic features as the MIFs.

Identifying an appropriate proxy by the channels through which it factually affects prices

On the other hand was the argument that one should instead consider the channels through which the sample of claimants that provided factual evidence considered MIFs when setting prices. Their evidence suggested that, in practice, many of them had treated MSCs as part of their overhead costs.[12] So, under this approach, the proxy cost should affect prices (if at all) through the same channel as the MIFs. In practice, and due to data availability, this meant that total overheads were considered as an imperfect yet best available proxy cost for the MIFs for these merchants.

That claimants treated MSCs as overhead costs mattered:

• Firstly, it suggests COGS would not be an appropriate proxy: such overhead costs had, at most, a more indirect effect on prices than COGS, and – even if they affected prices – they would do so via different mechanisms than COGS.
• Secondly, if a merchant treated the MIFs as part of a bucket of overhead costs in its pricing, then any proxy cost used to measure pass-on of the MIFs should also be an overhead cost: to ensure the proxy would affect prices via the same mechanism.[13]

The CAT's view

The CAT sided with the second argument and emphasised the importance of “trying to find a proxy that was treated similarly to the way that the Claimants treated the MSC in fact and would be passed on, if at all, through a similar mechanism.”[14] Although it pointed out that there were several issues with using total overheads as a proxy cost, it found that this was the best available proxy cost.[15]

It acknowledged that, for various reasons, using COGS would allow for a more precise estimate of the pass-on rate than using total overheads. I use “precision” here to refer to the degree of certainty around the estimated pass-on rate. For example:

• Assume the true pass-on rate is 27%.
• With COGS, one might estimate that, with 95% probability, the pass-on rate would be found in a very narrow corridor of between 75% and 80%, with the most likely estimate being 77.5%. That estimate is precise, but wrong.
• In contrast, for total overheads, a similar 95% confidence interval might span a range between 10% and 40%, with a central estimate of 25%. That estimate is less precise, but better reflects the true value. The estimate for overheads would thus be considered to be much less precise than the estimate for COGS.

Recognising that, the CAT emphasises that:[16] “The fact that a more accurate result can be obtained by using COGS does not justify using it if it is the wrong proxy”.

The key takeaway here is thus that when choosing an appropriate proxy cost, economic theory matters but it does not trump factual evidence. Where such evidence suggests that a firm undertook certain actions or treated certain aspects of its business in a way that seemingly departs from conventional economic theory, this does not mean that its behaviour can be dismissed out of hand. Instead, expert economists need to recognise and carefully consider the factual evidence before them.[17]

Sectorisation: extrapolating from analysed claimants to everyone else

Given the scale of the claimant pool, with over 2,000 firms, all experts agreed that it was not feasible to conduct individual analyses for each claimant.

Initially, the Parties discussed sampling a subset of claimants to analyse but the CAT eventually rejected this approach because it considered there was not enough time to implement it.

Instead, all experts grouped the claimants into different sectors such that all merchants in that sector would be assumed to have passed on the MIFs at the same rate.

Such aggregation inevitably generates inaccuracy because even otherwise very similar merchants may have passed on the MIFs at different rates. Indeed, there is a trade-off between accuracy (the similarity of merchants within each sector) and workability (the number of sectors with an estimate):

A short list of sectors prioritises workability, as it may allow experts to directly estimate a pass-on rate for each sector. This reduces the amount of work needed overall as no consideration needs to be given to matching sectors for which an analysis is conducted to sectors that are not directly analysed. However, it compromises accuracy, as there is greater scope for dissimilar merchants to be grouped together in the same sector.

Alternatively, having granular sectors prioritises accuracy, as there is a greater likelihood that all merchants in a sector will have at least similar pass-on rates and thus makes it likelier that any pass-on rate derived for a given sector is fairly accurate for all merchants in the sector.[18] However, experts may only be able to directly estimate pass-on for a subset of sectors. So, this approach can be challenging to work in practice, as it requires experts to extrapolate the results estimated for some sectors to those sectors for which no analysis is conducted.

Ultimately, the experts relied on two different approaches:

• An aggregated approach: Using Visa’s internal merchant classifications.[19]
• A granular approach: Using Standard Industrial Classification (SIC) codes to identify firms in similar industries and refining this list based on similarities in price setting behaviour by firms in different sectors.

I I briefly describe each of these approaches and then explain how the CAT’s view on this question is likely to shape other cases going forward.

Using an aggregated approach: the defendant's sector classifications

The key advantage of using the defendant’s existing sector classifications is that these were readily available. Created in the normal course of business, these grouped merchants into just over 10 categories.[20]

The small number of sectors considered means that there was at least one merchant from each sector who had disclosed data on their revenues and costs to allow for an analysis of pass-on of that merchant. Accordingly, little or no extrapolation across sectors was needed, i.e., any uncertainty around whether a given pass-on estimate is appropriate for a given merchant comes from the fact that different merchants in the same sector may have different pass-on rates.

However, it remained unclear whether merchant classifications that Visa and Mastercard used in their ordinary course of business would say anything about whether the firms in these categories were likely passing on the MIFs at similar rates.

The high degree of heterogeneity amongst merchants within some of these sectors therefore meant that even though this approach does not require any extrapolation across sectors, the within-sector inaccuracy might be very high, i.e., merchants in the same sector may actually have vastly different pass-on rates.

Relying on a granular approach: SIC codes

The alternative approach of grouping merchants based on SIC codes[21] resulted in close to one hundred different sectors.

The main advantage of this approach was that using narrower sectors based on claimants’ actual industries would make it more likely that claimants within a given sector would have had similar pass-on rates. Furthermore, as the SIC codes of all merchant claimants were known, the creation of these sectors did not involve substantially more work than using the defendants’ classification.

The downside of this approach was that the merchant claimants who had disclosed data for analysis only covered around ten sectors and accordingly, pass-on rates could only be directly estimated for these ten sectors.

To address this gap, a pricing expert was tasked with determining, on the basis of different firms’ pricing strategies, which of the sectors, for which pass-on rates were estimated, could serve as benchmarks for the other sectors. For example, the pricing expert might find that airlines and hotels pursue similar pricing models, in that both tend to procure capacity (i.e., offer a certain number of rooms or seats on a plane) and then try to fill that capacity in a revenue-maximising way. Accordingly, the pricing expert would suggest that the pass-on estimate for hotels would be the best proxy for airlines.

The CAT's view

The CAT has not made a final ruling on sectorisation, noting that there may still be certain claimants who would want to avail themselves of an “exceptions process”. However, it explained that it considered the approach of thinking about similarity of pass-on rates to be “credible and appropriate”.[22]

Accordingly, we expect that in future proceedings, parties who will want to extrapolate pass-on rates from one claimant or group of claimants to another, will not be able to point to claimant categorisations that were performed for entirely different purposes. Instead, they will need to come up with their own claimant groupings, based on a reasoned approach.

Supplier pass-on

Supplier pass-on refers to the possibility that merchants might offset higher MIFs by negotiating lower prices from their other suppliers. The argument is that changes in one cost component somehow affect other costs, as well.

This is a relatively novel theory that had been mentioned but not tested in litigation previously.[23] For a merchant claimant to engage in supplier pass-on, there would need to be some sort of connection between the cost on which they incur an overcharge (here: the MIFs), and their other costs.

Although untested, supplier pass-on is not necessarily inconsistent with economic theory. For example:

• A firm might not be paying the lowest price possible for its inputs because cost monitoring and negotiations are themselves costly activities. The firm may thus reach a point where further minimisation of the input’s price is possible but only at a cost greater than the possible reduction in the inputs’ price. At this point, the firm’s costs are minimised in general, but input prices may not be at their lowest possible level.
• In this situation, supplier pass-on may theoretically happen if there is a link between two costs A and B such that a cost shock to cost A somehow changes the cost-benefit analysis for cost B and thus prompts the firm to (i) carry out a (costly) off-cycle cost review or an off-cycle renegotiation, or (ii) negotiate more fiercely in the course of on-cycle interactions with its suppliers.[24]

However, supplier pass-on is difficult to demonstrate empirically. Firms often have hundreds of different suppliers, and numerous factors, such as demand, production costs, or product characteristics affect the prices they pay each of their suppliers. Establishing a causal link between MIFs and reductions of the prices paid to other suppliers would therefore require controlling for all these factors.

The CAT expressed scepticism about supplier pass-on. While leaving open the possibility that supplier pass-on may be proven in a different case, it also pointed out that there was simply no factual evidence it played any part in the present case. It further highlighted that “Mr Beal KC was right to submit that a likely reason for there being no evidence of supplier pass-on is that it simply does not occur.”[25]

This suggests that anyone trying to plead supplier pass-on will continue to face a steep uphill battle when it comes to actually demonstrating it empirically and may want to consider whether the juice is worth the squeeze.

Conclusion

I set out three key questions that the Tribunal had to answer in the Trial 2 judgment:

• How to assess pass-on of a very small cost by using proxy costs?
• How to assess pass-on for a large and highly diverse class of claimants?
• Does supplier pass-on offer defendants caught between direct and indirect purchasers a pass-on defence that actually works?

The factual answers given in the Trial 2 judgment are of course applicable to that judgment only. However, the key takeaway for other proceedings is that anyone addressing these questions should take a flexible approach that reflects the factual evidence available to them and transparently addresses its own shortcomings:

• When assessing pass-on of a very small cost by using proxy costs, it is important to consider not just the economic nature of the cost but also the factual reality of how it directly influences prices. Economic theory provides guidance in considering what characteristics of a cost are generally relevant but factual evidence is the key to getting the choice right.
• When assessing pass-on for a large and highly diverse class of merchants, it is acceptable to rely on analyses conducted on the basis of a small number of sample merchants and extrapolate the results to other merchants. However, when extrapolating results to other merchants, one should group together merchants that are expected to have similar pass-on rates for the cost in question, instead of relying on arbitrary categorisations.
• Supplier pass-on may theoretically offer defendants caught between direct and indirect purchasers an alternative pass-on defence but in reality the burden of proof seems very difficult to meet.

---

References