Economist’s perspective: recent CAT judgments yield lessons on the collection and submission of economic evidence

In summary

The UK’s Competition and Appeal Tribunal (CAT) recently issued judgments in three important competition matters: Compare the Market, Royal Mail, and Gormsen. This article discusses the key lessons from these judgments as they relate to economic evidence in competition damages matters. Several interesting themes have emerged from these judgments on the following topics: the market definition for two-sided platforms, the bar for establishing ‘by effect’ allegations, econometric analyses in damages estimations, proving overcharge mitigation and class certification.

Discussion points

• Market definition in two-sided platforms and the extent to which standard (SSNIP) tests do capture the substitutability of products for a given group of customers might yet fail to properly define competitive constraints, especially when one side has a zero price
• Challenges arising from deploying econometric analyses to estimate overcharge and pass-on
• Challenges getting the class action certified and ensuring that the right methodology is proposed for estimating damages for the class

Referenced in this article

• Compare the Market CAT Judgment
• Royal Mail and BT v DAF and Other CAT Judgment
• Dr Liza Lovdahl Gormsen v Meta Platforms, Inc and Others CAT Judgment

Compare the Market – CAT Judgment^[1]

In November 2020, the UK Competition and Markets Authority (CMA) fined Compare the Market (CTM),^[2] a home insurance price comparison service, approximately £18 million for using wide most favoured nation clauses (wMFNs) in its agreements with insurance providers between 1 December 2015 and 1 December 2017.^[3] The CMA found that these clauses constituted an infringement by effect of article 101 of the Treaty on the Functioning of the European Union (TFEU) because they reduced price competition between price comparison websites (PCWs) and between home insurers competing on PCWs.^[4]

The CMA defined the relevant market as PCW services for home insurance in the UK and characterised it as a single two-sided market comprising customer introduction services to home insurance providers and price comparison services to customers.^[5] The CMA reached this conclusion after considering the standard small but significant non-transitory increase in price (SSNIP) test of what substitution would emerge if a hypothetical monopolist of PCWs imposed a small price increase of 5–10 per cent on the commission rate paid by insurers. The CMA justified its approach of only applying the SSNIP to the insurer’s side by noting that both customers and home insurance providers face the same set of potential alternatives. By implication, the CMA assumed that there is no difference in the analysis of substitutability for either consumers or home insurers.^[6]

The CMA also noted that a 5–10 per cent increase in commission charges would translate to a very small cost increase for insurers and an even smaller increase in premiums for consumers (ie, 1.8 to 3.5 per cent), which would be too insignificant to incentivise them to switch to other channels.

CTM challenged the CMA’s decision, and on 8 August 2022, the CAT overturned the CMA’s decision primarily on the grounds that: (1) the CMA’s approach to market definition was ‘materially wrong’;^[7] and (2) the CMA presented ‘no reliable evidence to conclude that the existence of the wMFNs in the wMFN Agreements had any adverse effect on either Premiums or Commissions’.^[8]

While the CAT’s judgment has several important implications, three are of particular interest.

First, the failure to account for the matching services provided by a platform (ie, enabling transactions between distinct groups) risks disregarding the competitive pressure imposed by competing channels and a flawed market definition.

The CMA characterised PCWs as supplying ‘price comparison services’ to consumers, and ‘customer introduction services’ to insurance providers.^[9] However, the CAT found this characterisation to be incomplete. The CAT noted that, in addition to offering price comparison services, PCWs also ‘provide an efficient and convenient means for the consumer to conclude a contract of insurance with a chosen home insurance provider if they wish to do so, and a facility (assuming the consumer likes the outcome of the comparison) to buy’.^[10] In the CAT’s view, by disregarding this service provided by CTM, the CMA failed to consider the likely competitive pressure imposed by other competing direct channels (such as brokers) and other PCWs.^[11]

The CAT concluded that the CMA’s approach to market definition suffered from an inherent circularity as it was unduly influenced by the theory of harm that the CMA was investigating. Here, the CAT noted that if ‘one approaches market definition with a presumption that a lack of competition as regards Commissions arising out of wMFNs will result in higher Premiums, then the market will be defined in a manner that fails to test for that which is presumed.’^[12]

Second, when defining the market in the context of a multi-sided platform, it is imperative to consider competitive constraints on all sides of the platform.

The CAT flatly rejected the CMA’s position that the same set of competitive constraints is at work on both sides of a PCW. Here the CAT observed that:

whereas a consumer, a purchaser of home insurance, will be very well able to decide whether to renew or purchase through a direct channel or purchase through a price comparison website, a home insurance provider will have a far more constrained choice . . . . In other words, the issues of demand side substitutability are completely different, and must be tested for separately.^[13]

The CAT also rejected the idea that competitive constraints on the consumer side can be properly accounted for using the indirect SSNIP test only on home insurance products sold on PCWs.^[14] Both the CMA and CTM assessed substitution on the consumer side by assuming that a price increase of a hypothetical monopolist on the insurer’s side would be passed on to consumers.^[15] The CAT found such a test pointless and rejected this on the grounds that it only captures substitutability of consumers who actually end up buying an insurance policy and not for those who only browse the PCWs.^[16]

Third, a by-effect infringement needs to be evidenced by empirical analysis rather than as a likelihood based only on theory and qualitative evidence.

The CAT found that the CMA’s analyses of effects did not meet the burden of proof required to establish the anticompetitive effects of the wMFN clauses. According to the CAT, the CMA’s analyses were purely based on theories, assertions and qualitative evidence with no significant quantitative evidence.

A finding of a by effect infringement cannot be sustained simply on assertions based on economic theory and qualitative evidence. When feasible, it must be supported by quantitative evidence. According to the CMA, CTM operated a network of wMFNs with 32 distinct home insurance providers and the cumulative effect of these clauses was harming competition.^[17] The CMA arrived at this finding primarily based on qualitative evidence obtained from a small number of home insurance providers.^[18] The CAT found that a ‘great deal of the analysis [in the CMA’s decision] operates at the level of theory or (less helpfully) bare assertion’ and that its qualitative evidence lacked depth (ie, ‘bald explanations, without detail, are the order of the day’) and ‘was inconsistent with the CMA’s theory of harm’.^[19]

Additionally, a finding that the conduct in question (in this case wMFN) had some effect on market outcomes says nothing about whether the effect was anticompetitive.^[20] The CAT reiterated this finding in the context of excessive pricing in Gormsen (discussed below).

Royal Mail and BT v DAF and Others^[21]

For only the second time in the UK courts, the CAT reached a judgment in a cartel damages matter. This case follows on from the European Commission’s (EC) 2016 decision imposing a record-breaking fine (at that time) of €2.93 billion on five truck manufacturers (DAF, MAN, Daimler, Iveco and Volvo/Renault) for participating in a 14-year cartel. The cartel had involved the exchange of gross list prices for medium and heavy trucks and the costs of converting them to conform with more stringent EU emissions regulations.

Since then, there have been several follow-on damages claims against the manufacturers brought in the UK courts and a high volume of claims brought in the EU. Two separate claims brought by Royal Mail and British Telecom (BT) went to trial together last summer. In each case, the claimants alleged an overcharge of around 10 per cent for the trucks purchased during the cartel. They also claimed that they were unable to mitigate this loss. The defendants’ expert, however, found no evidence of overcharge and opined that any overcharge would have anyway been mitigated.

The CAT decided that damages should be awarded based on a 5 per cent overcharge with no discount for mitigation. Damages for both Royal Mail and BT amounted to some £17.5 million before adjustments for interest and tax.

The CAT’s 300-page judgment provides a detailed illustration of the issues the Tribunal had to grapple with, including plausibility, overcharge, pass-on, interest on damages and taxes. Discussed below are three important takeaways regarding plausibility of the cartel, estimation of the overcharge and the defendants’ mitigation defences.

Plausibility

Although the bar for establishing plausibility is low in a follow-on damages case, it is important that proposed theories are realistic and based on the facts of the case.

Both claimant and defendant experts submitted reports on whether, based on the factual evidence available, it was plausible the cartel could have had an effect. The claimant expert argued the plausibility threshold was easily met and cited the EC-commissioned study on quantifying damages that includes research on the typical impacts of cartels. By contrast, the defendant expert challenged whether, in this instance, coordination on transaction prices was feasible given the bulk of the information exchange was on list prices and that he had found no actual link between list and transaction prices.

The Tribunal, however, found it plausible that the information exchange inflated transaction prices in light of:

• testimony from a defendant factual witness that there was a material relationship between list and transaction prices;
• some evidence of the illegal exchange of transaction prices; and
• the fact that the cartel lasted 14 years, suggesting that the defendants got something out of the information exchange.

Ultimately, the Tribunal concluded that the defendant expert’s position on plausibility was in itself not plausible. Indeed, his position included an assertion that information exchange on list prices might even be pro-competitive – a position that he later rowed back from. Moreover, the expert had failed to interrogate the defendants as to why they had shared the information or how they had actually used the data. Furthermore, the expert’s position on plausibility may have also impacted the Tribunal’s consideration of how objectively the expert approached the econometric evidence, as more than once, the Tribunal mentioned the risk of confirmation bias.

Estimation of the overcharge

As the experts have a duty of independence to assist the court, the CAT will closely consider indicators of objectivity, such as how preferred models are selected and the reasonableness of the experts in recognising the merits of alternative approaches.

The overcharge analysis was dominated by complex econometric analysis to establish the impact of the cartel on truck prices and estimate the level of overcharge faced by the claimants. The Tribunal engaged fully with the detailed technical evidence and, in doing so, reached a view on how reasonable the experts’ positions were and, ultimately, how helpful their expert opinions could be.

Both claimant and defendant experts adopted an approach that compared cartel prices with those that prevailed outside the cartel period. A well-known challenge that arises is controlling for confounding factors (such as demand shocks) that also influence truck prices over some or all of these periods. Most of the technical debate centred on three key factors that affected truck prices during the 14-year cartel: the exchange rates;^[22] the 2008 financial crisis; and the costs of complying with more stringent emissions standards.

Regarding exchange rates, the experts adopted very different approaches, and the overcharge estimates were very sensitive to this choice.^[23] Indeed, the 10 per cent overcharge estimated by the claimant expert largely evaporated when switching to the defendant expert’s approach. The Tribunal concluded that there were significant challenges with both approaches and no single right answer, expressing disappointment that the experts did not countenance other (equally) plausible alternatives:

The tendency of both experts to defend their positions without acknowledging the inherent difficulties in their own approach was disappointing and inconsistent with their primary duty to assist the Tribunal.^[24]

Similarly, when controlling for the collapse in demand following the 2008 financial crisis, the claimant expert introduced additional controls over and above demand to account for the unprecedented nature of the financial crisis. The Tribunal agreed that there may be merit in that explanation but took exception to the claimant expert only doing so after the initial approach resulted in a much lower overcharge (1.5 per cent) – with this fact only emerging late on in the process. In the CAT’s view, this ‘did appear to have had the effect of shifting the goalposts ex post after his original model using the standard demand controls reached an inconvenient result.’ ^[25]

The experts also disagreed on the issue of increased costs of complying with more stringent emissions standards over time. The difficulty here was that this coincided with some manufacturers introducing added features to boost demand to compensate for customer reluctance to accept the costs of higher emissions standards. The claimant expert asserted that higher margins on vehicles that complied with new emission regulations resulted from the infringement, while the defendant expert argued that higher margins were due to improved overall product quality.

Ultimately, whils the CAT acknowledged that the experts’ detailed analysis enabled the Tribunal to make a better-informed judgment, the Tribunal concluded that it could not be assisted by the opinions of the experts as these were seen to lack independence:

[W]e cannot rely on either expert to have provided a fully reliable and unbiased stand-alone estimate of the effect. Both reached conclusions that, whilst they fell within the range of robustly arguable positions, were clearly influenced in favour of the commercial interests of their respective clients.^[26]

In essence, having engaged deeply with the analysis to satisfy itself that the respective positions were at least arguable, given its concerns about independence related to both experts, the Tribunal was unable to attach greater weight to one opinion. Accordingly, the Tribunal then applied the ‘broad axe’ to the opposing estimates to reach a finding of 5 per cent overcharge, based on ‘the evidence that was presented in the experts’ models, and on a wider appreciation of the factual context and witness evidence’.^[27] The Tribunal chose not to ‘score’ the experts point by point on the key swing factors to reach this conclusion as this would be spuriously accurate given clear solutions to some of the issues were not obvious.

There are many takeaways from the Tribunal’s treatment of the expert evidence on overcharge. What is clear is that the Tribunal will engage readily with all the detailed and technical issues. It will also look closely at indicators of independence, in particular the overall reasonableness of the expert’s final outcome, how the expert acknowledges the strength of their inferences in light of competing alternative methods, and how precisely the expert arrived at the findings on which they based their conclusion. All these elements of the CAT’s approach should not be surprising. Regarding the last point on process, it should be acknowledged that experts do not have perfect foresight, and as they get exposed to evidence and the facts of the case, unforeseen issues may arise that would require readjustments of proposed methodologies.

The defendants’ mitigation defences

The bar for successfully arguing mitigation of overcharges remains high and, while there is an expectation that experts will deploy sophisticated analyses, this needs to consider the proportionality of the analyses given the size of the effect being evaluated.

The defendant expert advanced three theories of mitigation, all of which the CAT dismissed. First, the defendants argued that the overcharge would need to be offset because claimants would have likely paid less for complementary products, such as bodies and trailers purchased from non-defendants. The theory here was that higher truck prices would depress not only the demand for trucks but also the demand for complementary products whose prices would then fall, thereby offsetting some or all of the overcharge. As is often the case with evidencing shifts in demand curves, the evidentiary burden was high and was not met. This point was, therefore, dismissed.

Second, the defendants argued that higher prices for new trucks would have also raised the price of that truck when it came to being sold as a used truck. On the facts, the Tribunal found that the time lag between purchase and resale – up to 12 years on average – was too great to have any confidence in a concrete link between the two. Given also that any effect would in any event be minimal (with used prices being around a 10th of new prices), the Tribunal rejected econometric evidence of this link.

Here, the Tribunal clearly expressed its comfort in dealing with complex issues such as endogeneity in the defendant expert’s methodology. Endogeneity arises if the same common factors affect both new and used truck prices, making it very hard to disentangle the effect of the used truck prices on new truck prices. The Tribunal expected that both experts would have engaged more with this issue and further examined the suitability of the ‘instrumental variable’ approach considered in the defendant expert’s methodology.^[28]

The third theory the defendants advanced was that the claimants passed on the overcharge to their retail prices, such as the price of stamps or the price of fixed landlines. The defendants noted that both claimants’ retail prices were subject to regulatory price controls that would allow for, among other things, cost recovery. Claimants, however, argued that truck costs were a miniscule proportion of their total costs and that regulatory mechanisms were not fine-tuned enough to deal with small changes that the overcharge would represent.

The Tribunal majority agreed, deciding that these costs – representing between a 1,000th and a 100,000th of the claimant’s revenues – were simply too small and too remote for any overcharge to be a direct and proximate cause of any increase in downstream prices. Also, there were too many steps, and too many examples of regulatory judgment that meant the pass-on of small cost changes could not be deemed mechanistic or automatic. The defendants’ position was also not helped by the absence of discrete line items for trucks in the regulatory accounts and no audit trail back to the specific truck invoices in hand.

Notably, the economist on the panel disagreed. In his minority view, no matter how small the cost, the regulatory judgments could be taken as a given, and slightly higher costs would have led to slightly higher prices.

The precedent value of this for the more usual situations outside of regulatory price controls needs to be clarified. In particular, where pricing decisions are not based on a regulatory process, it is possible that even a small change in costs can still result in a positive pass-on in downstream prices. For example, where the direct evidence shows that prices are set based on the customer lifetime value of subscribers, then any reductions in gross margin associated with even a very small overcharge will likely translate into higher prices. This is in keeping with the fact that less profitable customers are competed for less aggressively.

Gormsen v Meta^[29]

In this opt-out competition class action against the Meta group, the proposed class representative (PCR), Liza Lovdahl Gormsen, alleged that Meta, which owns and operates Facebook, abused its dominant position causing harm to users (members of the proposed class) of Facebook. Specifically, the PCR seeks damages citing three distinct alleged abuses by Meta.

• Unfair data requirement: by making ‘users’ access to Facebook contingent on the provision of personal data, Facebook abused its dominant position in order to advance its own commercial interests and objectives in the market(s)’.
• Unfair price: Meta imposed an unfair price on users (or made them an excessively low payment) because it took their valuable personal data without paying for them, offering only social networking services in return. Moreover, prices were ‘unfair because they enabled it [by undercompensating] to reap . . . supra-competitive profits, which it could not have obtained in conditions of workable competition’.
• Other unfair trading conditions: Facebook imposed terms and conditions (T&Cs) on users that were unfair and anticompetitive because, for example, ‘(i) users wishing to use Facebook’s platform were unable to avoid signing up to the T&Cs, (ii) the T&Cs were excessively long and/or complex, and (iii) Facebook failed to explain, adequately or at all, the nature, extent and/or scope of the personal data it collected and/or the way it utilised the personal data’.

The proposed class comprised an estimated 45 million persons in the UK who had a Facebook account and who accessed that account at least once during 14 February 2016 to 31 December 2019.^[30]

The PCR’s damages expert proposed calculating aggregate damages – due to all three distinct alleged abuses – as the commercial value of users’ personal data in a competitive market, less ‘the market price that users would have been charged by Facebook for delivering the social networking platform in a competitive market’.^[31] The expert:

suggested that the price that would be agreed in a ‘competitive’ market would lie somewhere between the maximum that Meta could afford to pay its Users (represented by [a profitability metric for Meta such as] ROCE [Return On Capital Employed] over WACC [Weighted Average Cost of Capital]) and the minimum that Users would accept (being a rate above their value of the data they were providing to Meta) [emphasis in original].^[32]

The expert added ‘that there are a variety of possible competitive counterfactuals, including ones in which users are paid for the data that they provide to the social network and Facebook continues to monetise the data through selling advertising’,^[33] and that any loss calculated using the proposed framework would be applicable to any potential competitive counterfactual.^[34]

In its judgment, the CAT held that the PCR had ‘unequivocally failed’ to meet the Pro-Sys Consultants v Microsoft test (the Pro-Sys test)^[35] and stayed the CPO application for six months. This was to give the PCR time ‘to file additional evidence setting out a new and better blueprint’, but if that was not forthcoming, the CAT would lift the stay and reject the CPO application.^[36]

Several important lessons, in respect of economic evidence, emerge from the CAT’s judgment.

First, a damages methodology must be consistent with the alleged abuse. While the PCR’s expert proposed a single damages methodology for calculating damages due to all three (distinct) alleged abuses, the CAT found the approach to be deficient because ‘the counterfactual in the case of each of the three causes of action pleaded . . . varies significantly from claim to claim’.^[37] As the CAT explained:

there is a nexus between (i) the exact breach of duty alleged, (ii) the framing of the counterfactual needed to put the claimant class in the position they would have been in had the tort not been committed, and (iii) the method of quantifying the damage sustained as a result.^[38]

Thus, the CAT reiterated that damages should be calculated relative to the specific counterfactual that is likely to prevail absent the alleged abuse. Ultimately, the Tribunal agreed with Meta’s criticism that the expert’s methodology could only be relevant for the alleged unfair price abuse and that no damages methodology had been advanced for the two other alleged abuses.^[39]

Second, to establish abuse, the CAT is clear, at least regarding excessive pricing allegations, that the question is not what prices would prevail in some unspecified competitive counterfactual. Instead, the relevant counterfactual is what prices would prevail absent excessive prices (which could be greater than competitive prices). This ‘involves not working out what price would pertain if the market were differently structured, but working out whether Meta’s price is too high’.^[40] Accordingly, the CAT found that it is inappropriate to rely on comparators ‘to infer the price that users would receive in the competitive counterfactual’ because ‘[t]hat is not the relevant counterfactual. The relevant counterfactual . . . is what price would Users pay if the abuse (the excess price) was removed’.^[41]

The CAT seems to be acknowledging the point that prices set by a dominant company – ie, prevailing in a market that is a departure from normal and effective competition as (in the words of Hoffman La Roche) the degree of competition has already been weakened by the very presence of the dominant company – cannot automatically be deemed excessive as that would make dominance itself automatically abusive. In other words, the United Brands test for excessive pricing does not stop with the question of whether a dominant company is able to ‘reap trading benefits which it would not have reaped if there had been normal and sufficiently effective competition’.^[42] The test also requires careful consideration of other aspects, including the company’s margins and the economic value of its products.^[43] Indeed, the CAT recognises that, ‘even in a competitive market, prices may significantly be in excess of cost, and not just in the short-run’.^[44] The use of the word ‘even’ here implies that the CAT considers that this is not the right counterfactual in any event. Again, the existence of a dominant position will already have weakened competition (ie, static competition), so this cannot be the benchmark for assessing excessive pricing.

Third, the damages methodology should not attribute damages that one side of the platform may have suffered to another side, thereby undercompensating the former and overcompensating the latter. The CAT recognises that the surplus (or value) that the different types of users (in this case, consumers and advertisers) derive from the platform ‘varies between buyers [and is] dependent on the particular buyer in question’.^[45] The CAT finds that in a two-sided market:

the potentiality for the same excess profits to be the result of unfair prices in two markets cannot be forgotten . . . . [The] Tribunal must be astute to avoid over- as well as under-compensation, and this is a point that needs at least to be addressed in any methodology advanced in this case.^[46]

Thus, the CAT seem to caution against incorrectly matching potentially excess profits earned by Meta from advertisers with alleged losses suffered by users.

Fourth, the CAT is alert to attempts to shoehorn consumer protection claims into the competition opt-out regime. The PCR claimed harm to users owing to unfair trading conditions because Facebook’s T&Cs were allegedly complex, far-reaching, opaque and misleading, and imposed on a ‘take it or leave it’ basis. The CAT, however, noted, ‘[t]hese allegations seem to us not to be competition law infringements at all, but rather some other – possibly consumer protection based – claim’.^[47] The CAT also observed that an allegation about misleading T&Cs or misrepresentation ‘turns on the extent to which individual class members were misled. That, in turn, strongly points against this claim being susceptible of collective proceedings.’^[48]

Ultimately, in its judgment in Gormsen, the CAT was critical of a failure by the damages expert to set out a clear blueprint for trial given the complexity of the respondent’s business and the precise focus of the excess pricing rules. It is also a shot across the claimants’ bows regarding the scope of some claims that embrace more consumer law protection issues.

Notes