Assessing Antitrust Damages in Follow-On Actions Against Cartels

Assessing damages in follow-on actions against cartels that have infringed EU or domestic antitrust law is still in its infancy in Europe. This Article highlights the reasons why this issue has only recently started reaching the courts and analyses the pitfalls and problems courts face when calculating or estimating the amount of loss caused by a cartel to individual plaintiffs. As this endeavour raises complex economic questions, the Article outlines the basic economic approaches to measure antitrust harm and derives general principles from EU law on how courts should deal with economic evidence in follow-on actions to ensure an effective enforcement of competition law.


I. Introduction
The use of the European or domestic antitrust 1 rules in private proceedings as a sword 2 to obtain damages and/or an injunction was historically confined to narrow specified scenarios.Here a sharp distinction could be drawn with US law, where private actions have been a central enforcement axis since the enactment of the Sherman Act in 1890.After the foundation of the European Economic Community ('EEC'), the European Commission focused on building a competition law centred on a strong public enforcement and worked towards the uniform application of EU law across the Community. 3At that time, this was a reasonable decision.The Commission was a new player on the field of antitrust law enforcement and the legislature had thus to ensure that public enforcement got effectively 'up and running'.In addition, the member states' laws to protect unfettered competition were of very recent origin or were only developed after the EEC was founded. 4As a result, the *Many thanks to Martin Fischer, Carsten Krüger, Jakob Rüggeberg and especially Adrian Fritzsche, for their helpful comments on an earlier version of this paper and Okeoghene Odudu and Louise Gullifer for helpful comments on my presentation at the 3CL/CELS Seminar at the University of Cambridge.All remaining errors are the author's own. 1 The terms 'competition law' and 'antitrust law', or rules respectively, are used interchangeably in this Article.
2 European variant of private enforcement concerned for long time primarily 'antitrust shield litigation', ie disputes in which one party relied on the nullity defence enshrined in Article 101(2) of the Treaty on the Functioning of the European Union ('TFEU') (ex Article 85(2) EEC and Article 81 (2) EC). 5 Sword litigation was rare, although not non-existent. 6he European centralised enforcement system came under pressure as the European Community grew.Prior to the Eastern enlargement, the Commission 7 called for decentralisation and the end of the notification system for the exemption of restrictive practices enshrined in Regulation 17. 8 The 'enhanced role' for national courts envisioned in the White Paper on the modernisation of the enforcement system of 1999 9 became law with the adoption of Regulation 1/2003. 10Yet what truly kicked off private enforcement came from the European Court of Justice.In Courage v Crehan, the Court held in 2001 that 'any individual' must be able to claim damages for breaches of the European antitrust rules.Otherwise the full effectiveness of the European antitrust rules 'would be put at risk '. 11 To further facilitate private enforcement, the European legislature finally harmonised-after years of intensive discussion on how much US-style litigation should be incorporated into EU law and on how public and private enforcement should be optimally synchronised-key issues of substantive and procedural law in the Damages Directive 2014/104/ EU ('Damages Directive'). 12s a consequence, the role and practical importance of private enforcement has changed significantly since the start of the new millennium. 13A regularly updated statistical overview published by Jean-François Laborde counted in 2021 an overall number of 299 private enforcement cases 14 brought before courts across Europe (including Great Britain, Switzerland, and Norway), 15 whereas in December 2009 this number was merely 46. 16ost of these cases are follow-on actions for damages that are brought by businesses (not consumers) 17 before national courts after the European Commission or a national competition authority has detected and fined a cartel.As such proceedings are regularly brought against one of the entities addressed by the (final) infringement decision of the European Commission or a national competition authority of an EU member state, the decision has a binding effect or serves at least as prima facie evidence in civil proceedings before national courts that an infringement took place (Article 16(1) Regulation 1/2003, Article 9 Damages Directive). 18Today (after many hurdles posed by national law were cleared in earlier proceedings), the parties thus essentially litigate about the issues of causation and damages in follow-on proceedings.
The analysis and quantification of harm is a topic that is strongly intermingled with economic theory and 'the marriage between economics and … antitrust policy becomes rocky when it reaches the law of damages'. 19This is so because damages assessment hinges to a large extent on how prices and quantities would have evolved in the hypothetical scenario in which the anticompetitive act had not occurred.To assess this 'but for' scenario, courts are faced with complex issues.Even though the beginnings of an effective antitrust enforcement in Europe date already back for some time, the quantification problem has only recently started reaching the courts in great numbers and many issues are still controversial.That 'it is early days in the quantification of cartel and competition damages' 20 has various reasons: many damages actions brought after Courage were dismissed for other reasons so that judgments did not have to decide on the amount of damages, 21 other disputes were settled out of court, 22 and in some countries procedural peculiarities further delayed rulings on the quantification of harm caused by a cartel. 23gainst this background, this Article explores the basic principles of assessing damages in follow-on actions against cartels.The analysis will focus on the most important head of loss, ie claims for overcharge charged by a cartel where there was no passing-on.
Many rules on the assessment and quantification of damages in antitrust cases are rooted in national law.That does, however, not mean that European law has no bearing on the interpretation and application of these rules.As will be explained in Part II of this Article, the right to claim If the action for damages is brought against a legal entity not addressed by the infringement decision, the defendant is entitled to dispute the infringement alleged.In practice, such a scenario occurs if a plaintiff sues, for example, a subsidiary of the parent company that was fined based on the economic entity doctrine, the former can then dispute that it belongs to the same undertaking as the parent company.For example, in Germany, courts showed a tendency to rule on liability without dealing with the quantum of damages by declaratory or interlocutory judgments (Feststellungsurteile, Grundurteile), thus postponing dealing with the quantum of damages, before the Bundesgerichtshof raised the bar for such an approach, see Rengier, note 21 above, pp 74-75.
damages for infringements of the European antitrust rules is based on a solid European foundation that has a strong impact on the application of national law.
Part III will briefly set out the key economic approaches for the determination of the counterfactual and highlight some meta studies on which courts have placed some reliance when estimating damages.
Part IV will develop some general principles flowing from European law that must be respected by courts of EU member states when assessing and quantifying damages.Since Brexit became effective, courts in Great Britain are no longer bound by these principles.But as long as the enforcement system is structured in a similar manner as it was when Great Britain was an EU member state, these principles should nonetheless remain persuasive.
Part V summarises the main conclusions of this Article.

II. Assessing Damages: The Legal Framework
A. Objectives of European Antitrust Law Enforcement: Compensation and Deterrence As the application and interpretation of law often makes recourse to its objectives, the analysis must start by recalling the two purposes of damages actions for breaches of the European antitrust rules.First, actions for damages should compensate victims of anticompetitive conduct to the full extent.This general guiding principle was already spelled out by the European Court of Justice in the first two rulings on antitrust damages, Courage 24 and Manfredi. 25Later, the European legislature incorporated this maxim into Article 3 Damages Directive.The compensation objective should also close a gap left by public enforcement.Fines imposed by public authorities usually go to the state budget.The money is used to serve the public good and is typically not used to indemnify the victims of anticompetitive conduct.To remedy such losses, private actions are necessary to achieve corrective justice. 26eyond compensation for the harm suffered, the European Court of Justice has underscored that private actions for damages also serve a preventive purpose.Potential antitrust infringers are deterred from engaging in unlawful activity.Damages actions then not only serve corrective justice but also help to maintain undistorted competition on the market. 27That is why 'actions for damages for infringement of EU competition rules are an integral part of the system for enforcement of those rules, which are intended to punish anticompetitive behaviour on the part of undertakings and to deter them from engaging in such conduct'. 28The Damages Directive, however, does not clearly spell out the object of deterrence. 29Even though the recitals reiterate the principle that the right to claim damages serves 'the full effectiveness of Articles 101, 102 TFEU', the rules of the Directive are geared towards the object of full compensation.Article 3(3) Damages Directive even rules out overcompensation by 'means of punitive, multiple or other types of damages'.It was therefore rightly pointed out that the Directive focuses on full compensation but not on punishment. 30owever, that does not mean that the old case law on prevention was abrogated.There is a difference between punishment on the one hand (which is ruled out in Article 3(3) Damages Directive) and prevention (deterrence) on the other.The latter aims at the establishment of an effective system of law enforcement without going so far as imposing punitive or treble damages US-style.In European tort law, it is widely accepted that prevention is to be distinguished from a punitive effect. 31Moreover, the effect of damages actions on free and unfettered competition on the market was underscored by the European Court of Justice as a paramount enforcement goal and so it is hard to argue that the Directive sought to change the double objective of private enforcement.Cutting back the object of prevention through the Directive would also not be in line with the legislature's intention to codify 'the acquis communautaire on the right to compensation for harm caused by infringements of Union antitrust law' in the Directive (cf Rec 12).Therefore the goal of prevention (deterrence) remains at least a secondary goal under the Directive. 32

B. The Europeanisation of Private Actions for Damages
From a tort law perspective, the significance of the Courage case law lies in the fact that the European Court of Justice derived the right to claim damages for breaches of EU antitrust law directly from Article 101 TFEU.This does not mean that all of the conditions of this right are based on European concepts.But given the various impediments to the effective enforcement at the national level, the European Court of Justice has over time grounded key conditions in European law to ensure effective enforcement.
First, European law determines the person having the right to claim damages for antitrust law infringements ('any individual'). 33The claim for damages is thus not restricted to persons operating as customers or suppliers in the same market but extends to persons that are suffer damages on related markets.Thus, as decided in Otis v Land Oberösterreich, an institution granting financial subsidies for housing projects based on the building costs can bring an action for damages against a cartel fixing prices for elevators that were built into these premises. 34uropean law also determines the question of from whom damages can be sought, thereby harmonising a key condition which was disputed in the member states even after the transposition of the Damages Directive. 35In Skanska and Sumal the European Court of Justice synchronised private and public enforcement.Article 101 TFEU applies to 'undertakings' and the Commission can impose fines on these entities under Article 23 Regulation 1/2003.This 'undertaking' is also liable for damages in private enforcement cases.According to the European Court public and private enforcement has insofar to run in parallel as the concept of 'undertaking' cannot have two different meanings in the different strands of enforcement. 36As a consequence, if an 'undertaking' is comprised of more than a single legal entity, all units forming the 'undertaking' can be sued for damages irrespective of their knowledge of or their 'fault' for the breach.Put simply, two entities form an 'undertaking' if one entity exercises decisive influence over the other unit.Such an influence is presumed when a parent company owns all, or nearly all, of a subsidiary's share capital. 37As a result all entities that are part of an economic unit are subject to ( joint and several) liability. 38Also a firm or person that essentially carries on the business of the (dissolved or restructured) law infringer can be held liable given that the economic continuity principle developed under Article 23 Regulation 1/2003 39 applies in private actions for damages. 40here is also a strong European bearing on the causation criterion even though it is not settled yet if causation is a truly European concept.Advocate Generals Wahl 41 and Kokott 42 argued in Skanska and Otis that all conditions relating to the question of whether damages can be claimed are rooted in European law.The European Court of Justice is less clear in this regard.In Manfredi and Kone, the Court explicitly stated that it is for the domestic legal system of each member state to lay down the rules for the application of the concept of a 'causal relationship'. 43In Otis, though, the court did not repeat its clear statement, but held that 'any loss which has a causal connection with an infringement of Article 101 TFEU must be capable of giving rise to compensation in order to ensure the effective application of Article 101 TFEU'. 44Basing the argumentation on the effective application of Article 101 TFEU is a step away from Kone towards a more European interpretation of the causation requirement.Yet the Court did not explicitly state that national concepts do not matter at all when assessing the causal link.
The European Court of Justice has also laid down the basic requirements for awarding damages.Even though it is for the domestic legal system of each member state to set the criteria for determining the extent of the damages-subject to the principles of effectiveness and equivalence-the court held in Manfredi that the right to full compensation amounts to compensation for actual loss (damnum emergens) and loss of profit (lucrum cessans) plus the payment of an appropriate amount of interest 45 (even though the precise amount is a matter of the applicable national law). 46Punitive or exemplary damages are not part of European law. 47These findings are mirrored today in Articles 3(2)-(3) Damages Directive.Domestic law governs the general rules of evidence and the rules relating to the costs of proceeding.These areas of law are however not a domaine réservé given that the principles of effectiveness and equivalence apply.Against this background, the European Court of Justice has recently held that the principle of effectiveness does not preclude the application of a domestic rule according 38 Sumal, note 18 above, para 44.This is in sharp contrast to US law where courts are reluctant to extend liability to a parent company for antitrust infringements committed by a subsidiary.See C König, 'An Economic Analysis of the Single Economic Entity Doctrine in EU Competition Law' (2017) 13 Journal of Competition Law & Economics 281, p 284.
to which each party has to bear its own costs and half of the common cost of the proceeding in the event that the claim is upheld in part only. 48 Antitrust Damages and Their Quantification under the Damages Directive Even though the Damages Directive is based to a large extent on pre-existing case law, it has introduced some additional assistance for private claimants regarding the assessment of damages.49 First, the Directive set forth a rebuttable presumption that cartels cause harm (Article 17(2) Damages Directive).This harm can be caused 'downstream' or 'upstream'.In the case of producer cartels it will thus be purchasers that rely on it and in case of buyer cartels the presumption works on behalf of the suppliers.50 The Directive presumes merely that the damages are larger than zero (unless the defendant can rebut this presumption).51 The European legislature has refrained from adopting a presumption with regard to the quantum.Some member states have however codified such presumptions. Thpioneer in this regard was Hungary.Already before the transposition of the Directive, Hungarian law provided a rebuttable presumption that cartels cause an overcharge of 10%.Latvia and Romania followed this approach when transposing the Directive and codified rebuttable presumptions of 10% and 20% respectively.52 However, the majority of member states refrained from introducing such a presumption.This is reasonable as cartels may have quite different economic effects so typical overcharges are difficult to define economically.But one should also not overestimate the value of a presumption.Even in countries having laid down such presumptions, they are rebuttable and if the defendant can rebut the presumption courts in those countries will have to carry out the complex analysis in the same way as courts in countries without such a presumption as to the quantum.
Without a presumption in place, assessing the quantum must be based on the facts of the case.At first sight, the assessment of damages seems to be a straightforward computing operation.To compensate the claimant, a judge needs to compare the financial position the claimant is in and the position he or she would have been in absent the infringement.In follow-on actions against cartels, the harm typically claimed flows from cartel overcharges but other losses can also be remedied, such as a loss of profits caused by fewer resales by the direct purchaser to indirect purchasers.The counterfactual ('but for scenario'), ie the development of prices and sales absent the conspiracy, that serves as basis for assessing the harm caused, must be based on a hypothetical economic market environment.As markets are often complex, it can only be based on estimations and forecasts.Against this background, the Damages Directive further provides that courts must also be able to estimate the amount of harm if it is established that a claimant suffered harm but it is practically impossible or excessively difficult precisely to quantify the harm suffered on the basis of the evidence available (Article 17(1) Damages Directive).As a form of assistance to national courts, the European Commission has commissioned a Practical Guide on quantifying harm in action in damages actions with economic methods. 53ven though it is of a purely informative nature, 54 this document lists accepted economic standards on which courts can rely.The Directive further provides for a disclosure mechanism to mitigate information asymmetries between the parties (Articles 5-6 Damages Directive). 55This asymmetry can also exist with regard to data that can be helpful for assessing damages.In addition, the European Commission and national competition authorities can cooperate with national courts (Article 15 Regulation 1/2003) and national authorities can, upon request of a court, assist that national court with respect to the determination of the quantum of damages where that national competition authority considers such assistance to be appropriate (Article 17(3) Damages Directive).In practice, antitrust authorities only give very general information to courts and do not calculate damages in the case at hand.For such a task, they would need significantly more resources. 56

D. Conclusion
Summing up, various layers of European concepts and principles govern the enforcement of competition law.Even the assessment of antitrust damages is not a domaine réservé of domestic law any longer as the goals of private enforcement and the principle of effectiveness have an impact on national rules on damages and evidence.Much of the difficulty in assessing the precise harm caused by cartels to single purchasers or sellers lies however in the economic complexity involved in determining the counterfactual based on the hypothetical development of the market absent the anticompetitive conduct.The key approaches to map this counterfactual will be described in the following Part of the Article.

III. The Economics of Assessing Cartel Overcharges in a nutshell
Economists have developed very different approaches and models to delineate the proper counterfactual ('but for scenario') to be used as the basis for the computation of the harm caused.The following overview sketches the most important methods for private enforcement cases without being complete.Generally speaking, there are three major economic approaches.The first uses data generated across markets or time to map the hypothetical but for scenario ('comparator-based models').The second bases the counterfactual on cost structures or financial information ('cost/finance-based approach').Finally, the third relies on Industrial Organization theories and models to estimate prices and quantities absent the anticompetitive conduct ('Industrial Organization simulation models'). 57In addition, there are meta studies on average cartel overcharge on which courts have placed some reliance when quantifying damages, even though most economists do not consider these studies to be helpful in individual cases.

A. Comparator-Based Models
Comparator-based models have been used frequently before national courts as basis for assessing overcharges.Such calculations are usually provided by party appointed experts. 58uch models base the counterfactual either on the development of prices in a comparable product or a different geographic market that is not affected by anticompetitive conduct ('yardstick approach'). 59Alternatively, they are based on the development on the market affected by the cartel but use data from a time before or after the cartel was active.The strength of the comparator-based approach is in grounding the analysis on robust real-life data observed in the same or a similar market and not on some made-up proxies. 60The complexity of comparator-based models can vary significantly given that the but for scenario can be based on very different data sets, ranging from a simple comparison of average prices to a far more complex regression analysis and also the depth of the studies can vary.An advanced approach is the 'difference-in-differences' technique.Under this approach one or more alternative market(s), so called 'control market(s)', has to be chosen.Each control market must have a similar structure to the cartel market 61 and must not have been affected by anticompetitive conduct.Control markets can either be a different geographic market of the same product (eg the truck market in the US instead of the EU) or a market for a related product (eg the market for light transporters instead of heavy trucks).After the control market(s) has/have been determined, the price changes in the control market(s) and in the market affected by the cartel must be measured over time.The model got its name from the type of analysis that has to be carried out.It does not compare average prices on two markets but looks at the difference in prices between the affected and the competitive market over time.This is so because the control market is seen as a surrogate for how demand and prices on the affected market would have changed absent the anticompetitive conduct.The estimated impact of the cartel 'is thus the difference between the price changes in the affected and unaffected markets-that is, the difference between the differences'. 62Even though this advanced estimation strategy can be applied in its simplest form as a comparison of four average prices, in real-world cases it is usually applied via a multivariate regression given the need to control for different confounding factors. 63In practice, such a regression analysis is however rarely applied because of a lack of reliable data.In addition, if no proper control market with a sufficient degree of similarity to the affected market can be identified, the model can also not be applied. 64Difficulties arise also when even the control market is affected by cartel activity, be it of the same cartel or another one.

58
As claimants have to specify an exact amount, usually an economic expert appointed by the claimant is required to introduce a first calculation to the court.This assessment is usually 'de-constructed' by an expert hired by the defendant and-depending on the applicable procedural law-the court might appoint an economic export to assess the presented calculations or to introduce a new economic assessment.Court-appointed and party appointed experts rely on the same approaches discussed in this Section.Similar structure means that the demand side and supply side must be comparable.Otherwise the markets are likely to evolve differently over time.Supply side differences can result eg from different technology built into products or differences with regard to regulatory requirements.Differences regarding the demand side can eg concern buyer structure (craftsmen or consumers vis-à-vis large firms).Given that the 'difference-in-differences' model often cannot be implemented in practice, parties regularly resort to comparisons of prices in the same market over time (before, during, and after the anticompetitive conspiracy) as there is a reasonable chance that it will be possible to collect a sufficient amount of data for such an analysis.Such a comparison over time works when stable market structures can be observed and it can take into account a multitude of firms operating in the market to assess the development of average prices. 65Comparisons over time pose more problems if the market is not stable.Price fluctuations can have various reasons (eg price increases for raw materials, spiking energy costs, price wars, etc).All these factors external to the cartel can affect the price level and thus the estimation might need to be adjusted.Difficulties further arise when the exact end of the cartel is unclear as then periods with inflated prices might be taken into account in the analysis, which would then underestimate the harm caused.The same holds true when the cartel has assisted market participants in establishing patterns of price leadership or parallel conduct 66 or if prices are kept high to reduce the risk of being sued for damages, 67 so an important issue in the analysis is the question from which point of time prices in the post-cartel period can be regarded as 'non-collusive'. 68s the model only works if both supply and demand side are comparable, the comparison can also be distorted for other reasons.If after the end of the conspiracy a new player enters the market, the stronger competitive pressure usually leads to prices that are lower as it would be the case with fewer players on the market.This might overestimate the anticompetitive overcharge given that the difference between the prices during and after the cartel period is larger than in case of a stable structure.In turn, if after the cartel came to an end, some cartel members exit the market or are taken over, the price level that can be measured will be higher than without this structural change so that the before and after model bears the danger that the harm caused by the cartel is underestimated.In such a scenario, the before and after model can only measure a type of minimum damage. 69Similar issues can occur when there are significant changes at the demand side.If, eg, demand shifts dramatically towards a new product that has entered the market the prices for the product that was once part of a cartel might fall and as consequence the harm might be underestimated.

B. Cost-/Finance-Based Approach
An alternative approach to assess the cartel overcharge is to look at cost or more broad financial criteria gathered from comparator firms or an industry.The idea behind the cost-based approach is to estimate the production cost per unit and add a reasonable rate of profit that would have been charged under competitive conditions as basis for modelling the counterfactual. 70The financebased approach would look at the financial performance of the claimant and/or defendant to assess the impact of the anticompetitive conduct. 71As accounting costs do not necessarily reflect economic cost, finding robust cost estimates is often a challenging task. 72or an antitrust authority that wants to assess the harm caused by a cartel, an assessment based on cost and finance structures can be a workable alternative.Public enforcers can order the incumbent to disclose relevant information and can also gather relevant data from other market participants.
For private parties, the cost/finance-based approach is a less viable option, given that often a lot of data is in the hands of parties other than the claimant.Even under a working disclosure mechanism this approach can be difficult to implement.In addition, the defendant will usually know its cost data much better than a claimant to which parts of it have been disclosed.If the assessment is predominantly based on the cost of the defendant, this model tends to favour the defendant.But if a sound analysis based on the comparator-approach fails for lack of sufficient data, the claimant may resort to a finance-based alternative to prove the quantum of the overcharge paid.

C. Industrial Organization Simulation Models
The Industrial Organization approach seems to be the approach from those described in this Section that has so far had the least impact in the courtroom. 73It is based on theories, assumptions, and models developed in the field of Industrial Organization to predict market outcomes-and not on a comparison of prices over time or across markets as is the case for comparator-based models. 74he Industrial Organization approach is often based on very general assumptions about markets that exclude the complexities of the real world.This raises the question of whether in the case at hand they can be adapted sufficiently to serve as a proper approximation of the counterfactual.Claimants seem thus far to be hesitant to resort to this more theoretical approach, but it might be necessary as last straw if it is not possible to assess the overcharge based on another approach.

D. Abstract Meta Studies
The approaches described above are based on information collected in the affected or a related market or on models that reflect to a certain extent the particularities of the market in question.However, private parties and courts have also-as a supporting argument-taken into account very abstract empirical meta studies on past cartels in order to estimate damages in private enforcement cases. 75These meta studies have calculated average overcharges charged by cartels which were extracted from the compilation of various sources such as judgments, estimations by government sources, or information contained in social sciences studies in various countries over time. 76The most comprehensive one was published by John M Connor and Robert H Lande based on various sources on cartels dating back in part to 1780. 77The study showed that international cartels can usually charge higher overcharges than national cartels and that bid rigging cases have in average a lower overcharge than conventional price fixing or market allocation conspiracies. 78The Oxera study of 2009 examined the data underlying this study and recalculated the estimation by, inter alia, focusing on more recent cartels that were active from 1960 onwards. 79Oxera concluded that in 93% of the cases, the overcharge is positive, ie above zero, and merely 7% of the cartels could not raise the price above the presumed competitive level. 80Around 35% of the cartels could fix an average overcharge of 10-20%, around 18% an overcharge of 20-30%, and around 73 See Veljanovski, note 20 above, para 11.39: 'In no case was an (…) simulation approach accepted by the courts as for its award of damages'.For an overview of different meta studies and their results, see Inderst and Thomas, note 57 above, p 89; Veljanovski, note 20 above, para 3.11.Ibid, p 91, also noting that 'it may be that the empirical studies tend to focus on cartels that have been operational and that are therefore most likely to have had an impact on the market'.
https://doi.org/10.1017/cel.2023.8Published online by Cambridge University Press 15% an overcharge of 30-40%.An updated study published recently by Connor and Lande on a wider regional basis showed a median average of 20,70% for national cartels and 22,50% for international cartels that operated from 1990-2018. 81

E. Conclusion
Economic scholarship and practice have developed a wide range of approaches to gauge cartel overcharges over the years.All models and techniques however just provide approximations of a purely hypothetical but-for scenario that cannot be measured with precision.It needs to be recalled that economics is not a hard science and therefore the outcome of any calculations depends-as always in economics-not only on the quality of data at available but also on the (subjective) assumptions, corrections, or proxies underlying the model.The three main approaches used to assess cartel overcharges described above can be tailored to some extent to the case at hand.Given that cartels come in very different shapes and sizes and that the application of all three major model has pitfalls, it is not possible to argue that one approach is superior to the other and must therefore be exclusively applied to private enforcement cases.Rather parties and courts can rely on any of the three main approaches and also other approaches to approximate the counterfactual.It is further accepted that several methods or tools can be applied alternatively or cumulatively when assessing the harm caused. 82Against this background, the following Part develops general principles on the thorny issue of how courts must make use of economic insights when ruling on the assessment of antitrust harm.

IV. General Principles of Damages Assessment
Under the principle of full compensation, any individual harmed by the cartel must be compensated for the losses caused by the anticompetitive conspiracy.The Damages Directive sets forth a presumption that cartels cause damage but this presumption does not extend to the quantum (Section II.C above).The burden of proof regarding the quantum is governed by domestic law.As a general rule it falls upon the claimant to demonstrate, with the necessary certainty, to what extent he or she has suffered losses from the anticompetitive conduct. 83Even though matters of proof and calculation are to a large extent governed by domestic law, it is possible to deduce some general principles of damages assessment from European law.

A. Reasonable degree of precision
It is a simple truth that courts when ruling on actions for damages must 'aim to compensate damage as precisely as possible, once the existence of and [liability] for that damage are established'. 84But precision comes at a price.In classic tort cases the price to establish a very high degree of precision is often relatively moderate.Taking the example of a car accident where the injured party claims the costs for repair of the car.The damage can be calculated based on the invoice presented and if the defendant argues that the repair was overpriced some offers of other repair shops in the vicinity will give the judge an impression of whether the claimed price is reasonable or has to be mitigated.In private antitrust cases things are much more complex.Today, cartels are covert operations and in practice it does not happen that claimants can get their hands on documents or witnesses that exactly reveal how the cartel calculated the overcharge over time. 85Therefore the counterfactual has to be approximated using other techniques with all the complexities attached (see Part III above).
If one sets aside what is generally available, the production of economic evidence based on precise data that is collected and modelled for the case in issue can also significantly raise costs.Judges will therefore have to balance the need for precision with setting an evidentiary burden that does not hamper private enforcement.Otherwise the deterrence goal can be weakened, given that a too high threshold causing excessive upfront costs will prevent harmed persons from pursuing their claims. 86n addition, as it is impossible to gauge hypothetical market developments with mathematical precision, it cannot be argued that courts must base their assessment on such economic evidence, that is-from an economic point of view-the very best approximation to the 'but-for scenario'.Rather a reasonable approximation must suffice.A German court recently quoted the British statistician George Box ('All models are wrong, some are useful') to underscore that the demand for the best possible (regression) analysis will not necessarily lead to more precise outcomes (given that a more refined model will inevitably also have some imperfections) but merely to more costs and complexities that could easily hinder the deterrence objective of damages actions. 87s economics is not a hard science and given the factual and economic complexities of antitrust damages claims, courts should accept a reasonable degree of precision and should not strive for 'precision at all cost'. 88The standards must be reasonable as otherwise the effectiveness of private enforcement will be in danger but also not too low as otherwise there might be incentives to lodge unfounded claims. 89Thus, as Lewison J summarised in Devenish Nutrition Ltd v Sanofi-Aventis SA (France), one of the first follow-on actions against the vitamins cartel: 'the restoration by way of compensation is often accomplished by "sound imagination" and a "broad axe"'. 90hat type of evidence has to be presented depends on the facts of the case.In actions against cartels that operated in complex markets more advanced studies seem to be needed in order to assess the overcharge with a sufficient degree of precision.By contrast in cases concerning stable market conditions much simpler types of economic evidence will suffice.

B. Estimation and Disclosure
'Access to information in the form of data, documents and analyses is critical to economists engaging with the substance of competition and regulatory cases'. 91Things do not differ much when it comes to the work of the economists who provide expert opinions on the amount of harm caused by a cartel to individual claimants.Some relevant data or documents might be in the hands of the defendants or third parties.The Damages Directive has therefore introduced a disclosure mechanism that empowers courts to order one of the parties or a third party to disclose 85 If the claimant can however prove with documents or witnesses that the cartel agreed on an overcharge of X% and there was such a rise of prices in the market, it is likely that the overcharge was X% at least in the short run.If the cartel was however active over many years one would have to consider whether the hypothetical price absent the infringement would have changed, for example through cost increases.See Inderst and Thomas, note 57 above, p 159. evidence that is in their hands in an attempt to limit the information asymmetry (Section II.C above).The European Court of Justice attaches great importance to this mechanism.The Court underscored in Tráficos Manuel Ferrer that 'the need to undertake a judicial estimation of the harm may depend, in particular, on the result obtained by the claimant following a request for the disclosure of evidence pursuant to [Article 5(1) Damages Directive]'. 92If the claimant did not do what he could have done to obtain the necessary information and as a result it is practically impossible for the court to assess the quantity of the harm caused, 'it is not for the national court to take the place of the [claimant] or to remedy its shortcomings'. 93ven though it cannot be disputed that it is up to the party on which the onus of proof lies to obtain and present the necessary evidence to enable the court to assess the overcharge it is also essential to take into account the limitations of the disclosure system.A comparative study on private enforcement has recently highlighted that in the area of disclosure 'there is an extraordinary heterogeneity between the legal orders of the different [member states], made worse by accentuated degree of legal uncertainty when interpreting and applying those national rules.This was so before the transposition of the Directive, and the effects of this heterogeneity are likely to linger in the interpretation of the new rules'. 94As key concepts are not clearly defined by the Directive (what type of evidence that can be obtained, confidentiality requirements, etc), at least in those member states that have had only limited experience with the concept of disclosure it will take a while before the system will run smoothly. 95Against this background, courts should not overestimate the effect of the disclosure mechanism when confronted with the question of how to estimate the quantum according to Article 17(1) Damages Directive and the respective national rules.
In addition, the counterfactual can be proven in very different ways.Comparator-based studies often rely on data collected by claimants alone and can still present credible findings at to the overcharge even without data provided by the defendant.Thus, courts should not conclude that because the claimants did not make use of the disclosure mechanism, that their presented evidence is insufficient.
In turn, if (some) data was disclosed to one of the parties, that does also not exclude the possibility of estimating the quantum.The European Court of Justice has with good reason pointed out that 'even where the parties are on an equal footing as regards the information available, difficulties may arise during the specific quantification of the harm'. 96

C. Meta Studies as Basis for the Damages Calculation?
In some court rooms the parties referred to results of meta studies in the process of assessing damages, especially the one published by Oxera. 97Although Spanish courts have rejected expert opinions based solely on such studies as being superficial, 98 they have nonetheless taken these studies into consideration when computing fixed standardised overcharges ranging from 5 to 15%.The amounts fixed by these standardised estimations of the overcharge were granted by Spanish courts to all claims against the truck cartel that were lodged before the respective courts once the standard had been set.The percentages between 5 and 15% were considered to be a '"prudent" estimation extracted from meta-studies on the average overcharge produced by cartels'. 99In a similar manner, the Landgericht Dortmund referred a decisions relating to the rails cartel to the results of the meta studies as a supporting argument for the estimation (overcharge of 15%). 100 The estimation itself was based on the cartel's structure and functioning, as the economic study presented by the plaintiff was-according to the court-too superficial to be of use for the calculation of damages.In addition, the Landgericht was of the opinion that the production of an economic study based on the comparator-based or cost-based approach model would be too burdensome for the claimant given the various uncertainties and complexities of these economic approaches 101 and their high cost ('at least a small six digit sum'), which was seen as being out of proportion given the low amount of damages being claimed. 102In a parallel case, the Landgericht Dortmund estimated on a similar basis the amount of damages at 10% overcharge. 103rom an economic standpoint, meta studies are very abstract and contain many cases from other jurisdictions which might differ from the scenarios in Europe.In addition, the Oxera study pointed out that the results might be slightly distorted as such studies focus on cartels that operated in the past over a longer period of time. 104Further, given the complexity involved in measuring the overcharge, it is intuitive that these studies cannot claim accuracy.Most economists therefore argue that courts should not look at them when dealing with antitrust damages as it is not possible to say that a cartel of type X generates a typical overcharge of Y% when the overcharge could instead even be zero. 105The empirical studies clearly indicate that cartels usually-albeit not always-affect the price but are less clear on the average magnitude of this effect.
From a legal perspective, one has to take into account that courts which have to decide on 'smaller' claims cannot order the plaintiff to present a comprehensive economic study as the cost of producing this would be out of proportion to the claim pursued and would thus hamper effective enforcement.In such cases the operation of the cartel (number of participants, degree of market coverage, duration of the conspiracy, geographic scope, etc) can give valuable hints on the chances of the cartel having been able to raise prices and with regard to the quantum it seems reasonable to prudently rely on meta studies for such claims as long as there are no credible indications that the cartel did not cause damage at all.Faute de mieux, in proceedings where expensive economic studies would be too burdensome for the plaintiff it seems justified that courts rely on other factors to substantiate their findings including average overcharges calculated in meta studies.
Courts in Europe seem to have understood this very well.The Laborde survey noted that out of 58 cases in which damages have been awarded, courts in Belgium, Greece, Germany, and-in particular-in Spain, have estimated a rate or an amount of overcharge on 14 occasions, 106 ie in approximately 25% of the cases in which damages were awarded.A figure, that certainly caused raised eyebrows amongst antitrust economists.Many of those courts which estimated the quantum have, however, 'explained that they considered this method appropriate in situations in which the cost of obtaining the opinion of a court-appointed expert would be disproportionate relative to the potential amount of damages'. 107The trend towards an estimation of damages is thus not a rejection of economic insights but is the result of a desire to keep the costs of private enforcement at a reasonable level.

D. The Cause of Economic Uncertainty
Finally, courts must take into account that the economic uncertainty that regularly surrounds the assessment of damages is caused by the defendant cartel members deliberately inhibiting competition by forming the cartel.This conspiracy is the cause of the disruption of the competitive process on the market and the cartel members are 'also responsible for the difficulty in determining the hypothetical prices on the market absent the anticompetitive agreement' as was correctly noted by the Bundesgerichtshof, the highest court for civil and commercial matters in Germany. 108gainst this background, the Bundesgerichtshof upheld a standard contract term in terms of which the supplier of goods or services which had participated in an unlawful cartel had to pay, as minimum damages, an overcharge of 5% to the purchaser unless the supplier could demonstrate that the amount of loss caused by the infringement was lower.The court further reasoned that-in the absence of better insights-standard contract terms providing for an estimation of the overcharge of up to 15% of the contract value were reasonable and thus valid, as long as the defendant was allowed to demonstrate that the damages were below that threshold. 109Such standard terms are common in contracts with public bodies that are purchasing goods or services but such terms are much less frequent in contracts concluded between two private businesses.
The reasoning of the Bundesgerichtshof is also of value with regard to the law of evidence in general.If the economic uncertainty surrounding the assessment of damages is caused by the defendant and his or her co-conspirators, and if one further considers that the assessment of harm largely depends on the operation and details of the anticompetitive cartel in the market, ie on information that is usually much better understood by the defendants than the claimants, it can be argued that uncertainties about the quantum should not play into the hands of the defendants who are the cause of this uncertainty. 110Therefore, courts must apply the law in a manner that reflects the reason why it is often difficult to assess damages.As the implementation of this reasoning concerns issues of (domestic) civil procedure, such as rules on evidence or provisions relating to how judges can steer the proceedings, courts in different jurisdictions can take different routes in implementing this reasoning, as long as the principles of equivalence and effectiveness are respected.A solution could be for judges to take a look at the structure of the cartel (type of agreement, duration, market coverage, etc) and to communicate at a rather early stage of the trial a preliminary assessment on a possible range of the overcharge as a type of anchor (eg around 10%, 5-10%, etc). 111This anchor can be based on the empirical assumption that cartels try to maximise their profits when engaging in unlawful conspiracies and therefore will try to obtain a substantial economic benefit from the arrangement.In this respect, the meta studies can serve as point of reference for this preliminary assessment.Setting such an anchor for estimating the overcharge will put pressure on the party that wants to deviate from that preliminary assessment, be it that the claimant wants a higher amount of damages or that the defendant wants the court to find that the damage was lower or even non-existent.The relevant party has to then bring forward an explanation supported by evidence for why the award of a higher or lower amount would be justified so that the 108 BGH -Schienenkartell VI, note 83 above para 41: 'Dieser ist für den Eingriff in die Freiheit des Wettbewerbsprozesses und für die sich daraus ergebende Störung des Preisbildungsbildungsmechanismusder zentralen marktwirtschaftlichen Koordinierungsfunktion, auf deren Funktionsfähigkeit redliche Vertragspartner bauenebenso verantwortlich wie für die daraus resultierende Schwierigkeit, den hypothetischen Marktpreis zu ermitteln, der sich ohne die Absprache eingestellt hätte'.court can adapt the preliminary assessment in the course of the trial.112Such a pre-assessment works better the more details about the structure of the cartel are laid down in the infringement decision.

V. Conclusion
Now that the European Court of Justice has "Europeanised" the right to claim damages to a significant extent and thereby cleared many obstacles for private plaintiffs (Section II.B above), the private enforcement of European and domestic antitrust law has gained much more importance.The Damages Directive has further facilitated such claims (Section II.C above) that today are brought in many jurisdictions in Europe.Currently, private litigants essentially litigate about the issue of causation and the problem of quantifying the harm caused by a cartel to individual plaintiffs.The latter issue hinges to a large extent on the economic counterfactual and various economic approaches have been developed to assess how the market would have developed but for the anticompetitive conspiracy (Part III above).As the application of these approaches face many difficulties, the right handling of domestic civil procedure rules will determine whether the two European objectives of antitrust damages actions, compensation and prevention (deterrence) (Section II.A above) can be met.In this Article, I have argued that even though the law of damages and evidence is to a large extent based on domestic rules, courts must apply these rules in a manner that is in line with the European objectives of private antitrust enforcements and the European principles of equivalence and effectiveness.Given the economic complexities inherent in determining the counterfactual, courts should accept a reasonable but sufficient degree of precision when assessing damages and should not strive for perfection 'at all costs'.Without reasonable standards the effectiveness of private enforcement will be put in danger (Section IV.A above).
In addition, the disclosure system set out in the Damages Directive should not be overestimated.It will not change the fact that the cartel members will usually have a better understanding of their data and better knowledge about the functioning of the cartel.The disclosure system should therefore not prevent courts from relying on an estimation when quantifying damages in follow-on actions against cartel members (Section IV.B above).
When estimating harm, courts have relied on abstract meta studies as an additional argument to justify their damages assessment.Even though these meta studies should be relied upon with caution, as they might display distorted results, their findings-coupled with other evidence-can be of value when dealing with "small claims" in which costly investment in economic expert opinions would be out of proportion (Section IV.C above).
Finally, courts should bear in mind that the economic uncertainty surrounding the assessment of damages is regularly caused by the cartel members, which also usually possess much more information about the operation and details of the anticompetitive conduct than the claimant.Therefore, uncertainties about the quantum should not play into the hands of the defendants (Section IV.D above).

48
Tráficos, note 45 above, para 49.49On the other hand, the Damages Directive has scaled back the possibilities for plaintiffs to seek disclosure from public authorities.Damages Directive Art 6(6) rules out the disclosure of information regarding leniency applications and settlement submissions completely, whereas the European Court of Justice allowed access under certain conditions ('weighting of interests involved').See Bundeswettbewerbsbehörde v Donau Chemie AG and Others, C-536/11, ECLI:EU:C:2013:366, paras 43-49.50 G Klumpe, T Thiede, and S Dethof, 'Kausaler Schaden' in F Stancke, G Weidenbach, and R Lahme (eds), Kartellrechtliche Schadensersatzklagen (2nd edn, Recht und Wirtschaft 2021), Kap.I paras 94-95.51 The Damages Directive therefore does not prevent a national court from finding that a cartel did not cause harm to individual plaintiffs.See P Van Cleynenbreugel, 'The Presumption of Harm and its Implementation in the Member States' Legal Orders' in M Strand, V Bastidas Venegas, and M C Iacovides (eds), EU Competition Litigation: Transposition and First Experiences of the New Regime (Hart Publishing 2019) 201, p 206. 52 See L Hornkohl, 'The Presumption of Harm in EU Private Enforcement of Competition Law -Effectiveness vs Overenforcement' (2021) 5 EU Comparative Law Issues and Challenges 29, p 36; P van Wijck and F Weber, 'The Abstract Presumption of Harm in the Damages Directive: Overconcern of Overcompensation' (2022) 18 European Competition Journal 204, pp 206-207.