Cartel Damages without Economic Experts? The Implications Of The German Federal Supreme Court Trucks III & 4 Decisions
Introduction
Cartel damages proceedings are a highly interdisciplinary endeavour between lawyers and economists in many jurisdictions, including Germany. In such proceedings, lawyers and courts must, among others, answer a fundamentally economical question – namely, what price would have been on a market without the infringement? The European Damages Directive of 2014 also has a goal of avoiding both under- and excessive compensation. The price without an infringement cannot be observed and can only by estimated. As the German Federal Court of Justice (“FCJ”) has stated in its judgments Paper wholesale (para. Liquid Gas I and Paper
Wholesale (para. 67), the goal is ‘to get as close as possible to reality through probability considerations’.The interdisciplinary collaboration is not frictionless, however. German courts have sometimes struggled to understand the economic expertise presented by the parties. Most court judgments have focused on whether or not damages exist, but not the quantum. In Germany, after more than 10 years and a number of follow-up damages proceedings that probably numbers in the four-digits, there are only a few judgments that deal with the quantum of damages. In many cases, economic studies that quantified damages played a key role in settlement negotiations between parties. In cases that went to court, there was a certain amount of uncertainty and frustration regarding how the courts should assess these reports, particularly when there were ‘expert report wars’ with conclusions which were far apart. There is a lively debate about this in the literature. For example, Klumpe/Paha (2024); Schliffke (2024); Kirchhoff (2024),; Klumpe (2024); Tolkmitt (2023); Bornemann/Suderow (21023); Haucap/Heimeshoff (2002) or Schweitzer/Woeste (2002) In Trucks III the FCJ made it (almost impossible) to refute an existing factual assumption of (some) damage. In Trucks IV the FCJ clarified that claimants are not required to submit an economic expert report. They only need to provide evidence that is readily available. This can be as little (or as much) as the summary decision by the cartel authority and, for example, the Oxera (2009)
study which presented data from meta-studies on cartel overcharges to establish that most cartels cause damages.Trucks IV, especially in combination with Trucks III, also has major implications for the work of economists. They might not even be needed anymore. However, as I will explain in the following paragraphs, things may be a bit more complicated. The Trucks III judgment and the inability to prove no damagesThe Trucks IV decision cannot be understood without a consideration of the previous judgment Trucks III as well as other earlier decisions (especially Steel, Abrasives, and Schlecker
). As a starting point, there is a long-established factual presumption of harm in favour of claimants in cartel damages proceedings in Germany (see FCJ judgment Schienenkartell
, para. 55). The defendants are not responsible for proving that there were no damages (see FCJ Schienenkartell judgment V
para. 27). To dismiss a claim, the defendants need to convince the court that, in a weighted average of all the evidence in favor and against damages, there is not enough probability of damages. Over time, it has become clear that rebutting a factual presumption can be difficult. In particular, the Steel Abrasives judgment and the Schlecker judgement, the FCJ held that the mere possibility of an alternative without damages was not enough to rebut a factual presumption. After these judgments, only very robust empirical proof could convince the court that there were no damages. The defendant may find this standard difficult to meet, but it is justified by the fact that most cartels are harmful. 37, 40). The FCJ also ruled that a regression with a statistically non-significant result could not prove damage absence. Statistic insignificance is simply the fact that an expert cannot refute zero with a large enough probability for him/her. Referring to Hurten/2022)
& Inderst/Thomas/2021) the FCJ continues to state that statistically insignificant results can even support statistically significant results (see below). According to the FCJ courts are not required to perform a detailed analysis of defendants regression analysis but only to ‘take an look’ (para. This ruling, I believe (here), has made it virtually impossible to refute the ‘if” of damages. The defendants cannot present a theory that there is no harm, which courts will consider. They can also not do more than a simple regression analysis that produces a statistically non-significant result. Profit-maximizing rational agents will not form ineffective cartels. Nevertheless, there are some cartels that do not cause damages. In economics, this is only possible if or when one or more assumptions are not true in a particular case. Economists are well aware that agents do not always maximize profits. They can also maximize sales or revenues, or take into account social preferences. The cartel members are also unaware of the counterfactual prices. They may be a little conceited and think that the prices would be lower if there was competition. It is unlikely that such explanations are true, but there is a small chance. It is unlikely that a court will accept an alternative explanation, since it cannot be separated from the purely protective claims. In this sense, defendants cannot succeed in providing a theory of no harm.Logically, it is also impossible to positively prove the absence of something. The defendants’ best bet from an empirical perspective would be to present a robust (regression analysis) with a small effect coefficient that is statistically insignificant. In accepted statistical terms, the coefficient is indistinguishable to zero. The literature discusses a specific case when the FCJ mentions insignificant results that may support damages. The FCJ cites articles by Hurten (2002) and Inderst/Thomas (2021) that refer to an Amrhein, et al. article published in Nature. (2019) discuss two studies. One shows a positive impact that is accurately estimated and significant. The other estimates the same effect in quantum but very inaccurately, such that it’s insignificant (as lower bound of so-called confidence range is below zero). In this case, it is important to note that an insignificant result doesn’t imply “no effect”, but rather supports the positive effect found by the other study. The majority of damages proceedings will be different as the defendants analysis will have a lower coefficient than that of the claimants. This lower effect could be so small that it is indistinguishable. According to Trucks III, the courts do not need to take this into account or go into the details of the defendants’ analysis when dealing with the ‘if’. The Trucks IV Decision
The Trucks IV DecisionIn a claim that led to the Trucks IV ruling, the claimant presented a list of allegedly affected trucks, along with proof of purchase. They also referred to the case decision of the European Commission and calculated the alleged damages using the meta-study results shown in Oxera (2009), based on median damages of 18%. In reality, they applied 15%. The claimant failed to submit a case-specific report by an economic expert. The regional court dismissed this claim. The higher regional court dismissed the case, despite stating that it was convinced that damages existed. It also argued that referring to Oxera (2009) is not sufficient to substantiate alleged damages. In Trucks IV the FCJ overturned this decision and referred the matter back to the higher court. The damages can be calculated and may differ from the true value. However, the estimation should be as accurate as it can be. The court may only refuse to estimate minimum damages when there is little or no evidence on which it can rely. If the court has enough evidence to determine the ‘if” of damages, it will have enough evidence to estimate the minimum damages (para. The FCJ continues to mention that damages are typically estimated using comparator methods. The FCJ continues to refer to damages that are typically estimated by applying comparator methods. 19). According to the FCJ however, claimants are not equipped with the necessary data or knowledge to carry out an assessment of damages in this manner. This is particularly true for methods that are otherwise recognized. According to the general requirements of proof (as stated in previous paragraphs), claimants do not have to submit an expert report. If needed, the court may rely on experts. 20). The FCJ concludes that the claimant can use any evidence they have at their disposal. Evidence can include damage assessments made using the standard method, but it is also possible to use other evidence – such as the case decision. 21).Therefore, the FCJ ruled that the higher regional court set the standard of proof too high, and that it erred in assuming that the claimant must put forward a damages assessment based on a comparator method. This type of analysis can only be done by an economic expert and the claimant does not have to hire one. If the court feels it needs an independent external expert report, they can request one themselves (para. 24). The FCJ also states that the argument of the higher regional court, that Oxera (2009) does not support damages at 15% or other minimum damages is incorrect. The court concluded that trucks cartel did not fall under the 7% of cartels that are ineffective. According to the FCJ, this conclusion is either contradictory, or misunderstands what the court’s freedom is in estimating (minimum damages). Damages without experts after Trucks IV? Consider a claimant who submits their claim with the value of commerce, the summary decision from the authority and the Oxera 2009 study but no other evidence. The defendants may also submit an economic expert report that contains (i) a section disputing the Airtours criteria for a stable monopoly and arguing there have been shifts of market shares that indicate competition and (ii), an econometric study showing damages of only 3% but statistically insignificant (by Trucks III and previous decisions);
the general arguments in the defendants’ expert report are dismissed as they only establish the possibility of no damages (by Trucks III and previous decisions); A potential course of action runs as follows:
the claim is not dismissed for not being substantiated (by Trucks IV);
the general arguments in the defendants’ expert report are dismissed as they only establish the possibility of no damages (by Trucks III and previous decisions);the regression analysis by the defendants is dismissed, even without going into detail, as it does not positively prove the absence of damages (by Trucks III’);
if the court concludes that there are damages, since the evidence by the defendants is not enough to outweigh the factual presumption, then it also has enough evidence to estimate at least minimum damages (by Trucks IV).
The above reasoning follows directly from Trucks III and Trucks IV. One important aspect is overlooked. The defendant’s expert’s opinion does not only make a statement on the ‘if’ but also a statement on the quantum. This quantum statement, for example, is completely independent of the question of statistical significance. 3%, is also completely independent of the question of statistical significance.
The quantum statement in the expert report is an indication which must be included in the overall weighting of all the evidence presented, assuming that such an overall weighting is also necessary regarding the quantum (see judgment Schienenkartell II
(para. 36)). This statement, to my understanding, also applies to the determination of the quantum. In my understanding, this statement also applies to the determination of the quantum.
There is a considerable qualitative difference between an empirical expert report tailored to the specific individual case, if it is sound, and a meta-study on cartel damages that is completely detached from the individual case. Oxera (2009 p.90) states that damages from cartels with a high degree of complexity are highly variable. There is also no reliable evidence that the proportion of violations with minor damage decreases noticeably the longer a cartel lasts or the greater the market coverage, as pointed out by Hellmann/Schliffke (2022)
.
Intuitively, the case-specific expert opinion should have a much greater weight and outperform the meta-study in the overall weighting of evidence. To accurately determine the evidence weighting, the court needs to be convinced that the defendants’ empirical reports are sound. The court must examine the expert report to determine if it meets the criteria outlined in the FCJ’s judgment in Trucks II
– i.e., that it is based upon a reasonable data base, is methodologically accurate, and has reliable results. 36). The court can only take evidence that is potentially relevant to the decision. It is therefore conceivable that a court assumes without examination that the (insignificant) 3% presented by the defendants in the example is correct, but nevertheless concludes that there are damages, and that those are greater than 3% based on the fact that most cartels cause higher damages (as shown by the Oxera (2009) study) and an assessment of the severity of the conduct as described in the case decision.
Therefore, there is now a potential way of handing down a quantum judgment without ever going into the details of an empirical analysis put forward by the defendants, and the claimants need not put forward such an analysis either.
- What remains unclear is whether and to what extent a court could go above the 3% from the example. To avoid having to deal with the regression analysis, the court will have to accept that it is correct. This will result in a situation in which the only evidence for the quantum is a meta-study. If this limits the quantum the court can potentially rule on, defendants will still be motivated to submit regression analyses. To counter the likely low estimate of damages put forward by defendants, claimants will almost certainly offer an expert opinion to prove that damages are in fact higher than the defendants’ view. This offer of evidence could force courts to commission a report from an independent court expert, since there is a significant difference between, say, 3% and 15%, and determining which estimate is closest to the actual damage requires independent economic expertise. It is unclear whether there is a threshold below which a judge can accept the defendants’ analysis as mathematically correct but dismiss it as implausible, thereby avoiding being bound by it, and possibly avoiding the need to appoint court experts. These questions will only be answered by subsequent decisions. As of yet, the exact nature of this impact is unknown. One possible future scenario is that defendants will try to submit the lowest number possible that the court won’t dismiss, and claimants will attempt to obtain some (maybe even very small) minimal damages without presenting expert evidence. This could lead to courts only obtaining evidence based on strategic considerations. In this scenario, no one would be trying to estimate the actual damage caused in the case at hand. This result is strange when compared to the FCJ’s goal of getting as close to reality as possible through probability considerations. 15). It is also strange in the context of the EU Damages Directive, which requires that compensation be accurate and not under- or overcompensated. Disclaimer: The author has worked as an economist on numerous cartel damages proceedings for both claimants and defendants. He was an expert for claimants before joining Oxera 2023. The views expressed here are those of the author, and not necessarily Oxera Consulting LLP.