§5.1.2 Implementing comparator-based approaches in practice

(100) Various techniques are available for estimating passing-on related price effects based on the comparator-approach. Certain factors, such as an increase in the raw material costs in the example above, may influence only the comparator market or only the market affected by the passing-on. As explained above, adjustments should be made to the observed data in order to account for such influences. These could be simple adjustments to the data in cases where the influencing factor and the magnitude of its effects can relatively easily be accounted for. (101) In certain cases, when the availability and quality of the data permit, adjustments of comparator data can be made on the basis of econometric techniques, in particular through the use of regression analysis. Regression analysis is a statistical technique which helps to analyse patterns in the relationship between economic variables. (102) In a regression analysis, a number of data observations for the variable under consideration and the likely influencing variables are examined. The relationship identified is usually expressed in the form of an equation. This equation makes it possible to estimate the effects of influencing variables on the variable under consideration and to isolate them from the effects of the infringement. Based on a regression analysis, it is possible to estimate how closely the relevant variables are correlated with each other, which may in some instances be suggestive of a causal influence of one variable on the other (87). (103) The different techniques available for adjustments of comparator data are described in more detail in the Practical Guide (88). By reference to examples and illustrations, the Practical Guide provides guidance on the concepts, approaches and conditions for the application of the different techniques. It should serve as the basis for dealing with issues regarding the approaches available to implement the comparator-based methods. (104) The court should be aware that the use of regression analyses can increase the degree of accuracy of a damages estimate and may thus help in meeting the required standard of proof (89). National courts have dealt with the probative value of statistical evidence, such as regression analyses, in cases where damages have been awarded (90). (105) In general, the probative value of statistical evidence depends on its ability to avoid the risk of so-called type 1 errors, i.e. finding a passing-on effect when none actually exists, and so-called type 2 errors, i.e. failing to find a passing-on effect where there actually is one. In order to deal with these risks, a statistical model may apply the concepts of statistical significance and statistical power. The concept of statistical significance involves setting a high bar for a passing-on effect to be considered existing. An approach widely used is to set a five percent risk for a type 1 error, using a so-called 95 percent confidence interval (91). However, regression models may also reduce the risk of type 2 errors (92). In a specific case where passing-on is estimated, it is for the court to decide whether a regression-model, that has been presented to it, avoids the risks of type 1 and type 2 errors to such an extent that the model has probative value. (106) The potential importance of adjusting the comparator data by using quantitative techniques such as regression analyses can be exemplified by reference to a judgment from a German court, illustrated in Box 8. Box 8 The German car glass case The claimant in this damages action was an indirect purchaser from the members of a car glass cartel. The members of the cartel infringed Article 101 TFEU and were fined by the European Commission in 2008. The claimants' experts carried out an analysis of the price developments before, during and after the cartel period. No regression or correlation analyses were run, rather, the experts aimed at establishing a link between the price of car glass (the cartelised product) and replacement car glass purely by observing the price patterns. The court however considered that this analysis failed to show any sufficiently direct causal link between the pricing of the two products mentioned above. This was particularly because of the data used in the claimants' expert analysis and the fact that it did not take due account of other effects on prices as well as market trends. The court also took into account the market conditions in the case at hand to dismiss the passing-on arguments. (107) However, techniques based on econometric analyses may in certain cases entail considerable costs. In such cases, the court may find it sufficient to estimate the passing-on by simultaneously assessing quantitative data without the use of regression analysis and by taking into consideration qualitative evidence. Moreover, the court may in most cases also find it useful to assess qualitative evidence, such as direct evidence on passing-on, also when employing the quantitative methods described in this section. (108) When estimating passing-on based on qualitative evidence, internal documents describing a firm's pricing policy may be of particular relevance. When assessing internal documents, the court should be aware of the fact that firms in different industries, or even within the same industry, may adopt different pricing policies. In some cases, a firm may have a clear policy or established practice which identifies the price adjustments that will result from specific changes in cost. For example, in some cases purchasers may link price adjustments to changes in certain indices which may not be affected by the infringers' anti-competitive conduct, e.g. consumer price indices. In other cases, purchasers may seek to achieve certain performance objectives, e.g. apply a specific margin to the pricing of the products they supply. In principle, the former policy may speak against the finding of passing-on whereas the later suggests that the purchaser would pass on cost changes. (109) Moreover, when assessing pricing policies, the court should also take into account whether the pricing policy of the relevant firm has actually been implemented, e.g. by considering price data to determine whether these correspond to the pricing policy in question. (110) Examples of cases in which national courts have taken into consideration qualitative evidence are given below. Box 9 Estimating passing-on based on qualitative evidence — Cheminova (2015) In this judgment, the court found that a producer of pesticide had passed on 50 percent of the initial overcharge to the indirect customers. This finding was based on economic theory. In this case, the court could rely on publicly available market studies characterising the market on which the direct customer was active as a monopoly market. In a report provided by the direct customer, it was argued that the market should in fact be characterised as competitive rather than a monopoly. The direct customer inter alia submitted that a large number of products were competing on the market and that moderate market shares indicated a competitive market. However, due to the facts of the specific case, the court disagreed with this approach. Box 10 Estimating passing-on based on qualitative evidence — DOUX Aliments (2014) In this judgment, the court found that the claimant had demonstrated the absence of passing-on. The overcharge in this case concerned lysine, an input into the production of chickens. The court found that lysine only represented one percent of the costs of chicken production. Such a small increase in costs was not sufficient evidence to convince the court that it would also lead to an increase in prices of chicken. The court found that the prices responded to other factors, such as competition with other meat products and buyer power. When concluding that the overcharge was not passed on to the indirect retailers, the court referred to the fact that chickens were sold on an international and competitive market and that grocery retail chains had strong buyer power. (111) When applying comparator-based methods for the estimation of overcharges, courts have also sometimes applied a so-called safety discount. This means they adjusted an amount that would sufficiently take account of uncertainties in the estimate from the observed data (96). If the implementation of econometric analysis is not feasible, such an approach may also be applied for the estimation of passing-on. The objective of such an approach would be to exclude the effects of other possible factors on the variable under consideration, for instance the price offered by the indirect customer.

← 5.1.1.3 · All articles · 5.1.3 →

Source: EUR-Lex (Cellar) · retrieved 2026-09-07