Cynthia Rios

“As the Burden Shifts”

The relationship between shifting burdens of proof, presumptions, and error types…

T-Mobile’s Causal Link Presumption

This section examines the presumption of a causal link between a concerted practice and negative effects on market conditions announced in the T-Mobile decision, and its effects on the burden of proof. [1]  The objective is to show how the burden of proof is shifted by this presumption and posit that on a broader level, this presumption may not be in line with the overriding presumption of innocence that the EU adheres to generally.

In order to violate Article 101, there must be a violation by either “object” or “effect.” For a violation by effect, actual anti-competitive effects must be demonstrated as a dispositive element of an Article 101 violation. This differs from a violation by object, since the CJEU has announced that certain actions will be presumed to have a negative effect on the market merely by virtue of their occurrence. A showing of actual negative effects on the market is unnecessary. Violations by object presume negative effects, thereby functioning as a per se rule and potentially making factual evidence irrelevant.

The presumption can be illustrated by comparison to another type of claim. For example, in a tort action, the plaintiff must prove all the elements of the cause of action (duty, breach, injury, and causation). The effect of the presumption of a causal link is akin to a presumption that the existence of a breach of duty likewise establishes the element of causation, without requiring proof. In effect, the burden of proving each element, specifically the causation element, is removed from the party bringing forth the claim and it becomes the defending party’s burden to disprove that presumption. For reasons to be discussed in a separate section, the court states that he presumption of a causal link is a rebuttable one under 101(3). There, however, it is the defending party’s responsibility to produce factual findings of positive effects on the market to defeat the presumption.

Hypothesis Formulation and Explanation

A presumption can be re-stated as a hypothesis in order to test outcomes. A null hypothesis states that there is no association between the predictors and outcome variables. In operation, the null hypothesis (H₀) is: Unilateral disclosure does NOT result in anti-competitive effects. The alternative hypothesis (H_a) is stated to reflect an association between the predictors and the outcome variables. Therefore, the alternative hypothesis is: unilateral disclosure DOES result in anticompetitive effects. The predictor is unilateral disclosure and the outcome is anti-competitive effects.

H₀: Unilateral disclosure does NOT result in anti-competitive effects on the market.

H_a: Unilateral disclosure DOES result in anti-competitive effects on the market.

H₀ in operation:

(1)   Not reject a null hypothesis that is in fact true (true negative).

The null hypothesis is similar to the presumption of innocence. The judgment/inference in “accepting” (failing to reject) a null hypothesis that is true is the result of the “test” and is accurate (not a Type I or Type II error). The presumption of innocence reflects an error preference: that it is better to “accept” (fail to reject) a false null hypothesis than to reject a true null hypothesis.

→ the guilty are convicted

•  Failing to reject (“accepting”) the proposition that there is no association between unilateral disclosure and anti-competitive effects on the market and finding no violation under EU Competition law where there in fact has been no anti-competitive activity.

(2)   Reject a null hypothesis that is in fact false (true positive).

 → the innocent are found not guilty

• Rejecting the proposition that there is no association between unilateral activity and anti-competitive effects on the market and finding a violation where the unilateral activity DOES in fact have anticompetitive effects on the market → there has been anti-competitive activity.

(3)   Reject a null hypothesis which is in fact true (Type I error).

 → the innocent are convicted

• Rejecting the proposition that unilateral disclosure indeed has no anti-competitive effect on the market and finding a violation where unilateral disclosure DOES NOT in fact result in anti-competitive effects on the market → there has been no anti-competitive activity.

(4)   Fail to reject a null hypothesis which is in fact false (Type II error).

 → the guilty is found not guilty and goes free

• Failing to reject the proposition that unilateral disclosure indeed has no anti-competitive effect on the market and finding no violation where unilateral disclosure DOES result in anticompetitive effects on the market → there has been anti-competitive activity.

Burdens

Overview

The burden of proof can refer to the duty on one party to prove or disprove disputed material facts of a case. The burden of proof can include a burden of production, a burden of persuasion, and a tactical burden. The burden of production refers to actual evidence in support of the asserting party’s position (claimant). The burden of persuasion refers to the quantum of actual evidence the party with the burden of proof must adduce to establish the disputed material fact. Finally, the tactical burden is the burden placed on the defendant to counter the evidence proffered against him or her. The tactical burden is created once the proponent of the claim has satisfied his burden of production.

In context, the burden of proof as to every element of a 101(1) violation lies with the Commission. However, in T-Mobile, the CJEU presumes the existence of a dispositive element (the causal link between unilateral disclosure of future pricing of remunerations and anticompetitive effects on the market).[2] Consequently, the Commission no longer bears the initial burden of proof as to this element has been shifted to rest with the defendant. This is different from the tactical burden that is created in rebuttal to evidence proffered against the party.

Indeed, a defendant must show a pro-competitive effect where no proof as to negative effects has been given – which is the case when violation by object is alleged – as an “affirmative” defense of sorts under 101(3). [3] AG Roemer makes this precise point in his opinion for the Grundig and Consten cases:

“[It would be] artifical to apply Article 85(1)[now Article 101(1)], on the basis of purely theoretical considerations, to situations which upon further inspection would reveal no appreciable adverse effects on competition, in order then to grant exemption on the basis of Article 85(3) [now Article 101(3)].”[4]

The burden of persuasion can be otherwise referred to as the standard of proof, i.e., beyond a reasonable doubt (U.S. criminal law), preponderance of the evidence or clear and convincing evidence (U.S. civil litigation). As opposed to the U.S., EU law (or more aptly, the Treaty) does not “constitutionally” mandate the standard of proof. That matter is left to be addressed at a national level among the Member States. This point is summarized in paragraph 94 of the AG’s opinion in the T-Mobile case. [5]

The CJEU did not set or address the standard of proof in its review. As a general matter, EU National courts must establish (1) whether evidence relied on is factually accurate, reliable, and consistent, (2) whether that evidence contains all the information which must be taken into account in order to assess the complex situation, and (3) whether that evidence is capable of sustaining the conclusions drawn from it. [6]

At the very least it is apparent that by presuming the causal link and not requiring proof of actual effects, the CJEU has weakened or “downsized” the standard of proof the Commission and national (enforcing) authorities meet to prevail in a 101 violation by object. [7]In a number of cases, the standard of proof is not equivalent to the standard of proof related to a presumption of innocence traditionally (beyond a reasonable doubt) or indicative of a heightened standard. [8]

In re Error Types

“[C]hanges in the [burden] may have altered the operation of the presumption of innocence, so that the defense now has the burden of proving the reasons for innocence.” [9]

Allocation of the burden of proof recognizes the tradeoff between Type I and Type II errors. Presumptions of a dispositive element of liability operate as per se rules. Per se rules entail more Type I errors. For practices that are almost certainly anticompetitive, the error cost – are arguably acceptable.  The outstanding question’s is how and why certain practices (i.e., unilateral disclosure of information) can be presumed, “by their very nature,” to be truly anticompetitive. In fact, the T-Mobile decision did not specifically reference empirical data to support its presumption (cf. GlaxoSmithKline).  [10]

Moreover, as the cost of Type II errors (letting the guilty go free) cannot be avoided without accepting the cost of Type I errors (convicting the innocent), would it not be helpful in deciding upon the tradeoff to approximate the cost of Type I errors, empirically speaking?

The question also remains as to whether punishing beneficial behavior (Type I errors), takes the behavior out of any possible market self-correction (e.g., the invisible hand)? Furthermore, it is worth considering whether the market auto-corrects uncompetitive behavior? Arguably, “[u]npunished anti-competitive practices, ones that result in higher prices, will only attract new entry.” [11]

The T-Mobile decision arguably broadens the category of information exchange concerted practices that may be found “by their very nature” to be 101 violations by object. Necessarily when the standard of proof is lowered, Type I errors are likely because such a low threshold for finding a violation leaves firms with two options: conducting economic valuation of possible negative vs. positive effects of every opportunity that arises for information exchange among competitors or refraining from any such activity. Naturally, the least expensive option would be preferred and potentially beneficial conduct for the market will not be engaged in by firms. Certainly, consumers do not benefit from that and the question arises of whether this is affects competition law enforcement (that which furthers the “normal functioning of the market”). [12]

---