Introduction

Last week, the Court of Justice delivered the long-awaited judgment in the Compass Banca case (C‑646/22). This case has been closely monitored in the EU consumer law community as it deals with one of the key concepts in the field – that of the average consumer. Indeed, under EU consumer law, the lawfulness of traders’ actions and omissions often depends on how they would affect the average consumer. The concept thus shapes the level of protection that the EU acquis affords to buyers and service recipients.

Traditionally, the average consumer has been understood as a person who is “reasonably well-informed and reasonably observant and circumspect”. But what exactly does this mean? Does the average consumer notion approximate a perfectly rational homo economicus or can it accommodate behavioural findings on bounded rationality? In Compass Banca, this question was expressly put forward to the Court of Justice. 

Facts of the case

The case concerned a decision of the Italian consumer authority (Autorita' Garante della Concorrenza e del Mercato, AGCM) against a bank. The authority took issue with the practice of offering personal loans together with partly unrelated insurance products. While it was possible to take out the loan without also buying the insurance, the authority believed that consumers may assume otherwise, due to what was described as “framing”. For this reason, the AGCM argued, consumers may be unfairly led to take out insurance products that they would not consider otherwise. 

To remedy the situation, the authority required the trader to grant consumers a seven-day cooling-off period between the signing of both contracts. The trader disagreed and challenged the decision. Eventually, the Italian Council of State decided to stay the proceedings and refer several questions to the Court of Justice. 

Our focus in this entry is primarily on the average consumer, that is on the first question. We will also touch upon the second question, in which the Court considered if the practice at issue could be deemed aggressive or misleading. We will leave the third and fourth questions, concerned with the powers of national authorities, for another time. 

Judgment

The Court of Justice understood the first question as asking, in essence, whether the concept of an ‘average consumer’, within the meaning of Directive 2005/29 on unfair commercial practices (UCPD), must be defined “not only by reference to a consumer who is reasonably well-informed and reasonably observant and circumspect, but also taking into account the fact that an individual’s decision-making capacity is impaired by constraints, such as cognitive biases” (para. 43). Departing from the preliminary reference, the Court thus stayed clear of the language of ‘homo economicus’ and ‘bounded rationality’.

The Court began its analysis by referring to recital 18 of the UCPD, which elaborates on the average consumer. It reiterated that the concept is “not statistical” (para. 50) but quickly added that it is also not “merely theoretical”. When determining the typical reaction of the average consumer, national courts and authorities should “exercise their own faculty of judgment” but also take account of “considerations that are more realistic” (para. 51). The Court thus appears to have opened the door to a more nuanced understanding of the average consumer and his or her behaviour. At the same time, it made sure to remain within the boundaries of the legal text: the average consumer reality check should be compatible with recital 18. 

In the remaining part of the judgment, the Court focused on what it means to be “reasonably well-informed” as well as “reasonably observant and circumspect”. It confirmed that the reasonable degree of being well-informed is not full information: specifically, the standard only covers information which can reasonably be presumed to be known to any consumer and is not transaction-specific (para. 52). Furthermore, the fact that the average consumer is reasonably observant and circumspect “does not exclude taking into account the influence of cognitive biases on that average consumer” (para. 53).

The Court further pointed to examples from its prior case law that already take such insights on board: 

• case C‑611/14, Canal Digital Danmark, in which it was accepted that an erroneous perception of a piece of information may be suggested to the average consumer;
• joined cases C‑54/17 and C‑55/17, Wind Tre and Vodafone Italia, in which it was acknowledged that, in certain fields, the average consumer may be technically incapable of understanding all the details of any offer in order to make a fully rational choice; and 
• case C-324/97, Lloyd Schuhfabrik Meyer, in which the Court remarked that the average consumer’s level of attention may vary according to the category of goods or services in question.

Overall, the Court concluded that the understanding of the average consumer as a person who is reasonably well-informed and reasonably observant and circumspect “does not exclude the possibility that an individual’s decision-making capacity may be impaired by constraints, such as cognitive biases” (para. 59). Commercial practices that leverage these biases may be deemed unfair, provided that other conditions of the fairness test are fulfilled. In particular, as the Court underscored, it should be verified if a given practice “is of such a kind as to affect the consent of a person who is reasonably well-informed and reasonably observant and circumspect, to such an extent as to materially distort his or her behaviour” (para. 57).

What does the judgment (not) bring?

First, it should be noted that the judgment does not bring a complete reform of the average consumer concept. It remains an objective, not statistical concept, the content of which is for the national courts and authorities to determine, taking into account the case law of the Court. Furthermore, the well-established description of the average consumer holds. We find this unsurprising because, as AG Emiliou has pointed out in his opinion (para. 42, see also previously on this blog), the average consumer is not perfectly well-informed and perfectly observant and circumspect. Indeed, because he or she is reasonably well-informed and reasonably observant and circumspect, there is leeway to change the interpretation of the notion without changing its wording. 

Although the judgment does not bring a complete reform of the average consumer, it opens the door for, what we call, a reality check. According to the Court, more realistic considerations must be taken into account to determine the reaction of the average consumer to a commercial practice, provided that they are compatible with recital 18 of the UCPD (para. 50). What the latter means can be inferred from what the Court considers regarding cognitive biases. First, it should be noted that the Court not only acknowledges that the average consumer could have cognitive biases. In general, it acknowledges that the average consumer’s decision-making capacity can be impaired by constraints, of which cognitive biases are just an example. This, of course, raises the question of what other constraints influence the decision-making of the average consumer. The Court seems to mention some examples from its previous case law (para. 54-56). It will be interesting to see what else can be added.

Cognitive biases

Although the court explicitly opens the door for taking cognitive biases into account, it does not define them. The term ‘cognitive bias’ is, among others, used by the psychologist Daniel Kahneman. Together with others, mostly through controlled experiments, he has mapped several systematic departures of human decision-making from the decision-making of the homo economicus described by rational-choice theory (e.g. Kahneman 2003). To avoid cognitive overload, people use heuristics, which are mental shortcuts or rules of thumb. A cognitive bias is a systematic deviation in human decision-making from the judgment of the homo economicus caused by the use of a heuristic. Some examples of cognitive biases are availability bias, recognition bias and status-quo bias (cf. this overview).

Interestingly, there is controversy around the use of the term cognitive bias. According to the psychologist Gerd Gigerenzer, we should not evaluate human decision-making through economic standards (e.g. Gigerenzer & Gaissmaier 2011). Instead, we have to acknowledge that it is rational to use heuristics, not only because they are fast and frugal but also because they generally lead to accurate results - something Gigerenzer describes as ecological rationality. Admittedly, the heuristics people use are not always well-adapted to their environment, yet the author views those instances as exceptions. In his view, controlled experiments – like the ones conducted by Kahneman – have unjustly painted a different, overly pessimistic picture. 

Considering the above critique, it might have been more neutral if the Court had ruled that the average consumer uses heuristics to make his or her decisions, instead of referring to cognitive biases. This would still imply that consumer decision-making can be flawed if the heuristics used have not been well-adapted to the context. Yet it would avoid implicitly adopting the unrealistic homo economicus as – still – the benchmark for human decision-making. 

To what extent should cognitive biases be taken into account?

That being said, let us consider to what extent cognitive biases should be taken into account according to the Court. This also provides some leads as to the compatibility of the discussed reality check with recital 18 of the UCPD. The Court stressed that not any risk of a cognitive bias necessarily has the effect of materially distorting the behaviour of the average consumer. It must be duly established that a commercial practice “is of such a kind as to affect the consent of a person who is reasonably well-informed and reasonably observant and circumspect, to such an extent that it materially distorts his or her behaviour” (para. 57). What this exactly means remains to be seen, but it is an important point because, although we are in favour of a reality check, we do see a risk of national courts and authorities assuming too easily that a commercial practice unfairly capitalises on a cognitive bias.

Most cognitive biases are established through controlled experiments, which does not mean that they also necessarily have an effect in similar real-life contexts. Illustrative in this regard are two experiments that were conducted after the Mars case (C-470/93). In this much-cited case, the Court ruled that the average consumer is not misled by a bonus package which contains an oversized indication of the “extra” volume in the package. In contrast, a controlled experiment showed that consumers do overestimate the extra volume of a bonus package when confronted with an oversized indication (Purnhagen & Van Herpen 2017). However, in a follow-up experiment, the same researchers found that oversized area indications on bonus packages failed to affect a consumer’s transactional decision (Purnhagen e.a. 2021). 

To some extent, the risk of national courts and authorities assuming too easily that a commercial practice unfairly capitalises on a cognitive bias can also be illustrated using the Compass Banca case itself. The reason for asking the preliminary question was that, according to the AGCM, there could be an influence of framing on the transactional decision. However, it is not clear from the facts of the case what the AGCM understood by framing. In the behavioural sciences, a distinction is drawn between risky choice framing, attribute framing and goal framing (Levin, Schneider & Gaeth 1998). None of these effects seems to be able to directly influence a consumer’s decision due to the practice of offering personal loans together with partly unrelated insurance products. 

One way to prevent national courts and authorities from assuming too easily that a commercial practice unfairly capitalises on a cognitive bias would be to instruct them, under certain circumstances, to ask for expert opinions or to commission empirical research. However, this option, which was mentioned in the Gut Springenheide case (C-210/96) but is currently only used in trademark litigation, is not mentioned in the Compass Banca ruling. In the latter case, the Court only emphasises that national courts and authorities themselves can determine how the average consumer would react to a commercial practice, taking into account the case law of the Court.

Information paradigm

In our view, a better understanding of consumers' cognitive biases would be especially important - and impactful - in the context of Article 5(2) UCPD. However, the Court did not include this provision in its analysis when responding to the second question. Instead, the Court went on to assess the practice of cross-selling personal loans and insurance products under Articles 6, 7 and 8, concerned respectively with misleading and aggressive practices. The Court concluded that the conditions of Article 8 UCPD were not fulfilled but seemed to suggest that the practice may have been deceptive if the trader failed to provide the consumer with “additional information” as to the absence of a link between the two offers (para. 79, 82). The Court did not elaborate on how this piece of information would need to be communicated to eliminate the risk of deception.

Against this background, those hoping that the more behaviourally informed notion of the average consumer would lead to the demise of the information paradigm may find the Compass Banca ruling disappointing. However, we believe that the case at issue was simply not the best proceeding to display what a more realistic picture of consumers’ cognitive capacities adds to the equation. The referring court was not quite clear about the type of problem that required solving. It referred, on the one hand, to “bounded rationality” and “framing” and, on the other hand, to “the fact that a loan applicant is normally in need”, thus hinting at the position of vulnerability (para. 81). Moreover, the preliminary reference failed to include Article 5(2) of the UCPD.

Potential and next steps

Overall, we see the Compass Banca case as an important step towards a more behaviourally informed fairness analysis. It is not the first step, as the Court has already leaned in this direction in its prior case law, e.g. concerning product information (case C-195/14, Teekanne). The Compass Banca ruling confirms that this direction has not been misguided but also cautions against rushed conclusions.

Where the average consumer reality check can be most disruptive is in the context of Article 5(2) UCPD. After all, improving information under Articles 6 and 7 can only take us so far and the Court’s interpretation of aggressive practices is rather restrictive. Article 5(2) UCPD is therefore where most potential lies. What, however, may prevent it from being activated is the lack of tangible evidence about the impact of specific practices. This is where researchers can offer a valuable contribution – as they have done most recently with regard to dark patterns.

Although we see the Compass Banca judgment as an important step towards a more behaviourally informed fairness analysis, in the end, we think that more is needed. To improve consumer protection, more and better exceptions to the average consumer benchmark would have to be recognized. Our first impression is that the Compass Banca ruling provides enough leeway to address, for example, some of the concerns regarding structural asymmetries in digital consumer markets, including dark patterns (cf. Helberger e.a. 2021). However, some might be more susceptible to certain commercial practices than others. As the average consumer remains an objective concept, it is important to have more and better exceptions to it than the current vulnerable and targeted group benchmarks.