Measuring and inferring anticompetitive effects from an exchange of information

In four rulings¹ dated 3 November 2025, the Spanish National Court overturned the Decision of the Comisión Nacional de los Mercados y la Competencia (CNMC) in case S/DC/67/17 Tabacos² (‘the Decision’).

These four rulings are significant because they close one of the first cases that the CNMC chose to sanction by effects. Specifically, the Decision considered that an information exchange held between 2008 and 2017 by the three main tobacco manufacturers in Spain (Altadis, Philip Morris and JTI) with the main wholesale distributor (Logista) led to actual restrictive effects.

The CNMC has sanctioned most of the information exchanges following the ‘by object’ route. To that end, the authority has undertaken an analysis of the main characteristics of the exchanges in their economic context, which led it to conclude that the exchanges served no other objective than to be restrictive. Through these analyses, which were essentially qualitative in nature, the authority defined the information exchanged as being inherently strategic. However, this classification was not accompanied by any sort of empirical analysis that could show whether or how the information exchanges had any actual market effect.

The frequent recourse to the by-object approach for the purpose of analysing certain practices, such as information exchanges, means that there is no significant body of cases available that could allow for a proper understanding of the mechanisms through which the sanctioned practices do, in fact, produce restrictive effects.

In this regard, the recent rulings in the tobacco case allow us to address the following three important questions, of a clearly economic nature, that concern the restrictive effects of information exchanges and their treatment under competition law.

• Why, in order to rule on the existence of effects, is it essential to start from a plausible theory of harm that, taking into account the economic and regulatory context, considers how the exchange may have affected the functioning of the market?
• Based on such a plausible theory of harm, how should the effects be quantified by interpreting the available evidence?
• Lastly, what characteristics of an exchange of information are truly useful in assessing the likelihood that it had actual effects on the market?

Before addressing each of these questions, we provide some background to the market and the controversial practice as described in the Decision.

The tobacco market and the contested practice

The tobacco market encompasses three main activities: manufacturing, wholesale distribution or logistics, and retail dispensing. Among the tobacco products marketed, cigarettes are the most sold, and the one in which the controversial practice occurs.

The market is concentrated at the manufacturing level, with the four main manufacturers (the three sanctioned companies plus BAT—British American Tobacco) holding a market share of over 95% during the period when the exchanges took place,³ while the three sanctioned manufacturers accounted for over 85%. Concentration is even greater in wholesale distribution, where Logista had a share of over 99%.

Retail distribution is carried out via a network of tobacconists, under a public monopoly regime that is subject, among other things, to an obligation to distribute all tobacco products under conditions of neutrality.

Manufacturers determine the retail price of their products, which becomes effective once it is published in the Spanish Boletín Oficial del Estado (BOE, the Official State Gazette). A significant proportion (up to 80%) of the final price consists of excise duties. The tax burden increased during the sanctioned period, which coincided with the economic crisis that began in 2008. Once published in the BOE, the price of a product is fixed throughout the retail network, and the network’s remuneration consists of a fixed percentage of the final retail price, with no discounts allowed. There is a second retail channel, consisting of vending machines, which are supplied exclusively by tobacconists.

Tobacco consumption is fairly inelastic to price, although some smokers are more sensitive to price, or have a stronger brand preference, than others. In addition, during the economic crisis the consumption of cigarettes fell while that of other tobacco products (specifically, rolling tobacco) increased.

Manufacturers compete for smokers’ spending by positioning their 200+ cigarette references (i.e. specific cigarette selling presentations), also known as SKUs (stock keeping units),⁴ along a price range of just €1 between the cheapest and the most expensive. Within that range, four segments can be identified. The lowest segment comprises the largest number of SKUs, at a price slightly above the minimum allowed by excise duties. At the other end of the range, the premium segment consists of the SKUs with the highest brand recognition. The low and medium segments sit in the middle of the range.

In addition to price positioning, manufacturers compete by launching new SKUs and through campaigns and promotions. Due to regulatory restrictions, these must be conducted only inside tobacconist outlets and must have been previously communicated to the Tobacco Market Commission (CMT).

The information exchange that was sanctioned consisted of Logista granting access to each of the three manufacturers to data on its sales to the tobacconist network, or ‘sell-in’ data. This data was provided on a daily basis, aggregated at the province level, and covered all of the cigarette SKUs distributed by Logista. It included both the sales of the sanctioned firms and those of other manufacturers that were subscribed to this service.

The CMT also made ‘sell-in’ data on provincial sales publicly available. However, this information had less coverage in terms of cigarette SKUs and was less disaggregated, being set at monthly instead of daily sales.

In addition, Logista sold, and offered as part of its complementary service to manufacturers, ‘sell-out’ data, which referred to daily sales from tobacconists to end-consumers. This data was compiled by Logista itself on the basis of information obtained from a panel of tobacconists. In other words, it consisted of paid-for information on consumer sales, similar to that offered in other mass consumption markets by well-known services such as Nielsen.

The exchange of the sanctioned sell-in data had been taking place since 1999, but the Decision considers that it was restrictive only from 2008 onwards. According to the authority, demand conditions changed during the crisis, when consumption fell sharply, making the access to the information more relevant to the functioning of competition.

The fourth-largest manufacturer, BAT, voluntarily decided to stop receiving sell-in information from its competitors in 2013, and its data was not made available to others from that point onwards. Consequently, its participation in the sanctioned exchange before 2013 was considered time-barred.

Logista also provided the same type of sell-in data in relation to tobacco products other than cigarettes, such as rolling tobacco, but the Decision did not consider that the exchange of sell-in information in that market had restrictive effects.

The importance of starting from a plausible theory of harm

It is well known that information exchanges can be harmful to competition, as they can lead to worse market outcomes. As the European Commission states, the main risk of information exchanges is that they may help to overcome one of the main problems that prevent collusive agreements from being effective. Specifically, they allow participants in such agreements to implement, monitor or enforce collusive arrangements.⁵

This means that certain types of exchange, such as those concerning future plans for prices or quantities, are generally considered restrictive by their very nature and are sanctioned as cartels.⁶

However, as the Commission also explains, information exchanges are a common feature in many markets and can also generate positive effects.⁷ Companies routinely seek and acquire information in order to make decisions. In fact, in this case, Logista itself made sell-out information available to tobacco companies on daily sales by tobacconists to consumers, and the purpose and effects of this were not questioned by the Decision.

In any case, as with any other conduct, its impact on—or restriction of—competition does not occur in a vacuum, and it is therefore fundamental to analyse each practice within its economic and legal context. Therefore, the assessment of how the exchange may have modified the market outcome (for example, by raising prices, reducing supply or degrading quality) must be based on how competition works in the specific market under study.⁸

Such an assessment must start by proposing a plausible hypothesis about how the exchange of information has altered the actual functioning of the market relative to the situation that would have existed if the exchange had not taken place. In other words, the hypothesis or plausible ‘theory of harm’ must be based on a counterfactual analysis, which is essentially qualitative in nature, and considers the characteristics of the exchange in its economic context. This analysis, although qualitative and in terms of plausibility, is the first step that should be taken in order to look for a relationship, if any, between the practice and its effects.

In this case, the Decision had proposed a theory about the impact on competition that the appeal judgement⁹ (the ‘Judgement’) concluded was not compatible with the actual functioning of the market. According to the Decision, the sell-in data would have allowed the three tobacco companies to perfectly identify the market outcomes of pricing decisions, and also to perfectly gauge the success of their commercial decisions, such as the launch of new products or advertising campaigns. This perfect knowledge would have enabled them to counteract any competitive action taken by their competitors and immediately nullify it, which would ultimately, according to the Decision, have removed the incentives to compete, giving rise to a series of restrictive effects that would have been observed in the market (as analysed below).

This theory of harm is explained in the Judgment in its Second Legal Ground when describing the position of the Decision:¹⁰

However, in its Eighth Legal Ground, the Judgment explains how the sell-in information would not have been useful for the manufacturers’ strategic decision-making and, therefore, could not have had the effect on competition that the Decision claimed.

The Judgment explains that, as supported by the expert reports provided by the manufacturers, they compete through three strategic mechanisms: movements in the prices of references, the launch of new products, and temporary promotional campaigns within tobacconists.

A detailed analysis of these mechanisms in their regulatory and economic context shows that sell-in data was not relevant to any of them, for two reasons. First, however disaggregated the data may have been, it concerned Logista’s provincial sales to tobacconists. Yet, as the ruling acknowledges:¹¹

Second, because the launch of new products or new promotional campaigns, as explained in Judgment, is assessed in the medium term, provincial sell-in data did not provide any information that would have made it possible to counter manufacturers’ incentives to continue with such activities:¹²

In short, the Judgment concludes that, although the nature of the information exchanged might, a priori, meet most of the criteria for being considered ‘strategic’, in reality it was not capable of actually affecting the way in which manufacturers compete:¹³

It is clear that if, when putting forward its theory of harm, the Decision had correctly carried out the counterfactual analysis of the way in which sell-in information could affect how competition works in the market, it would have concluded that it had not had the capacity to produce restrictive effects.¹⁴ After reaching that conclusion, any effect that was observed in the market could not have been attributed to the exchange of sell-in data.

This case shows that analysing the ability of an information exchange to produce restrictive effects not only involves analysing the nature of the information exchanged, as part of a mechanical checklist that allows the information to be labelled as ‘strategic’. It also involves analysing the information in its economic context and, in particular, determining whether it alters how competition actually works in the market.

However, in addition to being based on a flawed theory of harm to competition, the Decision did not correctly interpret the effects that it claimed to have identified. This is the subject of the next section.

The importance of interpreting empirical evidence objectively

The next step in testing the existence of restrictive effects¹⁵ involves taking the counterfactual analysis to an additional level, using quantitative techniques. This means the use of economic evidence to assess how the exchange of information led to a series of results in the market that would not have occurred in the absence of the exchange.¹⁶ Various techniques, which are generally statistical in nature, allow for such an informed analysis of causality. They usually involve comparing the situation of the market that is affected by the restrictive practice with another (or others) that is not.

In this case, it is crucial not to forget that the Decision stated that the exchange of sell-ininformation would have had an effect only from 2008 onwards, despite the fact that it had been taking place since at least 1998. The reason given was that, from that point onwards (coinciding with the economic crisis), demand conditions changed and there should have been increased competition, especially in terms of prices across manufacturers.

Under this approach, the Decision considers that there were three proven types of effect, observable from 2008 onwards, through which the restrictive effect of the exchange of sell-in information on the market would have materialised.¹⁷

• First, it considers that there was an abnormal stability in market shares, which did not correspond to what would have been expected. Specifically, it argues that manufacturers with a higher proportion of cheaper products should have gained market share.
• Second, it considers that manufacturers did not compete sufficiently on price, leading to higher prices and price parallelism, which would not have been logical, especially given the tax increases that occurred during the crisis. It also argues that, given the incentives that the collapse in demand would have produced, a greater fall in the prices of the leading, most recognised cigarette references would have been expected.
• Finally, it considers that manufacturers reduced competition by decreasing the intensity of their commercial actions, which was observable in a lower number of new product launches and promotional campaigns.

One problem that counterfactual analysis often faces is the identification or recreation of the scenario unaffected by the controversial practice, the so-called but-for scenario. However, in this case, as explained in the Judgment, the authority had at its disposal three scenarios in which the sell-in data exchange would not have had any effect, which provided the obvious options to rely on when proposing the empirical counterfactual analysis.

The first and most obvious counterfactual scenario was the behaviour of the market before 2008, i.e. before the restrictive effects began to be observed. The second was the behaviour of another closely related market, that of rolling tobacco, in which sell-in data was also exchanged throughout the period analysed. Lastly, the CNMC could have analysed the behaviour of the fourth manufacturer, which had voluntarily decided to stop receiving and, therefore, stop using sell-in information from 2013.

Any of these three scenarios would have provided useful information on the performance of the market and the role of the sell-in data in the absence of the restrictive effects. The fact that these scenarios are not identical to the conditions of the affected market is not necessarily an impediment to their use as comparators, since various analytical mechanisms could allow for possible differences to be taken into account.

The Judgment criticises the Decision for dismissing these three counterfactual scenarios. For this reason, it names its Ninth Legal Ground ‘Absence of a counterfactual analysis in the appealed Decision’, and uses it to justify the validity of these three comparators, especially as, as explained in the Judgment:¹⁸

The ruling concludes that, in the absence of an adequate counterfactual analysis, the Resolution:¹⁹

However, in addition to rejecting the validity of the three counterfactual scenarios, the Decision interprets the data presented in a way that, as described in the Judgment, is also inconsistent with the evidence itself.

Starting with the first of the three effects identified, the Judgment explains how the analysis of market shares offered by the Decision as alleged evidence of abnormal stability can not be read in that way. The Judgment explains that, on the contrary, the market share information presented by the Decision shows a continuous decline in the share of the historical market leader, which can hardly be equated with abnormal stability in market shares:²⁰

With regard to the second effect, concerning abnormal pricing behaviour, the Judgment explains that the alleged parallelism was also observed before 2008 and was entirely reasonable in light of the oligopolistic conditions of the market:²¹

The last of the effects identified by the Decision, the decline in commercial activities, was also inconsistent with the available evidence. Also, as criticised in the Judgment, the Decision did not provide any examples of this alleged loss in commercial aggressiveness, while the available data on the number of campaigns, launches and advertising investment showed the opposite:²²

In short, the analysis of the evidence proposed by the Decision did not allow for the conclusion that the market exhibited anomalous behaviour. The features identified by the Decision as being potentially restrictive could be rationalised by the normal functioning of the market, which was highly conditioned by regulation, its oligopolistic structure and the way by which manufacturers competed. Also, the Decision failed to recognise the true relevance to other competition drivers observed in the market, which also affected its outcomes and were key for manufactures as effective competition tools.

The usefulness of the features of an information exchange in concluding on their possible restrictive effects

This case shows that not all exchanges of information produce restrictive effects, even though some of their essential features might suggest otherwise. The sell-in exchange in the tobacco market sanctioned by the CNMC had many of the characteristics that, as detailed in the European Commission’s Communication on Horizontal Agreements, make an exchange of information more likely to be harmful. The information exchanged referred to quantities sold, was highly disaggregated, and was very recent. Furthermore, the market was highly concentrated, the exchange was carried out by manufacturers that dominated most of the market, and it spanned a long period of time. All of these features would, a priori, be considered to be likely to lead to restrictive effects.

As the case shows, however, the key is not only to analyse the characteristics of the exchange and the market. The most important task is to rigorously justify, taking into account the reality of the market, how that information can change the way in which companies compete and thereby alter the outcome of the market. That is, finely understanding the mechanism through which information leads to negative outcomes is fundamental.

Such a plausibility analysis is particularly useful (indeed, essential) when seeking to infer the existence of effects in cases where the exchange of information has been sanctioned by its anticompetitive object. This is because the two levels of analysis—object and effect—are independent, and the sanctioning of conduct by its object does not necessarily require proof of, or a relation to, the existence of effects.²³

In fact, some of the most notable cases on information exchanges in European Court of Justice case law show that considering them as restrictive information exchanges by their object is compatible with situations in which it is not unlikely that they could have produced no restrictive effects.

One of the most relevant is the T-Mobile case,²⁴ where, through a preliminary ruling, the Court analysed the extent to which a single meeting among the five mobile telephone operators in the Netherlands could be considered a restriction by object. At that meeting, information was exchanged on standard remuneration for distributors of line subscriptions, and the referring national court considered that this was unlikely to have had any effect on the market, particularly on prices. The Court of Justice noted that, in order to determine whether an exchange of information is restrictive by its object, it is not necessary to analyse its effects, and it is not relevant whether there is a connection between the practice and market prices.

Consequently, without a plausibility analysis that puts forward a theory of harm that is compatible with the functioning of the market, it is not possible to make a well-founded inference about the likelihood that an exchange of information, or any other restrictive practice, has produced effects.