4 - The Boundaries of the Firm and the Reach of Competition Law

Summary

While company law makes great efforts to maintain the separation of different legal entities, other areas of law increasingly emphasise the common responsibility of corporate groups. One of the fields shaping this emerging principle is competition law, where focusing on the whole group rather than its individual members is increasingly becoming the norm. But this approach is still far from being uniformly accepted. While EU competition law is pushing ahead, US antitrust law is said to take a rather critical stance. Against this background, this chapter examines the functions performed by a unitary perspective on corporate groups. The main goal is to show that at least three important functions must be distinguished in EU competition law, each of which has its own implications. Only when these are properly understood can it be determined where the group perspective is appropriate and where it is not. This is shown by the example of current discussions, for example, on the liability of sister companies and on possible applications in the area of liability for damages. Finally, the analysis in this chapter also aims to contribute to a better understanding of the different approaches in EU and US competition law.

4.1 Introduction

It is an essential feature of modern company law that the shareholders are not personally liable for the debts of the company. This form of asset partitioning, often referred to as ‘limited liability’, fulfils a number of important functions.¹ Any exception to the rule is therefore controversially discussed. In recent times, however, the number of exceptions seems to be increasing. An important example can be found in EU competition law. Under the so-called ‘single economic entity doctrine’, parent companies can be held liable for competition law infringements by their subsidiaries. The doctrine only applies to controlling shareholders or, in the words of the EU Courts, to shareholders who exercised a decisive influence over the conduct of a subsidiary at the time when the subsidiary infringed the competition rules.² The application of the doctrine, which is generally considered to have no equivalent in US antitrust law, does not presuppose that the parent company itself has done anything wrong. It does not matter whether the parent company knew about the subsidiary’s infringement and tolerated it. Nor does it matter whether the parent could have prevented the infringement. The liability of the parent company is based on structure, not behaviour, and is justified above all by the possibility of control.

There is of course no doubt that European company law – like all advanced commercial law systems – recognises the corporate law principle of limited shareholder liability. The European Court of Justice (ECJ) has even ruled that shareholder liability may, under certain conditions, infringe Article 63 of the Treaty on the Functioning of the European Union (TFEU).³ That provision governs the free movement of capital, a central element of the European internal market, itself the economic backbone of the European Union. However, the ruling of the Court related primarily to non-controlling shareholders. Moreover, the ECJ has explained that, while limited shareholder liability seems to be accepted as the general rule,⁴ ‘it cannot be concluded therefrom that this is a general principle of company law applicable in all circumstances and without exception’.⁵ It seems that the liability of parent companies for antitrust fines is precisely one of these exceptions. In any event, the ECJ has clearly rejected all attempts to challenge the antitrust liability of parent companies by invoking the principle of limited shareholder liability.⁶ Although the Court’s arguments in this context often sound very formalistic, it seems obvious that the main reason for rejecting such attacks on parent company liability simply is that the judges give greater weight to the objectives of the single economic entity doctrine than to the principle of limited liability, at least in the specific setting of competition law offences.

It is therefore regrettable that the various functions of the single economic entity doctrine are still not properly appreciated. Although the liability of parent companies for infringements by subsidiaries has been recognised in European competition law for decades, much of the literature continues to lose itself in fundamental criticism.⁷ As a result, a thorough examination of the functions of the doctrine and its essential components is still lacking. Against this background, the purpose of this chapter is to examine in more detail the contexts in which the European Commission and the EU Courts invoke the single economic entity doctrine. As we shall see, the use of the doctrine is more nuanced than is generally thought. In particular, the doctrine is not only used to establish the liability of parent companies but also serves important functions in the calculation of fines and their enforcement. Only when these functions as well are appreciated is it possible to make informed statements about the proper reach of the doctrine. It is important to emphasise this because various extensions of the doctrine are currently being discussed. Unfortunately, it often remains unclear what purposes they would actually serve.

Section 4.2 starts with a brief look at the conceptual background of the single economic entity doctrine, which will allow us to better understand the approach of the courts. Section 4.3 then explores three distinct functions of the doctrine: to induce parent companies to control the conduct of their subsidiaries, to ensure the correct calculation of fines and to ensure that fines are actually paid. All three purposes will be analysed in detail and it will become clear that it is important to distinguish the different functions, as they each have their own implications. Section 4.4 discusses borderline questions of group liability: whether companies other than parent companies should also be liable for infringements committed by affiliated companies and whether it can be justified to hold parent companies liable not only for fines but also for damages, i.e. claims brought by private actions. Section 4.5 then broadens the view and asks how the problems solved in the EU by the single economic entity doctrine are addressed in US antitrust law. It will become clear that while on paper there is indeed no liability of parent companies for antitrust infringements by their subsidiaries in the US, the reality is more complex. Section 4.6 concludes.

4.2 Legal and Economic Entities

The starting point for all discussions on parental liability in the case law of the EU Courts is the concept of an undertaking. According to a widely used formulation, that notion ‘must be understood as designating an economic entity, even if, from a legal perspective, that unit is made up of a number of natural or legal persons’.⁸ That phrase was first used by the ECJ in Hydrotherm (1985) to explain that competition is ‘impossible’ between a legal person and its sole owner, as they necessarily have identical interests and act jointly on the market.⁹ The variety of forms offered by company law is simply not relevant for competition law purposes and the same applies to the exact distinction and delineation of legal entities. The EU Courts have held since ICI (1972) that the formal separation between two companies with different legal identities is not decisive for applying the competition rules.¹⁰ If one company controls the other and they therefore behave uniformly in the market, liability may be attributed to the controlling company.¹¹ For the purposes of applying the competition rules, it is the unity of those companies’ conduct on the market that matters, and not their separate legal personality. Similarly, the Courts have always rejected the idea – repeatedly put forward by companies and their lawyers¹² – that the corporate law principle of limited shareholder liability could have any meaning in this context. Instead, insisting on different legal personalities and separate assets is seen as a misplaced form-based approach that ignores economic realities and does not meet the needs of competition law enforcement.¹³

The Courts have even gone so far as to describe the joint and several liabilities of multiple legal entities constituting a single economic unit as ‘merely the manifestation of an ipso jure legal effect of the concept of an “undertaking”’.¹⁴ However, it has rightly been pointed out that the Courts do not automatically hold liable all entities that act jointly in the market as an economic unit.¹⁵ At least so far, the focus has been on the liability of parent companies. This liability is not primarily justified by the uniform market behaviour of the undertaking, but by the parent company’s control over the subsidiary. According to the case law, the conduct of a subsidiary may be imputed to the parent company ‘where, although having a separate legal personality, that subsidiary does not decide independently upon its own conduct on the market, but carries out, in all material respects, the instructions given to it by the parent company’.¹⁶ A parent company will only be liable for offences committed by its subsidiary if the parent is in a position to exercise a decisive influence over the commercial policy of the subsidiary and does in fact exercise that influence.¹⁷ This is presumed, however, for the specific case that the parent company holds all or almost all shares in the subsidiary.¹⁸ It is then up to the parent company to rebut the presumption of control by adducing sufficient evidence to show that the subsidiary acted independently on the market at the time it committed the competition law infringement.

Ironically, in recent years, an almost form-based use of the concept of an undertaking has emerged. While initially the instrumental approach was strongly emphasised, in recent years, the Courts have sometimes tried to find meaning in words instead of exploring the relevant problem. In particular, the substantive examination has occasionally been replaced by a reference to established definitions and (alleged) precedents.¹⁹ Moreover, the Courts have begun to treat the economic unit as akin to a legal entity, sometimes even seeming to endow it with legal personality. The apparent de facto personification of the economic unit seems to be linked to the growing number of arguments before the Courts based on fundamental procedural rights. For example, the Courts have countered the argument that parental liability infringes the principle of personal responsibility with the claim that this principle only applies to the economic unit, and they have made the same argument for the principle that the penalty must be specific to the offender and the offence. The ECJ held in Siemens Österreich (2014) that these principles ‘relate only to the undertaking per se, not the natural or legal persons forming part of the undertaking’.²⁰ Similarly, the General Court (GC) explained in Nynäs (2012) that parental liability ‘does not in any way constitute an exception to the principle of personal responsibility, but is the expression of that very principle’, as the parent company and its subsidiaries form a single undertaking and are thus collectively responsible.²¹ This reasoning has been rightly criticised in the literature.²² The undertaking is not a legal entity and therefore cannot be a bearer of rights and duties, let alone of fundamental rights.²³ To equip the undertaking with legal personality would confuse the economic and the legal level and would inevitably lead to misconceptions. The notion of undertaking is merely a legal mechanism by which certain objectives can be achieved – attribution, parental liability – and the concept of a single economic unit helps to explain why these objectives are appropriate. But a simple reference to the economic unit can never replace substantive arguments and it should not be used to disguise the solution of legal problems.

Even if the approach of personifying the undertaking de facto is not convincing, it is understandable that the EU Courts try to counter the often fundamental, uncompromising criticism by companies and their lawyers of the single economic entity doctrine with similarly heavy artillery. And apparently, the Courts believe that it would be more difficult to justify parental liability as a form of vicarious liability. Ultimately, however, the concept of the Courts will only gain acceptance if it goes beyond a mere interpretation of words. If the Courts really want to convince, and maybe even provide an example for other jurisdictions, they must show that their approach is good policy. This is actually not that difficult. As I will show in the following section, most of what the Courts do on the basis of the single economic entity doctrine, is justified by good reasons. However, it will also become obvious that it is important to clearly distinguish between the different purposes that the unitary perspective on corporate groups is intended to serve.

4.3 Three Distinct Functions

Much has been written about the concept of undertaking and its implications for the liability of corporate groups, but the exact functions of the single economic entity doctrine continue to be underexplored. Most of the debate still is concerned with the legitimacy of the doctrine, often ignoring that legitimacy cannot be separated from purpose. A good starting point for the question what purpose parental liability serves is provided by Advocate General (AG) Kokott in her opinion in Akzo Nobel (2009).²⁴ She relied on four main arguments:

Where […] a parent company exercises decisive influence over its subsidiaries […] it accords with [1.] the principle of personal responsibility and with [2.] the objectiveof effective enforcement of the competition rules to hold all the companies of the group […] jointly and severally liable […]. Only in that way can it also be ensured that, when assessing the amount of a fine to be imposed, [3.] the true economic strength of the whole undertaking is correctly taken into account and that [4.] the successful enforcement of the fine is not jeopardised by any transfers of assets between the parent company and its subsidiaries (enumeration and emphasis added).²⁵

In Section 4.2, above, I have already explained that I do not consider it appropriate to apply the principle of personal responsibility to economic units. The other three points, however, are as correct as they are important and they certainly deserve more attention – both by the courts and in the literature. They are discussed in more detail below.

4.3.1 Controlling the Conduct of the Subsidiary

Promoting the effective enforcement of competition law is probably the most important objective of the single economic unit theory. But how exactly does the liability of parent companies contribute to this objective? The clearest statement by one of the EU Courts on this point is to be found in the GC’s judgment in Dow Chemical (2012). In that decision, the GC stated that ‘as a result of the parent company’s power of supervision, the parent company has a responsibility to ensure that its subsidiary complies with the competition rules’. The Court further explained that ‘[a]n undertaking which has the possibility of exercising decisive influence over the business strategy of its subsidiary’ may be presumed ‘to have the possibility of establishing a policy aimed at compliance with competition law and to take all necessary and appropriate measures to supervise the subsidiary’s commercial management’.²⁶ Unfortunately, the GC made these statements in a case in which the parent company’s control was not self-evident – the case concerned a 50/50 joint venture. On appeal, the ECJ did not support the GC’s view, but instead maintained that the latter had pondered about the parent’s responsibility ‘purely for the sake of completeness’ and that the statement was therefore not contestable by legal means.²⁷ Regrettably, the GC did not see this as an encouragement for any further reflection of this kind.

But the GC was right. As I have explained elsewhere in more detail,²⁸ holding parent companies liable for competition law infringements by their subsidiaries serves a number of important objectives. Three reasons should be highlighted here.

First, parental liability restores effective deterrence where subsidiaries are underdeterred, for example, because they lack sufficient assets to pay a fine or because they misjudge the situation on the basis of insufficient information.²⁹ From a deterrence perspective, it can make sense to involve the parent company in compliance efforts, if the subsidiary does not respond to the threat of monetary sanctions. However, two conditions should be met: First, it must be likely that the parent company will respond better to incentives than the subsidiary, for example, because the parent has more assets or better information. Second, the parent company must be able to influence the behaviour of the subsidiary and deter it from infringing the law. Typically, both will be the case. Parent companies often possess more assets and better information, and they are in a good position to steer their subsidiaries in the right direction. They can select and replace their subsidiaries’ management, they can establish group-wide compliance mechanisms, and they can set up compensation and promotion schemes that reward compliance and discourage any form of illegal behaviour. In this respect, holding parent companies liable for their subsidiaries’ offences serves a similar function to other forms of vicarious liability, such as the liability of employers for wrongs committed by their employees or the liability of parents for the behaviour of their children.

A second reason for parental liability is that it prevents parent companies from opportunistically exploiting limited liability to separate their assets from their risks.³⁰ It is well known from corporate law and economics research that limited liability may induce shareholders to externalise risks to third parties, in particular to involuntary creditors who cannot insist on contractual protections.³¹ A strategic use of limited liability allows shareholders to fully benefit from the opportunities for profit, but to disassociate themselves from the company in the event of losses. Such incentives may also exist with regard to infringements of competition law. For example, companies could be tempted to bundle their sales activities in markets particularly susceptible to cartels in weakly financed subsidiaries. If an infringement of competition law were to occur, the damage could then be contained to the subsidiary, while the parent company would remain unaffected. Extending liability to the parent company undermines such strategies. It eliminates the incentive for opportunistically exploiting limited liability and re-internalises all competition law risks to the corporate group and its ultimate shareholders.

The third reason for parental liability is that enforcement is expensive and that parent companies often have a comparative cost advantage over the state.³² Many of the most serious competition infringements take place in secret. They are difficult to detect and investigate. As a result, competition authorities have to devote considerable resources to ensure that infringements are established in a way that withstands judicial review. High costs of enforcement are an important reason why competition law relies heavily on cooperative measures such as leniency programmes and settlements. The liability of parent companies can also be seen as a means to save enforcement costs. Parent companies are in a good position to take over part of the investigation work that would otherwise have to be carried out by authorities. They can obtain reports from their subsidiaries, examine contracts and accounts and, if necessary, conduct internal investigations. As internal control mechanisms usually already exist for the purpose of group management, parent companies can typically implement investigative measures at a lower cost than the state. Given the greater cost-effectiveness of internal reviews, this not only turns public costs into private costs but saves administrative costs overall.

Occasionally, it is claimed that the actual design of the single economic entity doctrine does not fit these policy objectives. For example, it is argued that the doctrine does not reward compliance efforts but punishes them because the parent company is liable even if it does everything in its power to prevent the subsidiary from committing competition law violations.³³ Moreover, it has been questioned why the parent company is also liable if the subsidiary is solvent³⁴ and why the application of the doctrine does not depend on whether the parent could actually have prevented the infringement.³⁵ However, the strongest incentive to commit to compliance comes from the possibility of avoiding liability altogether. The parent company is rewarded for its compliance efforts by not being liable if no infringement occurs. If the parent company succeeds in deterring the subsidiary from committing infringements, the question of liability does not arise at all – neither for the parent nor for the subsidiary. Furthermore, the same questions could be asked about the liability of employers under the doctrine of respondeat superior. Both concepts cannot be reduced to a single objective and they are based on fairly general conditions to cope with a large number of different cases. The fact that efficient deterrence can be achieved in individual cases even without vicarious liability does not mean that vicarious liability as a whole is not important for deterrence. Hardly any doctrine does justice to every individual case. It would be possible to differentiate the conditions for the liability of parent companies to a greater extent, but legal certainty would suffer as well. Given this trade-off, it seems justifiable that the doctrine is based on rather general conditions. Moreover, in accordance with the case law of the EU Courts, the Commission has discretion as to whether or not to invoke the parent company’s liability.³⁶ Where the application of the doctrine does not contribute to general or specific deterrence, it can therefore remain unapplied.

4.3.2 Ensuring the Correct Calculation of the Fine

Another objective of the single economic entity doctrine is to enable a proper calculation of fines. As we shall see, it is very important to distinguish this function from the one discussed in the previous section. In case law, however, both functions are sometimes confused. The starting point for all consideration is that the benchmark for the calculation of fines is the objective of deterrence. The Courts have repeatedly stressed that the Commission must ensure that the fine has the necessary deterrent effect.³⁷ They have also explained that, as far as the specific deterrence of the infringing undertaking is concerned, the deterrent effect must be assessed in relation to the size and the economic power of the undertaking, for which it is necessary to take into account its global resources.³⁸ As pointed out by AG Kokott in her statement quoted above, in this context, it is important to correctly assess the ‘true economic strength of the whole undertaking’, i.e. the corporate group. The Courts, therefore, assume that – in so far as turnover is relevant for the calculation of the fine – it is the consolidated turnover of the group as a whole that must be taken into account. In Group Gascogne (2013),³⁹ the ECJ explicitly referred to Article 1(1) of Directive 83/349⁴⁰ for the purpose of determining the 10% turnover cap according to Article 23(2) of Regulation 1/2003⁴¹ and paragraph 32 of the 2006 Fining Guidelines.⁴² Today, the relevant provision is to be found in Article 22(1) of the Accounting Directive 2013/34,⁴³ which, like its predecessor, obliges parent companies and subsidiaries to prepare consolidated financial statements. The aim is to provide ‘a true and fair view of assets and liabilities, the financial position and the profit and loss’ of the whole group of companies.⁴⁴ It should be noted, however, that the conditions for determining ‘parent undertakings’ and ‘subsidiary undertakings’ within the meaning of Article 22(1) of the Accounting Directive are set out exclusively in that provision. The criteria listed there are similar to those used in EU competition law to determine economic units – but they are not identical. It is therefore not necessarily the competition law concept of an undertaking that decides on the attribution of turnover.

The conclusion that only consolidated group turnover can be decisive in so far as turnover is relevant for the calculation of the fine is already apparent from the fact that turnover can easily be reallocated within the group. Competition law investigations take a long time, so companies usually know that a fine will be imposed well in advance of the actual imposition of the fine. If the managers of a corporate group could be sure that only the turnover of the subsidiary will be taken into account, they could adapt to this and shift the turnover to other companies in the group. This is particularly true where the Commission does not normally take into account the turnover during the cartel infringement but, as in the case of the 10% turnover cap, the turnover in the year preceding the prohibition decision. At this stage, the company is usually aware that a fine will soon be imposed. Against this background, it is surprising that the ECJ apparently wants to limit the relevance of consolidated group turnover to cases in which liability is attributed to the parent company. In the Groupe Gascogne judgment, at least, the Court held that the Commission is entitled to rely on group turnover where it ‘has established to a sufficient legal standard that an infringement may be attributed to a company which heads a group’.⁴⁵ If that was meant as a condition, the Court would be wrong. For the reasons set out above, it must always be the consolidated group turnover that is taken into account, even if only a single subsidiary is liable. Interestingly, the ECJ recognises that the Commission cannot be required to demonstrate the decisive influence of the parent company for each subsidiary whose turnover it wishes to include in the calculation of the fine, as these are ‘two separate issues serving different purposes’.⁴⁶ This observation is correct and the conclusion must ultimately be that the calculation of the fine should not depend at all on the economic unit as developed for the attribution of liability. Instead, the Commission should simply always be allowed to rely on consolidated accounts as defined in Chapter 6 of the Accounting Directive.

There is an important difference between the liability function of the single economic entity doctrine and its use for the correct calculation of fines, and that is the point in time that is decisive. The liability function is linked to the parent company’s control and its ability to prevent competition law infringement. Liability based on this ground should therefore apply only to corporate entities which were able to exercise a decisive influence during the infringement. For the calculation of the fine, however, the relevant point in time should in principle be the date of the prohibition decision. The reason is that fines serve primarily as a deterrent and thus have a forward-looking purpose. In particular, the objective of specific deterrence, i.e. ensuring compliance with the undertaking addressed by the prohibition decision, requires the fine to be set in such a way as to anticipate the undertaking’s future behaviour. This is an important finding because the undertaking at the time of the decision may be quite different from the one at the time of the infringement.⁴⁷ The objective of setting the fine in such a way as to prevent future infringements may even justify taking into account the economic strength of entities which were not part of the undertaking at the time of the infringement. In contrast to the liability function, it is irrelevant for the calculation of the fine whether other members of the group could have prevented the competition law infringement. Instead, it is simply a matter of correctly taking into account the actual economic strength of the group when calculating the fine.

4.3.3 Ensuring the Payment of the Fine

Another important objective of the single economic entity doctrine, derived from the case law of the EU Courts, is to ensure that fines are actually paid. Even though the Courts adopt an approach which comes close to personifying the economic unit formed by a group of companies,⁴⁸ they accept that the Commission can only address its decisions to natural and legal persons.⁴⁹ This is the only way to ensure that decisions imposing fines are enforceable under Article 299 TFEU. In this context, the ECJ has described the joint and several liabilities of parent companies and their subsidiaries as ‘an additional legal device available to the Commission to strengthen the effectiveness of the action taken by it for the recovery of fines imposed for infringement of the competition rules’.⁵⁰ Moreover, the Court has explained that extending liability to the parent company ‘reduces for the Commission, as creditor of the debt represented by such fines, the risk of insolvency […]’.⁵¹ On other occasions, the ECJ has emphasised that joint and several liability ‘cannot be reduced to a type of security provided by the parent company in order to guarantee payment of the fine imposed on the subsidiary’.⁵² Thus, the Court has clarified that securing the enforcement of the fine is one purpose of parental liability, albeit not the only one.

Since fines for competition offences are often very high, it is not surprising that companies go to great lengths to avoid them. Past experience has shown that this sometimes includes corporate restructuring. A fairly well-known example that happened in Germany is discussed in detail in Chapter 5 of this book.⁵³ In this case, the Bundeskartellamt had to close its proceedings because a restructuring had made it impossible to enforce the fines. This case was one of the reasons why the German legislator decided to introduce parental liability based on the European model as part of the Ninth Amendment to the German Competition Act.⁵⁴ As far as the particular function of ensuring the enforcement of fines is concerned, parent company liability competes with other approaches, such as the liability of legal and economic successors or simply faster enforcement of fines. However, it is clear that it becomes more difficult and therefore less attractive to avoid fines through restructuring if liability is also extended to the parent company.

The Courts have in the past linked the problem of collecting the fine with the objective of deterrence.⁵⁵ Such a connection does indeed exist since a fine which is not collected cannot contribute to deterrence. According to this logic, if the subsidiary defaults on payment, liability should be extended only to those who could have prevented the infringement, i.e. the parent company at the time of the infringement (or several successive parent companies if there has been a change of control during the period of the infringement). However, as illustrated by successor liability,⁵⁶ deterrence is not the only possible objective of such a form of contingent liability. It could also be seen as an attempt to ensure that a large financial claim by the Commission is met in order to protect the EU budget and ultimately European taxpayers. Seen in this light, there would not even need to be a link between the party liable and the competition law infringement. Instead, the substitute debtor could be anyone who held a significant stake in the subsidiary at some point between the beginning of the infringement (when the debt was first incurred) and the payment of the fine (when it is fulfilled). If the main objective is to avoid subsequent restructuring, it makes sense to focus on a point in time before the investigations become known.

To be clear, such liability, which is not linked to the control during the infringement, has not yet been established under EU competition law.⁵⁷ Yet, that does not mean that it could not exist.⁵⁸ First steps in this direction can be found at the Member State level. Germany has created a form of substitute liability in 2017, as part of its strategy to close loopholes in its former liability law. Under Section 81a of the German Competition Act,⁵⁹ a sum equal to the fine can now be demanded from parent companies as well as legal and economic successors if the fine cannot be enforced against a subsidiary or predecessor responsible for an infringement.⁶⁰ This type of liability has the advantage that it does not contain any accusation regarding the infringement – the substitute debtor need not have been affiliated with the subsidiary at the time of the infringement – so that the procedural guarantees of the law on fines need not be applied. For this reason, the German legislator suggests, for example, that the provision can also be applied retroactively, even though the law on fines is characterised by the principle that acts can only be punished if they were prohibited before they were committed.⁶¹ In fact, the same approach could be used for other high liabilities, such as tax debts – without it being relevant whether any of the companies have infringed the law. Such liability would not be a sanction but would be justified solely by the need to prevent evasive conduct aimed at circumventing high payment obligations.

4.4 Borderline Issues of Liability

After these remarks on the distinct functions that the single economic entity doctrine fulfils in EU competition law, it is now time to reflect on the recent debates on extensions of the doctrine. As we have seen, it is important to distinguish between the question of the calculation of fines and that of liability. The fact that, for example, the turnover of another company should be taken into account in the calculation of the fine does not mean that this company should also be liable. While the calculation of fines depends on a correct assessment of the economic strength of all affiliated companies at the time of the prohibition decision, attribution of liability should be concerned with the question of who could have prevented the competition law infringement at the time when it occurred. This difference must also be kept in mind in the following considerations. It means, in particular, that, at least as a general rule, a much narrower approach is required with regard to liability than with regard to the calculation of fines.

4.4.1 Sibling Liability

The first issue to be looked at is the liability of sister companies. It has been claimed in the literature that the liability of sister companies already follows from the concept of the single economic unit.⁶² Since sister companies form an economic unit with the parent company and with each other, they shall, according to this view, be mutually liable for their respective competition law infringements. This position is closely linked to the view that it is the economic unit as such which commits the infringement and that the infringement can therefore be attributed to all the legal entities constituting the economic unit, i.e. essentially all members of the corporate group.

The Courts have so far acknowledged the liability of a sister company only in the rare case where one sister company exercises a decisive influence over the other.⁶³ For other constellations, the case law is not yet clear. The ECJ has held in Aristrain (2003), that it is not possible ‘to impute to a company all of the acts of a group even though that company has not been identified as the legal person at the head of that group with responsibility for coordinating the group’s activities’.⁶⁴ The GC has apparently understood this to mean that liability can only be attributed bottom-up in the direction of (direct and indirect) parent companies, but not top-down in the direction of subsidiaries or even horizontally in the direction of sister companies. For example, it held in Parker ITR (2013) that the Commission cannot attribute to a subsidiary the responsibility of its parent company for the unlawful conduct of another subsidiary, which ultimately also means that the subsidiary is not liable for the infringement of the other subsidiary.⁶⁵ In other decisions, the GC has indicated that it interprets Aristrain as meaning that liability can only be attributed to the parent company.⁶⁶ Christian Kersting, on the other hand, has claimed that the real problem in Aristrain was that the existence of a single economic unit had not been proven, because there was already no ‘parent company’ which would have managed the companies of the group in a uniform manner.⁶⁷ That is correct, but it is unclear whether this was a decisive point for the ECJ.

Other judgments, such as the GC’s Michelin decision⁶⁸ and both Courts’ judgments in the Versalis litigation,⁶⁹ are less significant, as they do not concern the question of liability but the calculation of fines – in particular, the relevant turnover to be taken into account and the question of recidivism. As has been repeatedly stressed in this chapter, these are distinct issues that have little to do with the establishment of liability. Nor does it seem sensible to look for the solution in the notion of the undertaking or the concept of the single economic unit.⁷⁰

The decisive factor must ultimately be the purpose of declaring a particular legal entity liable. It has been argued that the liability of sister companies is necessary to properly capture the economic strength of the group.⁷¹ But that is not correct. As already shown,⁷² it is true that the calculation of fines should be based on the group as a whole. This does not mean, however, that all members must also be individually liable. Remember that the ECJ has made it clear in Group Gascogne (2013) that, where the liability of the ultimate parent company is established, the Commission ‘is entitled, for the purposes of assessing the financial capacity of that company, to take into consideration the latter’s consolidated accounts inasmuch as they may be regarded as constituting a relevant factor of assessment’.⁷³ The consolidated accounts, however, automatically include the assets, liabilities, financial positions, profits, and losses of all subsidiaries in terms of Chapter 6 of the EU Accounting Directive 2013/34.⁷⁴ Furthermore, the ECJ has pointed out that it is, in this context, not necessary for the Commission to show that each subsidiary was controlled by the parent company.⁷⁵ Thus, the Court has clearly separated the question of liability from that of attribution of turnover – and that was the right thing to do.

As explained in detail above,⁷⁶ extending liability to legal entities that were not directly involved in the infringement serves to induce them to influence the conduct of the real perpetrator. If this is understood, it is obvious that the mutual liability of sister companies for their respective infringements makes little sense. The same applies to holding subsidiaries liable for infringements committed by their parent companies.⁷⁷ Instead, liability should depend strictly on the possibility of control. In corporate groups, however, control is typically exercised from top to bottom. A parent company can take steps to ensure that its subsidiaries do not commit any infringements and holding the parent liable can therefore contribute to deterrence. But sister companies cannot control each other any more than subsidiaries can control their parent companies. The approach of extending liability to companies which are at the same or lower level in the corporate hierarchy than the actual perpetrator is therefore misguided.

4.4.2 Actions for Damages

Another issue is whether parent company liability is also justified in actions for damages. This has been the subject of much debate since the entry into force of the EU Directive on Antitrust Damages Actions in 2014. The issue has gained further attention since the ECJ recently held in Skanska (2019) that the concept of ‘undertaking’ within the meaning of Article 101 TFEU constitutes an autonomous concept of EU law and cannot have a different scope with regard to the imposition of fines on the one hand and actions for damages on the other.⁷⁸ However, the answer to the question of parental liability for damages cannot be found in terms and definitions, but only in a purposeful interpretation of the relevant provisions. If one assumes, as seems reasonable, that liability for damages does not only serve the purpose of compensation but is also supposed to contribute to deterrence, there is little reason not to apply the principles of parental liability also in actions for damages. This is particularly true if one further assumes that damages actions concern, at least in part, harm that is not adequately taken into account in fine proceedings. A rule of holding parent companies liable for damage caused by their subsidiaries then ensures that the necessary incentives for compliance are also set in relation to these positions.

4.5 The EU–US Divide

Finally, given the comparative perspective of this book, it will now briefly be discussed how the issues of this chapter are treated in US antitrust law. It is often stressed that there is a strong contrast between the EU and the US system with regard to the liability of parent companies.⁷⁹ Indeed, in the US law on antitrust sanctions, there is no general rule of parent company liability for the antitrust infringements of their subsidiaries. Sections 1 and 2 of the Sherman Act both address ‘persons’, defined in Section 7 of the same Act as including ‘corporations and associations’. These terms are based on a rather narrow legal understanding, which in no way resembles the functional approach developed by the ECJ with regard to the term ‘undertaking’ in Articles 101 and 102 TFEU and Article 23 of Regulation 1/2003. As a result, in US antitrust law, the focus with regard to the establishment of liability is very much on the specific legal entity whose employees committed the violation. However, the requirements of the Sherman Act are not so strict that it would not have been possible to develop a different approach. The reasons for the striking US–EU divide must therefore lie elsewhere. Apparently, up to now, there has simply been no need in US antitrust practice for holding parent companies liable. I have tried elsewhere to help answer the question of what might be the reasons for this phenomenon.⁸⁰ Besides the fact that complex group structures are simply much rarer in the US than in Europe (which makes the advantages of parental liability less obvious),⁸¹ the greater significance of other liability mechanisms is likely to play a major role. In particular, the higher relevance of individual liability of managers and employees can be expected to fill some of the deterrent gaps, to which the EU is responding by extending liability to other legal entities.

However, I would like to use this opportunity to draw attention to two other important points. First, it is clear, in this context as well, that the question of liability should not be confused with that of the calculation of fines. While it is true that US antitrust law does not generally recognise parent companies’ liability for fines, it is equally true that the rules on the calculation of fines do make a difference according to whether a company is part of a larger group or not.⁸² This can be seen, for example, in the rules on recidivism.⁸³ According to §8C2.5(c)(1)(A) of the US Sentencing Guidelines (USSG),⁸⁴ an organisation’s culpability score is to be increased by one point if the organisation committed any part of the offence less than ten years after a criminal adjudication based on similar misconduct.⁸⁵ This is comparable to the European Commission’s approach under paragraph 28 of the 2006 Fining Guidelines, except that these do not provide for a maximum period. In our context, it is important that the official comment to §8C2.5(c), published by the US Sentencing Commission, explains that in determining the prior history of an organisation ‘the conduct of the underlying economic entity shall be considered without regard to its legal structure or ownership’.⁸⁶ This shows an openness to approaches that would be comparable with the broad European concept of ‘undertaking’. However, the approach under §8C2.5(c) appears to be more nuanced and not primarily aimed at including other legal entities in the analysis. On the contrary, the Guidelines make it clear that even within one legal unit, there can be several economic units, each of which may or may not be classified as a repeat offender. This is already hinted at in §8C2.5(c), which not only mentions the ‘organisation’ as a reference point for recidivism but also a ‘separately managed line of business’. The official comment explains that, where separately managed lines of business exist, ‘only the prior conduct or criminal record of the separately managed line of business involved in the instant offense is to be used’.⁸⁷ Furthermore, the comment gives the example of two companies merging and becoming separate divisions and separately managed lines of business within the merged company. In such a case, ‘each division would retain the prior history of its predecessor company’.⁸⁸ As this makes clear, the focus is still on the divisions and not simply on the new company as a whole (as it would be under EU competition law).

An expansive approach can be found in §8C2.5(b), which concerns the involvement or tolerance by high-ranking managers. The provision refers to the size of the workforce for qualifying how much the culpability score is to be increased. However, it does not only focus on the ‘organisation’ but also on the ‘unit of the organisation’, which is defined in the comment as meaning ‘any reasonably distinct operational component of the organisation’.⁸⁹ It is further explained that, if a large organisation has several large units such as divisions or subsidiaries, ‘all these types of units are encompassed within the term “unit of the organisation”’.⁹⁰

In other contexts, it is less clear whether the entire group can be considered or whether the focus is strictly on legal entities. This is the case, for example, with regard to the inability to pay. According to §8C3.3(a) of the US Sentencing Guidelines, courts shall reduce the fine below that otherwise required to the extent that imposition of such fine would impair the organisation’s ability to make restitution to victims. Furthermore, §8C3.3(b) allows for the same reduction if a court finds that the organisation is not able and, even with the use of a reasonable instalment schedule, is not likely to become able to pay the minimum fine under the Guidelines. It is further stated that the reduction under any of the two provisions shall not be more than necessary to avoid substantially jeopardising the continued viability of the organisation. However, neither the Sentencing Guidelines nor a recently published memorandum makes it clear whether the economic situation of affiliated companies should also be considered.⁹¹ As a matter of policy, this is essential, as companies might otherwise try to influence their liability by shifting assets to other members of the corporate group. Other important parameters, which are strongly influenced by the single economic unit doctrine in EU competition law, have no equivalents in US fining practice. This applies, for example, to the 10% turnover limit and the deterrence multiplier.

All in all, however, it can be said that the unitary perspective on corporate groups plays a role under Chapter 8 of the US Sentencing Guidelines. The second important qualification to be made here is that, when it comes to the US American fining practice, there is a big difference between the law on the books and the law in action. For the 20 years from 2000 to 2019, the Corporate Prosecution Registry created by Brandon Garrett lists 293 antitrust decisions.⁹² Only three of these cases were resolved with a conviction. In contrast, 278 cases (almost 95%) were resolved by non-prosecution agreements (12), deferred prosecution agreements (2), or plea agreements (264). The fact that by far most cases are resolved by negotiated settlements gives prosecutors considerable leeway, also with regard to the sanctioning of groups of companies. This can be seen, for example, by looking at some of the international cartel cases that, in the EU, have led to the application of the single economic entity doctrine. In Hydrogen peroxide (2006), for example, the European Commission imposed a fine of 25.2 million euros on Akzo Nobel NV, Akzo Nobel Chemicals Holding AB and EKA Chemicals AB.⁹³ The latter was considered the actual perpetrator, whereas the other two companies were held liable as parent companies.⁹⁴ With regard to the US American part of the hydrogen peroxide cartel, a plea agreement was concluded between the US Department of Justice and Akzo Nobel Chemicals International BV,⁹⁵ at the time a wholly owned subsidiary of Akzo Nobel NV and the parent company of Akzo Nobel Chemicals Holding AB, i.e. two of the three companies addressed in the European decision. Interestingly, the plea agreement refers to ‘the defendant, including its subsidiaries’ as a producer and seller of hydrogen peroxide.⁹⁶ Furthermore, even though the agreement was only signed in the name of Akzo Nobel Chemicals International BV, full and truthful cooperation is promised by ‘[t]he defendant and its parents and subsidiaries that are engaged in the sale or production of hydrogen peroxide’.⁹⁷ This should probably include EKA Chemicals AB (or its successors), which are, however, not mentioned in the agreement.

Similar broad language can also be found in other plea agreements.⁹⁸ It seems that although a specific legal entity is always singled out as the contracting party, in reality, the agreement is effectively concluded with the whole group. There is no indication that the US Department of Justice would relieve parts of a unified group of companies of their responsibility simply because they are organised in separate legal entities. This applies, in particular, to cooperation obligations. The relevant passages in plea agreements typically also refer to parent companies, subsidiaries, and successors collectively referred to as ‘related entities’.⁹⁹ They all need to provide documents and make their employees available if so requested by the Department of Justice, and they all benefit from preferential treatment as long as they comply with the plea agreement. It is therefore obvious that the use of plea agreements avoids some of the problems that need to be solved in the EU with complex attribution mechanisms. Prosecutors in the US use the threat of conviction as leverage to secure the voluntary cooperation of the whole group. Since appeal procedures do not play a role where cases are resolved by negotiated settlements, judicial review of antitrust decisions is much less important in the US than in the EU. Enforcers can therefore take the liberty of differentiating somewhat less carefully between distinct legal entities and it is usually not important to prove specific contributions by individual members of the group.¹⁰⁰

In conclusion, the assertion that there is no liability of parent companies under US antitrust law must therefore be refined in two important ways. On the one hand, the US Sentencing Guidelines allow the particularities of corporate groups to be taken into account at least in some respects. While this cannot solve the problem of liability gaps, it will ensure that the calculation of fines does not ignore the economic reality. On the other hand, and this is even more important, the use of plea agreements gives the US authorities considerable leeway, which they can also use to prevent evasive behaviour by groups of companies. This is a very effective way – albeit less transparent than the European Commission’s much more formalistic approach – to ensure that all responsible actors are actually addressed.

4.6 Conclusion

The findings of this chapter can be summarised as follows:

1. 1. In a clear departure from fundamental principles of company law, EU competition law emphasises the common responsibility of corporate groups regardless of separate corporate personalities and limited shareholder liability. This is because competition law is about assessing the actual effects of corporate behaviour, for which company law formalities play no role.

2. 2. The challenges posed by corporate groups cannot be solved by metaphysical considerations of the undertaking or the single economic unit. The de facto personification of the undertaking by the EU Courts is also mistaken. The undertaking is not a person, but merely a legal concept that is supposed to be open to economic realities. As such, it can be used to answer questions of attribution and liability, but it does not answer these questions by itself.

3. 3. The single economic entity doctrine serves three important objectives: (A) to induce parent companies to control the conduct of their subsidiaries, (B) to ensure the correct calculation of fines, and (C) to ensure that fines are actually paid. It is important to clearly distinguish between these functions, as they each have different implications.

4. 4. The threat of liability can be used to encourage parent companies to prevent infringements by their subsidiaries. But that presupposes two things: (A) the parent company must respond better to incentives than the subsidiary and (B) the parent company must be able to deter the subsidiary from infringing the law. It is therefore correct to make the parent company’s liability dependent on its control over the subsidiary at the time of the infringement.

5. 5. Fines will only have a deterrent effect on groups of companies if the calculation takes into account the economic strength of the whole group at the time of the prohibition decision. When the Commission imposes a fine on a company which is part of a group, it should therefore be entitled to rely on consolidated accounts as defined in Chapter 6 of the Accounting Directive. This should not depend on whether it also holds other members of the group liable.

6. 6. In order to ensure the payment of fines, other entities within a group may be held liable as substitute debtors. In theory, this type of liability could be completely independent of any link to the infringement. The absence of such a connection could even mean that the procedural guarantees of the law of fines do not have to be applied.

7. 7. Holding sister companies mutually liable for each other’s competition law infringements or holding subsidiaries liable for infringements by parent companies does not contribute to the objective of deterrence. The reason is that control in groups of companies is typically exercised only from top to bottom, but not the other way round or horizontally.

8. 8. Extending the rule of parental liability from the law of fines to the law of damages can contribute to deterrence as it ensures that the necessary incentives for compliance are also set in relation to harm that is not adequately taken into account in the calculation of fines.

9. 9. While it is true that parent company liability does not exist as a general rule in US antitrust law, it must also be recognised that in the vast majority of cases this is not decisive because they are resolved by negotiated settlements. This practice provides the US authorities with the necessary flexibility to take due account of the specificities of corporate groups.