1.

In July 2015, a train consisting of six locomotives derailed in Kufstein Station (Austria), resulting in damage to two locomotives. The repair of the locomotives took six and eight months respectively, and during this time the carrier leased replacement locomotives. The basis and scope of liability for the ensuing lease costs are at the centre of this litigation.

2.

The preliminary reference submitted by the Oberster Gerichtshof (Supreme Court, Austria) concerns the interpretation of the part of the Convention concerning International Carriage by Rail (COTIF) of 9 May 1980, as amended by the Vilnius Protocol of 3 June 1999, relating to the liability of the railway infrastructure manager. As a preliminary issue, the referring court inquires whether the Court has jurisdiction to interpret provisions of the COTIF, concluded by both the EU and its Member States in the field of transport, in which the EU and the Member States share competence. This is not the first time the Court is invited to decide on its own jurisdiction in respect of mixed agreements. ( 2 ) Given the complexity underpinning mixity in external action, ( 3 ) it will also most likely not be the last.

3.

In the event that jurisdiction is established, the Court is further invited to interpret the extent of liability of railway infrastructure managers under the Uniform Rules concerning the Contract of Use of Infrastructure in International Rail Traffic (‘the CUI Uniform Rules’) ( 4 ) as well as possible derogations therefrom. The Court has not yet had the opportunity to interpret the CUI Uniform Rules.

4.

All Member States of the European Union, with the exception of the Republic of Cyprus and the Republic of Malta, are parties to the COTIF, establishing the Intergovernmental Organisation for International Carriage by Rail (OTIF). The European Union ratified the COTIF with effect from 1 July 2011. ( 5 )

5.

The CUI Uniform Rules ( 6 ) apply to any contract of use of railway infrastructure for the purposes of international carriage, such as the contract concluded between the parties in the main proceedings.

6.

Article 4 of the CUI Uniform Rules, entitled ‘Mandatory law’, reads as follows:

‘Unless provided otherwise in these Uniform Rules, any stipulation which, directly or indirectly, would derogate from these Uniform Rules, shall be null and void. The nullity of such a stipulation shall not involve the nullity of other provisions of the contract. Nevertheless, the parties to the contract may assume a liability greater and obligations more burdensome than those provided for in these Uniform Rules or fix a maximum amount of compensation for loss of or damage to property.’

7.

Article 8 of the CUI Uniform Rules, entitled ‘Liability of the manager’, in paragraphs 1 and 4 thereof, provides:

‘§ 1   The manager shall be liable

caused to the carrier or to his auxiliaries during the use of the infrastructure and having its origin in the infrastructure.

…

§ 4   The parties to the contract may agree whether and to what extent the manager shall be liable for the loss or damage caused to the carrier by delay or disruption to his operations.’

8.

Article 19 of the CUI Uniform Rules, entitled ‘Other actions’, in paragraph 1 thereof, provides:

‘In all cases where these Uniform Rules shall apply, any action in respect of liability, on whatever grounds, may be brought against the manager or against the carrier only subject to the conditions and limitations laid down in these Uniform Rules.’

9.

The Accession Agreement ( 7 ) sets out, inter alia, the relationship between obligations stemming from EU law as well as those derived from the COTIF and its appendices. More specifically, Article 2 thereof provides:

‘Without prejudice to the object and the purpose of the Convention to promote, improve and facilitate international traffic by rail and without prejudice to its full application with respect to other Parties to the Convention, in their mutual relations, Parties to the Convention which are Member States of the Union shall apply Union rules and shall therefore not apply the rules arising from that Convention except in so far as there is no Union rule governing the particular subject concerned.’

10.

Furthermore, Article 7 of the Accession Agreement specifies the method for determining the scope of EU competence:

‘The scope of the competence of the Union shall be indicated in general terms in a written declaration made by the Union at the time of the conclusion of this Agreement. That declaration may be modified as appropriate by notification from the Union to OTIF. It shall not replace or in any way limit the matters that may be covered by the notifications of Union competence to be made prior to OTIF decision-making by means of formal voting or otherwise.’

11.

In recital 2 thereof, Decision 2013/103 provides that:

‘The Union has exclusive competence or shared competence with its Member States in the areas covered by the Convention concerning International Carriage by Rail (COTIF) of 9 May 1980, as amended by the Vilnius Protocol of 3 June 1999 …’

12.

In addition, Annex I to Decision 2013/103 includes the Declaration by the European Union concerning the exercise of competence (‘the Declaration’) referred to in Article 7 of the Accession Agreement. The Declaration contains an appendix providing a list of instruments where, to date, the EU has exercised its competence.

13.

According to the referring court, Paragraph 1293 et seq. of the Austrian Civil Code (Allgemeines bürgerliches Gesetzbuch) (‘the ABGB’) codify the liability for compensation for losses on the basis of fault of the person who caused the loss or damage. In contractual relationships, the liable party is required to prove that it is not at fault with regard to non-compliance with its contractual obligations (Paragraph 1298 of the ABGB). The liable party is liable for the fault of its employees (Paragraph 1313a of the ABGB). Subject to the condition of fault being attributable to the defendant, the lease costs claimed for the replacement locomotives are eligible for compensation in accordance with national law.

14.

In December 2014, Lokomotion Gesellschaft für Schienentraktion mbH, a private railway company with a registered office in Germany (‘Lokomotion’) and ÖBB-Infrastruktur Aktiengeselschaft, an Austrian railway infrastructure firm (‘ÖBB-Infrastruktur’) concluded a contract on the use of railway infrastructure for international transport, in return for which Lokomotion pays a fee.

15.

That contract includes a reference to ÖBB-Infrastruktur’s General Terms and Conditions of Business relating to the Infrastructure Use Contract (‘the GTCs’).

16.

The GTCs provide that liability is to be regulated by the ABGB, the Unternehmensgesetzbuch (Company Code), the Eisenbahn- und Kraftfahrzeughaftpflichtgesetz (Austrian Law on railway and motor vehicle third-party liability) and the CUI Uniform Rules, unless those are contrary to the GTCs.

17.

Following the derailment of Lokomotion’s six locomotives on 15 July 2015, two damaged locomotives were unfit for use for the duration of the repair work. Lokomotion leased two replacement locomotives, thereby incurring costs.

18.

Lokomotion is claiming from ÖBB-Infrastruktur EUR 629110 (plus interest and expenses) for the costs of leasing the replacement locomotives as a result of the accident. It claims that the accident was caused by a shortcoming in the rail infrastructure provided by ÖBB-Infrastruktur. The latter, according to Lokomotion, unlawfully and culpably breached its duties of proper manufacture, inspection, service, maintenance and repair of the rails as codified in the provisions of railway law. The costs of leasing the replacement locomotives are therefore to be regarded as loss of or damage to property as referred to in Article 8(1)(b) of the CUI Uniform Rules.

19.

Conversely, ÖBB-Infrastruktur claims the accident was attributable to a loose coupling hook on the derailed locomotives that was already under excessive strain before derailment, and that Lokomotion is at fault. ÖBB-Infrastruktur additionally claims that the losses are of a pecuniary nature and consequently not eligible for compensation under Article 8(1)(b) of the CUI Uniform Rules.

20.

The first-instance court dismissed Lokomotion’s claim by partial judgment, holding that only the CUI Uniform Rules are applicable, to the exclusion of the ABGB’s liability provisions. It then agreed with ÖBB-Infrastruktur in holding that lease costs cannot be considered ‘damage to property’ within the meaning of Article 8(1)(b) of the CUI Uniform Rules.

21.

The appellate court, however, set aside the partial judgment by the first-instance court and ordered it to reach a new decision after conducting additional proceedings. It took the view that Article 8(1)(b) of the CUI Uniform Rules is to be interpreted broadly and is to include ‘secondary loss of or damage to property’ as well.

22.

ÖBB-Infrastruktur brought an appeal against that decision and requested the Oberster Gerichtshof (Supreme Court) to set it aside.

23.

In those circumstances, the Oberster Gerichtshof (Supreme Court) decided to stay proceedings and to refer the following questions to the Court of Justice:

24.

Written observations were submitted by the parties in the main proceedings and the European Commission.

25.

By its first question, the referring court asks whether the Court has jurisdiction to interpret the CUI Uniform Rules, which form part of the COTIF, an international agreement concluded as a mixed agreement.

26.

Mixed agreements are international agreements concluded simultaneously by the EU and by all or some of its Member States. ( 9 ) Not expressly envisaged by the Treaties, mixed agreements are the creation of the political and legal reality of the European Union. They reflect the uniqueness and complexity of the construction of an integrated Europe. ( 10 )

27.

When an international agreement regulates areas both within and outside its conferred competences, the EU lacks the constitutional power required to conclude such an agreement in its entirety. For that reason, Member States must join as parties alongside the EU to assure the valid conclusion of the agreement in respect of parts that pertain to exclusive Member State competence. This so-called ‘obligatory mixity’ ( 11 ) is a result of the application of the principle of conferral, ( 12 ) which governs the powers of the EU in both internal and external action. ( 13 )

28.

However, in other situations mixity is a choice. This so-called, ‘facultative mixity’ ( 14 ) occurs when the principle of conferral is satisfied in relation to the entire agreement. That means that the EU possesses necessary competences over all provisions of an international agreement, even if some of those competences were not yet exercised, and are, therefore, only potential. Facultative mixity does not, therefore, result from the application of the principle of conferral, as does obligatory mixity, but is rather a question of the exercise of conferred competences.

29.

If the EU shares competence in the areas covered by the agreement with its Member States, such an agreement may be concluded, as has been confirmed by the case-law of the Court, either as a mixed agreement or by the EU alone. ( 15 ) As I will explain in more detail under heading 2 below, that also means that the EU can exercise its shared competence not only by adopting internal legislation, but also by concluding an international agreement about the question in which no internal legislation yet exists.

30.

Facultative mixed agreements are thus a consequence of a political choice on how to exercise conferred shared competences.

31.

The agreement at issue in the present case is a facultative mixed agreement in the area of transport policy, an area in which the EU shares competence with its Member States. ( 16 ) No part of the agreement at issue belongs to an area of exclusive Member State competence, nor is that disputed in the case. In other words, all areas covered by the COTIF could be regulated by the European Union. A number of questions were indeed regulated internally to a large extent even before the EU’s accession to the COTIF. ( 17 )

32.

However, the CUI Uniform Rules cover an area which, at the time of the EU’s accession to the COTIF, belonged to potential shared competences. ( 18 ) That means that the competence to legislate in that area was conferred on the EU, but that the possibility of doing so had not (yet) been used. If formulated more precisely, the question of the referring court therefore asks whether the Court has jurisdiction to interpret parts of a mixed agreement concerning subject matter for which competence is conferred by the Treaties on the EU, but in respect of which the EU has not yet legislated internally.

33.

That question does not yet have a clear answer.

34.

The Court’s jurisdiction over mixed agreements has been discussed in a surprisingly small number of judgments, and for different reasons. ( 19 ) The Court’s case-law has not yet offered a general answer as to the justification and limits of the Court’s jurisdiction over mixed agreements. On the contrary, a number of Advocates General ( 20 ) and legal scholars ( 21 ) have pointed to the difficulties arising from the Court’s approach to jurisdiction, arguing in essence that it creates legal uncertainty concerning the applicable standard.

35.

That is also evident from the range of answers offered by participants to these proceedings in relation to the first question. Lokomotion denies the Court’s jurisdiction, arguing that the CUI Uniform Rules do not belong to the Treaties as they are not an act of an EU institution, and that, in any event, the Court has never interpreted the CUI Uniform Rules and thus lacks jurisdiction. Conversely, ÖBB-Infrastruktur and the Commission are of the opinion that the Court can interpret the CUI Uniform Rules. However, they offer different arguments and rely on varying case-law. On the one hand, ÖBB-Infrastruktur refers to the judgment in COTIF I ( 22 ) and concludes that the Court has jurisdiction to interpret the CUI Uniform Rules because the EU has competence in respect of the COTIF. The Commission, on the other hand, considers that it is not enough that the EU has competence, but since the CUI Uniform Rules concern an area largely covered by EU legislation, the Court may establish jurisdiction. In that respect, the Commission mostly relies on Lesoochranárske zoskupenie. ( 23 )

36.

In my opinion, the complexity of the constitutional division of powers which lies behind mixed agreements has to be taken into consideration in answering the question of the Court’s jurisdiction. Therefore, the answer whether the Court has jurisdiction in relation to a particular provision of a mixed agreement must take into account the reasons for which the EU became a party to such an agreement. In my view, the Court has jurisdiction only in relation to those provisions of a mixed agreement in the adoption of which the EU has exercised its competence.

37.

Notwithstanding the lack of explicit explanation to that effect, I believe that the existing case-law can be understood as allowing the conclusion that the Court does not have automatic jurisdiction over all provisions of a mixed agreement, but may only interpret those provisions in relation to which the EU has exercised its competence. I will present this analysis under heading 1.

38.

Not only is such an understanding possible under the existing case-law, but it is also, in my opinion, the only explanation that is acceptable under the constitutional division of competences as contemplated under the Treaties. That, however, begs the question of how the EU can exercise a competence externally and what the consequence of the exercise of a shared competence is. I will deal with those questions under heading 2.

39.

Under heading 3, I will then apply that reasoning to the case at hand to conclude that the Court has jurisdiction to interpret the CUI Uniform Rules because the EU has, by acceding to the COTIF, exercised its competence in relation to the rules set out therein, including those on the liability of the railway infrastructure manager.

40.

The explanation of the Court’s powers to interpret international agreements started with the judgment in Haegeman in 1974. ( 24 ) That case was concerned with the jurisdiction of the Court to interpret international agreements more generally, not only mixed agreements. The question arose because Article 177 of the EEC Treaty (now Article 267 TFEU) does not expressly mention international agreements as acts which the Court may interpret. The Court decided that it has jurisdiction to interpret the accession agreement between the EEC and Greece because it was concluded by the Council and therefore forms an integral part of EU law. ( 25 ) However, as noted by Timmermans, the ‘integration of a mixed agreement into the Union legal system does not entail … unrestricted jurisdiction of the Court regarding the entire agreement’. ( 26 )

41.

The question of the Court’s jurisdiction to interpret mixed agreements because they are part of EU law, or rather because the particular provision to be interpreted falls within the scope of EU competences, arose for the first time in Demirel. ( 27 ) The German and United Kingdom Governments claimed that the Court does not enjoy jurisdiction in relation to those provisions of the mixed agreement with Turkey in relation to which Member States have exercised their own powers. ( 28 ) The Court replied to those arguments by explaining that the question at issue did not fall within the exclusive competence of the Member States and discarded the argument that the EU had not yet exercised its potential competence as irrelevant, however, without explaining why. ( 29 )

42.

Although Demirel could, consequently, be read as if the Court established its jurisdiction to interpret all provisions of a mixed agreement, except those covering an area still in Member States’ exclusive competence, the Court did not clearly state or explain such a position. That early judgment cannot, therefore, be seen as taking a final stance on the general question of the Court’s jurisdiction to interpret mixed agreements. ( 30 )

43.

Later judgments relating to the Court’s interpretative jurisdiction may be divided, in my opinion, into two basic groups. The first group consists of those judgments that propose a broad reading of the Court’s jurisdiction, on the basis that mixed agreements are part of the corpus of EU law and as such necessitate uniform interpretation. In this group are several judgments relating to the Agreement on Trade-Related Aspects of Intellectual Property (‘the TRIPS Agreement’), such as Hermès ( 31 ) and Dior, ( 32 ) the judgment in Lesoochranárske zoskupenie, ( 33 ) relating to the Aarhus Convention, and the recent judgment in Republic of Moldova, ( 34 ) relating to the Energy Charter Treaty (ECT). The second group of cases adopts a more reserved approach and looks for the link between the provision whose interpretation is being sought and the exercise of EU competence in order to establish the Court’s jurisdiction. In that group is Merck, ( 35 ) also concerning the TRIPS Agreement, as well as a number of judgments resulting from infringement proceedings, such as Commisson v Ireland, ( 36 ) relating to the Bern Convention for the Protection of Literary and Artistic Works; Etang de Berre, ( 37 ) relating to the Convention for the protection of the Mediterranean Sea against pollution; or MOX Plant, ( 38 ) relating to the UN Convention on the Law of the Sea.

44.

The question of jurisdiction arose in those judgments for different reasons and they cannot therefore be properly understood outside their particular contexts. However, all of them may, in my opinion, be read as leading to the conclusion that the Court links its jurisdiction over a particular provision of a mixed agreement to the exercise of EU competence.

45.

That reading is not immediately obvious, especially in the light of the first group of judgments. Let me, therefore, explain my statement on the basis of Hermès. ( 39 ) In that case, the Court was faced with a situation in which it was unclear whether a national interim measure relating to a Benelux trade mark fell within the scope of Article 50 of the TRIPS Agreement, regulating provisional measures. The only EU legislation in the area covered by the relevant provision of the TRIPS Agreement was a regulation on Community trade marks. ( 40 ) However, the dispute concerned a Benelux and not a Community trade mark. Nevertheless, the Court found that it had jurisdiction because Article 50 of the TRIPS Agreement can apply equally to situations within the scope of national and EU law. Its uniform interpretation was, the Court explained, clearly in the EU interest. ( 41 )

46.

The Court’s reliance on the ‘interest in uniform interpretation’ could lead to the conclusion that such an interest was the main justification for establishing jurisdiction to interpret Article 50 of the TRIPS Agreement.

47.

However, in that case, as well as in others falling within the first group, the Court had to answer two different questions relating to its jurisdiction, which must be separated. The first is whether the Court, in principle, has jurisdiction to interpret a specific provision of a mixed agreement (Article 50 of the TRIPS Agreement in Hermès). The second question is whether the Court may interpret the relevant provisions of the mixed agreement where that interpretation is to be applied in a dispute before the national court outside the scope of EU law.

48.

In its answer to the former question, it was the exercise of powers by the EU in relation to the provision of a mixed agreement of which the interpretation was sought that enabled the Court to establish its jurisdiction. By acceding to the TRIPS Agreement, Article 50 thereof became part of EU law in the area of Community trade marks. Thus, the EU’s accession to the TRIPS Agreement made Article 50 thereof part of EU law.

49.

The Court relied on the ‘interest in uniform interpretation’ only to justify the second question relating to its jurisdiction. That was necessary because the dispute before the national court did not concern the Community, but the Benelux trade mark. In that case, therefore, Article 50 of the TRIPS Agreement was to apply as national law, and not as EU law. Citing and applying the line of case-law developed in Dzodzi, ( 42 )Leur-Bloem ( 43 ) and Giloy, ( 44 ) the Court justified its jurisdiction to interpret EU law in situations falling outside the scope of its application by the necessity of avoiding diverging interpretations of EU law that might occur if the Court does not assist national judges in domestic situations. That, however, was only possible because Article 50 of the TRIPS Agreement became a part of EU law in the first place.

50.

Similar reasoning was applied in all the cases that I mentioned as belonging to the first group. In all of them, the underlying reason justifying the Court’s jurisdiction to interpret Article 50 of the TRIPS Agreement, Article 9(3) of the Aarhus Convention, or provisions relating to foreign direct investments of the ECT, respectively, lay in the choice made by the EU to make those provisions part of EU law by acceding to the respective mixed agreements.

51.

That logic was expounded more clearly in the second group of cases, in which the Court made a more direct connection between the content of the agreement and the exercise of EU competences. Jurisdiction was established either because the question covered by a mixed agreement was regulated by existing EU rules, or, if not, was considered to fall into an area that was to a large extent covered by EU rules. If, however, no such link was established, the Court found that it had no interpretative jurisdiction, even if the potential shared competence existed under the Treaties.

52.

That was the situation in Merck. ( 45 ) The Court was again tasked with the interpretation of the TRIPS Agreement, this time Article 33 thereof governing the expiration period for patent protection. The Court found that its jurisdiction depends on whether the EU has exercised its competence in the relevant area of the TRIPS Agreement. After concluding that patents were not (or only sporadically) regulated under EU law, the Court considered that the question whether Article 33 of the TRIPS Agreement has direct effect was for the national court to decide. In the words of the Court itself, ‘it must be concluded that, since Article 33 of the TRIPs Agreement forms part of a sphere in which, at this point in the development of Community law, the Member States remain principally competent, they may choose whether or not to give direct effect to that provision’. ( 46 )

53.

Translated into the language of jurisdiction, this meant that the Court did not have jurisdiction to interpret Article 33 of the TRIPS Agreement because the EU had not yet exercised its shared competence in the field of patent protection.

54.

A number of judgments, listed in point 43 of this Opinion, adopted in infringement proceedings, rather than as preliminary rulings, also connected the Court’s jurisdiction to the exercise of EU competence. Those judgments posed the question of jurisdiction in a slightly different manner. What the Court was requested to resolve as a preliminary issue in those cases was whether the provision of a mixed agreement that was allegedly breached was part of EU law or not. Only in the former case could a Member State be found to have committed a ‘failure to fulfil its obligation under the Treaties’. Those judgments introduced an apparently novel factor: jurisdiction is established if the area at issue is already largely covered by EU legislation. ( 47 )

55.

The finding that a policy area is already largely covered by EU law should be understood as an indicator which enables the Court to hold that, by acceding to a mixed agreement, the EU also chose to exercise its shared competence concerning the particular question that does not feature in its internal legislation. The exercise of competence thus happened by way of conclusion of the mixed agreement at issue. ( 48 ) That is what enabled the Court’s jurisdiction.

56.

If read in the proposed way, both lines of case-law allow for the conclusion that the Court’s jurisdiction to interpret a provision of a mixed agreement in relation to which the EU has not yet exercised its competences internally depends on the finding that the EU has exercised its competence in relation to that particular issue by concluding the mixed agreement.

57.

I will further explain why that conclusion is necessary and the only one possible under the current EU constitutional order.

58.

The exercise of competences which the EU shares with its Member States is regulated by the Treaties. The constitutional choice of how to organise competences is both a justification and a limit to the Court’s jurisdiction to interpret mixed agreements.

59.

The exercise of EU competence might happen internally, through the adoption of internal legislation, or externally, by the very fact of accession to the international agreement which regulates areas of shared competence. The ability to conclude a mixed agreement in an area of shared competence does not depend on the condition that such competence was already exercised internally. It may be exercised for the first time by concluding an international agreement.

60.

Such a view was already proposed by Advocates General. In Mox Plant, Advocate General Maduro considered that ‘the conclusion of an international agreement can itself be a form of exercising a non-exclusive competence of the Community, independently of the previous adoption of Community internal legislation’. ( 49 )

61.

Two recent Court judgments, COTIF I ( 50 ) and Antarctic MPAs, ( 51 ) offer a confirmation that a shared competence may be exercised for the first time through the conclusion of an international agreement. ( 52 ) In COTIF I, the Court dispelled doubts created by a statement in Opinion 2/15, ( 53 ) and confirmed that the EU can always choose ( 54 ) to conclude an international agreement in the area of shared competences alone. ( 55 ) That means the EU can exercise its shared competence for the first time by signing an international agreement.

62.

Therefore, whenever it can be concluded that by acceding to an international agreement the EU chose to exercise (up to that moment only potential) shared competences, the relevant provisions of that agreement become an exercised EU competence as of the moment of its conclusion, and therefore part of EU law.

63.

Some misunderstanding, it seems, results from confusing the question of the exercise of shared competences externally (by conclusion of an international agreement) with the question of whether the external competences are exclusive. That is a result of how the ERTA principle ( 56 ) was embodied in today’s Article 3(2) TFEU. That article provides that the EU has ‘exclusive’ competence to enter into an international agreement if such an agreement may affect or alter those competences that were already exercised by common rules.

64.

In contrast to external EU relations, in internal situations, the FEU Treaty talks about exclusive or shared competences. Exclusive competences prevent Member States from acting, but, similarly, shared competences, once exercised by the EU, also preclude Member States from acting. However, in the internal sphere, that is not called exclusive competences; it is called pre-emption (even though the FEU Treaty does not use that expression). Pre-emption is envisaged in Article 2(2) TFEU.

65.

The inability of Member States to enter into international agreements if doing so would affect common rules is also the result of pre-emption. However, in the organisation of the Treaties, such pre-emption is placed under the provision relating to exclusive competences (Article 3(2) TFEU). That might be the reason why the exercise of competences in external relations is often seen and explained differently from the exercise of competences in internal relations.

66.

Additionally, Article 3(2) TFEU deals with pre-emption in a situation when shared competence is first exercised internally. It determines that it is the EU, and not the Member States, that is to conclude the agreement relating to the area in which Member States’ competence has been pre-empted. However, Article 3(2) TFEU does not resolve the question of consequences that follow if the shared competence is for the first time exercised externally.

67.

When the more appropriate terminology of pre-emption is used, ( 57 ) the situation becomes clear. Article 2(2) TFEU envisages that, in the area of shared competences, Member States may exercise their competences to the extent that the EU has not exercised its competence. Stated conversely, Member States cannot exercise their competence (they are pre-empted from doing so), if the EU has exercised such competence on a particular issue. Therefore, if the EU has exercised a shared competence internally, Member States cannot act to affect common rules, either by adopting internal legislation or by concluding an international agreement. That is the reason behind the ERTA principle embodied in Article 3(2) TFEU.

68.

Whether the EU exercises a shared competence by concluding an international agreement or by legislating within its legal order, the effect should be the same: Member States cannot alter that rule unilaterally. It is therefore irrelevant in that regard whether the EU has exercised its competence internally or externally. The reason why Member States are prevented from unilaterally altering the provisions of an international agreement within their internal legal order is the same as that for which they are prevented from unilaterally altering measures adopted within the EU legal order: ensuring the coherent and uniform application of EU law.

69.

Relying on the notion of pre-emption allows for the resolution of another common confusion in EU external relations, the one by which the question of the EU’s implied external competence is conflated with that of the exclusivity of its external competence. ( 58 ) Namely, the EU has implied external competence in all areas in which it has internal competence (exclusive or shared). The exercise of a competence envisaged internally as shared precludes Member States from exercising that competence internally or externally. ( 59 )

70.

It is further apparent from the foregoing that the search for existing internal EU rules or an assessment whether the area is already largely covered by EU rules serves to determine whether Member States’ attempts to regulate the question at issue have been pre-empted. That also justifies the Court’s interpretative jurisdiction.

71.

Common rules may be affected not only by adopting contrary rules or amendments in their text, but also by their interpretation. Independent interpretation of common rules by Member States’ authorities could affect or alter them. The Court has the final authority within the EU legal order to interpret those provisions of an international agreement which become common rules because the EU exercised its shared competence by acceding to an international agreement.

72.

Saying that the Court has jurisdiction to interpret provisions of an international agreement through which the EU has exercised its shared competence is, therefore, equivalent to saying that the Court has jurisdiction to interpret a regulation or a directive adopted by EU institutions. The relevant parts of such an international agreement are of the same nature for the EU legal order as internal secondary law. The Court’s jurisdiction naturally follows.

73.

The recognition of the EU’s power to exercise its shared competence for the first time at the international level is constitutionally sensitive because this choice influences the regulatory ability of Member States in the field at issue. That is so because of the pre-emptive effect of the majority of shared competences under the Treaties: once the EU has regulated an issue, Member States cannot regulate it on their own for as long as the EU regulation remains in force.

74.

It is an appealing idea to simplify the question of the Court’s jurisdiction by extending it to all provisions of a mixed agreement simply because such an agreement became part of EU law and because that enhances uniformity in the interpretation of EU law. ( 60 ) However, as rightly pointed out by Advocate General Cosmas, ( 61 ) that EU interest cannot justify abolishing the existing division of competences between the EU and its Member States. That division was agreed upon in the Treaties, the constitutional charter for the EU’s legal and political system. Therefore, even though linking the exercise of EU competences to the Court’s jurisdiction makes the latter a ‘hostage to the complexity of the former’, ( 62 ) that shortcoming alone cannot justify ignoring the constitutional choices made on how the powers of the EU and its Member States are delimited.

75.

That is why the Court may establish jurisdiction only in relation to those provisions of a mixed agreement in which the EU exercised its regulatory competence. In internal legislation adopted in an area of shared competences it is always clear that the EU aimed at regulating everything contained in such acts. In mixed agreements, on the contrary, that is not self-evident, but must be confirmed through the analysis of each particular agreement in its own context.

76.

The decision whether the shared competence which pre-empts Member States from acting is to be exercised by the EU or not is a political choice. ( 63 ) However, that choice is subject to EU constitutional principles. First, the EU can act only in order to achieve the objectives of EU policies. ( 64 ) Second, it can exercise a shared competence only if the principles of subsidiarity and proportionality are satisfied. Those principles govern the exercise of EU competences both internally and externally. ( 65 )

77.

Subsidiarity protects decision-making at Member State level. That becomes particularly important if one takes into consideration that shared competences, once exercised at EU level, preclude the action of Member States. If the EU exercises its competence externally, the principle of subsidiarity is not ensured by the same processes as in internal decision-making. ( 66 ) That principle must nevertheless be respected. Thus, breach thereof can be invoked as a reason for questioning the conformity of an international agreement with the Treaties ex ante in the procedure under Article 218(11) TFEU, or subsequently in an annulment procedure or a preliminary ruling on validity.

78.

Advocate General Cosmas pointed out another constitutional feature that should not be neglected in determining the Court’s jurisdiction over mixed agreements. It concerns the delimitation of powers on the horizontal axis, between the Court and other EU institutions. As he pointed out, even though the law-creating role is inherent in the Court’s general role, it does not mean that the Court should take a legislative initiative. ( 67 ) The decision whether to harmonise national legislation in a certain area, in the present case in relation to the liability of a railway infrastructure manager, belongs to EU political institutions and the decision-making process as envisaged by the Treaties. If the Court were automatically to conclude that all provisions of a mixed agreement are the result of the exercise of EU competences (the result of an automatic establishment of the Court’s jurisdiction), the Court would in effect be substituting its own assessment for that of the competent political EU institutions as to whether harmonisation in the field is necessary or not.

79.

In COTIF I, the Federal Republic of Germany argued that exercising shared competence in the external field without first doing so internally creates a risk of avoiding the participation of the European Parliament and the circumvention of the ordinary legislative procedure. ( 68 ) This argument has its merits in light of the EU’s constitutional order.

80.

However, the procedure under Article 218 TFEU ensures that EU institutions enjoy similar powers when competences are exercised in external relations as those that they enjoy in internal legislative procedures. The Court has recognised as much in its case-law. It confirmed that Article 218(6)(a)(v) TFEU ( 69 ) lays down the procedure applicable in the case of agreements which cover areas to which the ordinary legislative procedure applies. ( 70 ) It has considered that the two procedures ensure a similar institutional balance. ( 71 ) In both cases, the Commission enjoys the power of initiative, the Council decides by qualified majority and the consent of the European Parliament is required.

81.

Given the constitutional importance of connecting the jurisdiction of the Court to interpret a mixed agreement with the exercise of EU powers in the adoption of such an agreement, the only possible conclusion is that the assessment of the reasons for and the circumstances in which an agreement was concluded as mixed is necessary before deciding on the existence of the Court’s interpretative jurisdiction. That, of course, presupposes that the Court has the jurisdiction to interpret the entire agreement in order to decide which of its provisions are provisions of EU law. Such interpretative jurisdiction was indeed already confirmed by the Court. ( 72 )

82.

The task of determining which provisions of a mixed agreement are those in relation to which the EU has exercised its shared competence is a complex one that can only be performed on a case-by-case basis. Can anything be concluded from the very fact that an agreement was concluded as mixed? It might be possible to claim that Member States participate in such an agreement because it has been decided that only some of the shared competences are exercised by the EU, while others remain the matter of Member States’ regulatory responsibility.

83.

What could otherwise be the reason for Member States joining the agreement? In my view, Member States might have other reasons to join an international agreement as parties with the EU, even if such an agreement results in the pre-emption of their unilateral action in all the areas it covers. For instance, Member States might wish to retain control over negotiations of such an agreement, even if they would lose it after its conclusion. Member States might also simply want to remain visible at the international level. Finally, international law (as opposed to EU law) might have dictated their participation. I do not think, therefore, that much can be concluded from the mere fact that an agreement is concluded as mixed.

84.

Mixed agreements in which the EU enters as a party can be very different. ( 73 ) They may be bilateral or multilateral, negotiated as new agreements, or have existed prior to the time when the EU acquired a competence to become a member thereof. The circumstances surrounding the conclusion of each agreement individually might differ. Answering the question whether the EU has exercised its shared competence is particularly difficult in multilateral agreements in which the Member States were parties before the EU joined such an agreement. Whether the Court has jurisdiction to interpret a mixed agreement cannot, therefore, in my view, be determined in advance as a general rule applicable to every agreement.

85.

The Court must assess the circumstances surrounding the EU’s participation in a particular agreement in order to conclude whether the EU has exercised its shared competence in relation to a particular provision of such an agreement. If that was the case, the Court acquired jurisdiction to interpret such a provision of the mixed agreement for the purposes of its application within the EU legal order.

86.

It is conceivable that the Court may not be able to infer from the preparatory documents, the procedure leading to conclusion, the text of a mixed agreement, or any other relevant circumstance, that the EU intended to exercise some of its shared competences by acceding to the agreement at issue. I am of the opinion that it would then have to conclude that, in order to respect the constitutional organisation of competences, the powers remained with the Member States. In such a case, the Court would have to decline jurisdiction. Here, however, that is not the case.

87.

The COTIF, including the CUI Uniform Rules, is, as explained at the outset of this opinion, a mixed international agreement in the area of transport, for which the Treaties envisage that the EU shares its competence with the Member States. All the issues covered by the COTIF fell within the competence, either exclusive or shared, of the EU, at the time when the EU acceded to it. There is, thus, no area in which Member States have exclusive competence.

88.

Twenty-five EU Member States were parties to the COTIF before the EU joined as a party in 2011. It is clear that the EU took on the responsibility for those parts of the COTIF in which it possessed exclusive external competence by virtue of internal legislation (through pre-emption). ( 74 ) A list of EU legislative acts was attached to the Declaration. ( 75 )

89.

The Declaration cannot, however, be understood as an exhaustive list of the areas in which the EU exercised its competence when acceding to the COTIF, since the exercise of internal competences by the EU is constantly evolving. ( 76 ) It consequently does not preclude an interpretation according to which the EU has exercised its shared competences (for the very first time) by the very act of accession to the COTIF.

90.

The CUI Uniform Rules, as part of the COTIF, cover an area of EU transport policy that was not yet (or else only marginally ( 77 )) regulated internally. The CUI Uniform Rules concern contracts for the use of rail infrastructure and the liability for damages of either the operator or the rail network manager that might arise during such use.

91.

Has the EU, by joining the CUI Uniform Rules as part of the COTIF, exercised its shared competence in the area of transport and pre-empted Member States from regulating contracts for the use of railway infrastructure and related liability for damages in the situations of intra-EU contracts? If yes, the Court does have interpretative jurisdiction over the relevant provisions of the CUI Uniform Rules.

92.

The CUI Uniform Rules, by harmonising the conclusion of contracts for the use of railway network infrastructure and the liability of parties in intra-EU cross-border transport may be deemed to contribute to the objectives of the EU relating to transport policy.

93.

Harmonised regulation of liability eliminates divergence in cross-border operation of rail transport, thus contributing to the removal of obstacles to the smooth functioning of cross-border transport, an aim that cannot easily be achieved if left to divergent legislation of Member States. Action at EU level may thus be justified as falling under one of the objectives of transport policy and satisfying the principle of subsidiarity.

94.

Article 2 of the Accession Agreement reads as follows: ‘Without prejudice to the object and the purpose of the Convention to promote, improve and facilitate international traffic by rail and without prejudice to its full application with respect to other Parties to the Convention, in their mutual relations, Parties to the Convention which are Member States of the Union shall apply Union rules and shall therefore not apply the rules arising from that Convention except in so far as there is no Union rule governing the particular subject concerned.’ ( 78 )

95.

I read Article 2 of the Accession Agreement as an expression of the EU’s intention to regulate liability related to the performance of contracts on the use of rail networks in intra-EU cross-border transport by applying the CUI Uniform Rules to those relations. In other words, the EU decided to regulate the liability of the network manager by accepting the CUI Uniform Rules also for intra-EU situations.

96.

Additionally, Annex II to Decision 2013/103 ( 79 ) explains the meaning of the underlined phrase in the following way: ‘The term “governing the particular subject concerned” is to be understood as applying to the specific case which is governed by a provision of the Convention, including its appendices, and is not governed by European Union legislation.’

97.

The wording ‘except in so far as there is no Union rule governing the particular subject concerned’ must in my opinion be read as meaning that the EU is to use the COTIF as its own rules governing intra-EU transport. Thus, the EU applies that convention in areas that it has not yet regulated internally, for the purposes of also regulating intra-EU situations over which it possesses shared competence.

98.

That convention includes the CUI Uniform Rules, which means that the EU has decided to apply those rules to the liability of railway infrastructure managers in intra-EU situations. At the same time, it is clear that by that choice, the EU has not prevented itself from changing those rules in relation to intra-EU transport. Therefore, the CUI Uniform Rules apply to intra-EU relations only if and for as long as there are no different rules adopted by the EU. But, until such rules are adopted, the CUI Uniform Rules also become EU law.

99.

For that reason, I am of the opinion that the EU has exercised its competence in the area of transport policy, which it shares with its Member States, to apply the system of liability of railway infrastructure managers as envisaged by the CUI Uniform Rules also to intra-EU situations. This has resulted in the pre-emption of Member States’ regulatory powers in that area. The Court therefore has jurisdiction to interpret the CUI Uniform Rules.

100.

By its second question, the national court asks whether Article 8(1)(b) of the CUI Uniform Rules includes costs incurred by the carrier for leasing replacement locomotives during the repair period of the damaged locomotives. The Court is therefore invited to interpret the concept of damage to property under this provision. As already mentioned in the introduction, the Court has not yet had the opportunity to interpret the CUI Uniform Rules.

101.

ÖBB-Infrastruktur argued that the lease costs of replacement locomotives represent damage to the assets of Lokomotion rather than damage to locomotives themselves, and should thus be distinguished from costs related to the damage to the locomotives themselves. In addition, it argued that the structure of the CUI Uniform Rules rests on the strict liability of the railway infrastructure manager, which should result in a narrow interpretation of the scope of liability in Article 8(1)(b) of the CUI Uniform Rules. ÖBB-Infrastruktur highlighted that, contrary to Article 15 of the CUI Uniform Rules, ( 80 ) which allows for surpassing liability limits when fault is established, the objective liability requirement from Article 8(1)(b) of the CUI Uniform Rules calls for a narrow interpretation of ‘damage to property’.

102.

The Commission sided with that interpretation, noting that the structure of the COTIF and the CUI Uniform Rules support a distinction between material and pecuniary damage, only the former being covered by Article 8(1)(b) of the CUI Uniform Rules.

103.

Lokomotion, on the contrary, argued that lease costs for the replacement locomotives form part of costs resulting from damage to property under Article 8(1)(b) of the CUI Uniform Rules, given that they also contribute to the restoration of the locomotives to their original state.

104.

I do not agree.

105.

When interpreting a provision of an international treaty, the Court has previously relied ( 81 ) on Article 31 of the Vienna Convention of 23 May 1969 on the Law of Treaties ( 82 ) and Article 31 of the Vienna Convention of 21 March 1986 on the Law of Treaties between States and International Organisations or between International Organisations, ( 83 ) which express customary international law. According to those provisions, a treaty is to be interpreted in good faith in accordance with the ordinary meaning to be given to its terms in their context and in the light of its object and purpose. ( 84 )

106.

In the first place, and contrary to Lokomotion’s assertions, nothing in the structure and context of Article 8(1) of the CUI Uniform Rules suggests that its purpose is to fully restore the original situation of the operator, including the removal of any and all impediments to the regular functioning of its business caused by the damage to property.

107.

Rather, Article 8(1)(a) of the CUI Uniform Rules refers to bodily loss or damage (such as death, injury or any other physical or mental harm). For example, in the event of personal injury to the operator’s personnel, it does not appear to me consistent with this provision that a carrier would be able to claim costs related to employing additional personnel due to such injury.

108.

In the same vein, the repair costs incurred by Lokomotion are in my view the ones to be compensated under Article 8(1)(b) of the CUI Uniform Rules. Leasing replacement locomotives does not, in my opinion, target loss of or damage to property, but is an additional cost arising from Lokomotion’s intention to continue providing its service without disruption.

109.

Article 8(1)(c) of the CUI Uniform Rules is the only provision that refers to pecuniary costs (that is to say, those beyond concrete damage to humans or property) that the carrier is able to pass on to the infrastructure manager.

110.

In addition, the wording of Article 8(1)(b) of the CUI Uniform Rules states that the manager (in the present case, ÖBB-Infrastruktur) ‘shall be liable’, thus providing for strict (objective) liability. I would agree with ÖBB-Infrastruktur and the Commission that objective liability calls for a narrow interpretation of the concept of ‘loss of or damage to property’.

111.

Finally, I also consider Article 8(4) of the CUI Uniform Rules relevant for a narrow interpretation of the scope of liability of the manager under Article 8(1)(b) of those rules. Specifically, that provision allows for an extension of the scope of a manager’s liability for the ‘loss or damage caused to the carrier by delay or disruption to his operations’. Those losses or damage are incurred by the carrier as a result of its inability to use the lost or damaged vehicle in a timely and regular manner.

112.

My suggestion is not that the costs incurred by Lokomotion fall under this provision, given that its operation was not, according to the information available, delayed or disrupted. However, costs susceptible to compensation outside the damage to property itself under Article 8(1)(b) of the CUI Uniform Rules are explicitly listed in the scheme of those rules.

113.

Article 8(1)(c) of the CUI Uniform Rules, finally, refers to ‘pecuniary loss resulting from damage payable by the carrier under the CIV Uniform Rules [ ( 85 )] and the CIM Uniform Rules’. ( 86 ) Pecuniary losses that can be recovered under (strict) objective liability are those that arose for the carrier as a consequence of its inability to comply with the obligation of carriage under the CIV and CIM Uniform Rules. Those pecuniary losses can, therefore, be passed on to the infrastructure manager.

114.

That interpretation appears to me compatible with the Explanatory Report to the CUI Uniform Rules, ( 87 ) which states that pecuniary losses are not included in Article 8(1)(b) of the CUI Uniform Rules.

115.

It is true that by leasing the replacement locomotives, it is hard to conceive of a situation in which pecuniary losses listed in Article 8(1)(c) or Article 8(4) of the CUI Uniform Rules would arise. Nevertheless, that does not mean that such losses are covered by Article 8(1)(b) of the CUI Uniform Rules.

116.

I am therefore of the opinion that the leasing costs for replacement locomotives do not fall under the concept of damage to property under Article 8(1)(b) of the CUI Uniform Rules.

117.

By its third question, the national court asks whether, in the event that Article 8(1)(b) of the CUI Uniform Rules is not applicable to the costs incurred by Lokomotion in replacing the damaged locomotives, a blanket reference in the contract to the ABGB constitutes a valid assumption of greater liability than that in the CUI Uniform Rules under Article 4 and Article 19(1) thereof.

118.

This question requires the Court to decide on the appropriate balance between freedom of contract and the mandatory nature of the CUI Uniform Rules in respect of liability. ( 88 )

119.

As explained above, the parties agreed to include the GTCs in their contract on the use of railway infrastructure. ( 89 ) These refer to a number of sources, including the ABGB, which in turn provides for fault as a requirement of liability on the one hand, but allows for a broader scope of claims, such as the leasing costs in the present case.

120.

Against this background, can it be concluded that the parties have, by including a subsidiary application of the ABGB, contracted greater liability as allowed under Article 4 of the CUI Uniform Rules?

121.

In my opinion, the answer to that question is yes.

122.

Article 4 of the CUI Uniform Rules allows the parties to assume a liability greater and obligations more burdensome than those provided in the CUI Uniform Rules.

123.

At the outset, I do not consider that that provision requires both the basis and the scope of liability to be greater than what is provided in the CUI Uniform Rules. Rather, I understand Article 4 of the CUI Uniform Rules to provide the parties with the freedom to contract an extension of liability.

124.

This interpretation aligns with the Explanatory Report to Article 4 of the CUI Uniform Rules, which states that the purpose of that article is to allow parties to the contract to ‘extend their liability’. ( 90 ) The Explanatory Report makes no distinction between basis and scope of liability.

125.

Therefore, the parties can make provision in the contract for extending the liability of a network manager for the losses incurred by the operator by leasing replacement locomotives during the repair of those damaged. Equally, such liability may be based on fault.

126.

ÖBB-Infrastruktur put forward the argument that a general reference to several pieces of national legislation in its GTCs provides a list of examples, rather than an exhaustive list, thus rendering it too imprecise to be considered a derogation under Article 4 of the CUI Uniform Rules. ÖBB-Infrastruktur also claimed that the GTCs refer to several national laws that could be applicable to the situation at hand, thus resulting in a need to apply the CUI Uniform Rules as the sole instrument regulating liability arising from the contract.

127.

Article 4 of the CUI Uniform Rules does not limit the manner in which the more extensive liability may be contracted by the parties. There is nothing to suggest that that could not be achieved by a blanket choice of the rules of a national system. It is for the national court to decide whether that was indeed the intention of parties under the contract.

128.

That conclusion is not called into question by Article 19(1) of the CUI Uniform Rules, which prescribes the mandatory application of the CUI Uniform Rules to other actions in respect of liability. The Explanatory Report to Article 19(1) of the CUI Uniform Rules ( 91 ) states that its aim is to ‘protect fully the liability system … by limiting extra-contractual claims’. However, I consider that the dispute in the main proceedings concerns contractual claims (an extension of liability) stipulated under what Article 4 of the CUI Uniform Rules allows. It follows that Article 19(1) of the CUI Uniform Rules does not apply to the case at hand.

129.

Thus, I consider that the Court should answer the third question to the effect that Article 4 and Article 19(1) of the CUI Uniform Rules allow the parties to contract an extension of liability by a blanket reference to national law. Such an extension may provide for liability that is of greater extent, but is, however, conditional on proof of fault.

130.

In light of the foregoing considerations, I propose that the Court answer the questions referred for a preliminary ruling by the Oberster Gerichtshof (Supreme Court, Austria) as follows: