JUDICIAL SCRUTINY OF MERGER DECISIONS IN THE EU, UK AND GERMANY

Abstract The appropriate role of the courts in controlling the discretion of merger authorities has become one of the key issues in European merger law and policy in recent years. This article investigates judicial review of merger decisions, taking a comparative approach by examining cases from the EU, UK and Germany. We observe an apparent increase in the willingness of the EU and UK courts to scrutinize merger decisions, and a long-standing tradition of close scrutiny in Germany. In respect of the EU and UK, we consider agency theory offers a convincing explanation—that increased scrutiny is explained by the need to enhance the credibility of merger policy. In Germany, the constitutional basis of judicial review differs significantly, and the relatively close scrutiny exercised by the court is better explained by the very different constitutional context.


I. INTRODUCTION
Courts are often left out of institutional analysis. This is despite the fact that judicial decisions may have profound consequences for the way in which agencies exercise their powers. 1 Judicial review and the limits of the court's jurisdiction are matters for the court. 2 Moreover, the way in which the court exercises its supervisory jurisdiction, and the degree to which the court defers to the public body under scrutiny, is highly dependent upon context (the nature of the power, the rights affected by the decision, the type of error alleged to have been made, and so on). For the purposes of this article, two contextual factors are of particular significance: the legitimacy and expertise of the court relative to the decision-maker. We explore these questions throughout, and this article has wider significance in terms of the judicial supervision of independent expert agencies.
The appropriate role of the courts in controlling the discretion of merger authorities has become one of the key issues in European merger law and policy in recent years. This article investigates judicial review of merger decisions, taking a comparative approach by examining cases from the EU, 3 UK and Germany. We observe an apparent increase in the willingness of the EU and UK courts to scrutinize the substance of merger decisions. At first glance, this is a surprising observation. The principles underpinning merger law and practice are increasingly drawn from economics (the so-called 'effects-based' approach) which, in turn, requires increasingly demanding levels of expertise and specialization on the part of the competition agency. Given this, it would be reasonable to expect more, not less, deference being shown to agencies by reviewing courts. Exploring this apparent paradox is the central motivation for this article.
We argue that what we observe for the EU and UK is consistent with an agency perspective of judicial decision-making, whereby the court is focused upon the control of agency discretion in order to promote credible decisionmaking on the part of the agency. While having very different foundations in both theory and law, through judicial review the courts in these two jurisdictions attempt to strike a balance between agency autonomy on the one hand, and the need for credible decision-making, on the other. We contrast the position with Germany where there is a tradition of the courts exercising a high level of supervision over administrative agencies, especially in the merger area.
This article is organized as follows. In section II we set out the theoretical foundations of the article and our central hypothesis. Section III explains and compares the institutional context for merger control in each jurisdiction, and in particular the role of the reviewing courts. Given our theory, we think it essential to understand the constitutional context for judicial review in each country, which varies considerably, as we show. Section IV contains our comparative case law analysis. 4 As a preliminary to our comparative analysis, we set out the different ways in which the reviewing court controls agency decision-making. The nature and extent of judicial review varies according to the nature of the question being addressed by the agency-be they questions of law, fact or decisions involving the exercise of discretion. Section V concludes.

II. THE COURTS, AGENCY THEORY AND CREDIBILITY
We argue that the increased scrutiny exercised by the respective courts in the EU and UK is motivated by a perceived need on the part of the courts to enhance the credibility of merger control. We utilize principal-agent theory to describe the role of the court in this regard. 5 First let us consider why delegation to an agency has occurred. Since merger decisions involve complex legal and economic analyses, delegation to an expert body is seen as functional from the principal's (legislature's) perspective; it neither has the resources nor the knowledge to implement the policy. 6 Another key motivation for delegation to an independent agency is the need in certain policy areas to make credible, long term policy commitments. In the field of mergers, functional independence from the principal is viewed as being of crucial importance, with a high level of autonomy given to competition agencies, and few ex post controls. 7 There is a general acceptance, and a move over time, to more transparent processes and decisionmaking based on sound economic principles, giving market players appropriate signals and incentives in respect to efficiency enhancing transactions. 8 There are a number of risks to credibility. First, there are agency risks. Errors may result from a lack of resources and expertise. Rent-seeking behaviour on the part of firms may result in capture. Secondly, there are risks associated with the principal, due to an inherent problem in the political process of prioritizing short-over long-run policy preferences. 9 Independence (or autonomy) of the agency will address the second type of risk. An increase in agency autonomy, however, brings with it the greater potential for shirking or 5 Principal-agent models have been adapted and applied by political scientists to the problems of delegation. They are used to identify and illuminate the phenomena of hidden information and hidden action with the hope that, in so doing, it will become possible to design mechanisms of control which will better align the actions of the delegatee (the agent) with the preferences of the delegator (the principal). slippage on the part of the agency because it is likely to be granted more discretion and be subject to weaker ex post controls by the principal. 10 Within this framework, how should we conceive of judicial review? 11 The courts may play an important role in these regards, both controlling discretionary powers and ensuring that decisions are made in a way consistent with the legislature's intentions. 12 The principal-agent framework is, however, too simplistic given the doctrines of institutional balance and separation of powers which inform the relationship between the legislature, the executive, and the courts. As Shapiro points out, in contrast to the administrative agency, the court is not an agent in the sense of being either normatively or empirically subordinated to the principal. 13 From the normative perspective, within the constitutional traditions of Western democracies based on notions of the rule of law and the independence of the judiciary, the courts benefit from sources of legitimacy distinct from those of the legislature. 14 Where the court's role is to ensure that powers are exercised in the manner intended by the legislature (as in the UK), courts can more appropriately be characterized as acting on behalf of the principal, whereas when the reviewing function stems more from ensuring that core constitutional precepts are followed (as in Germany), the linkage between the reviewing function and legislative intent becomes weaker.
So, credibility may require intense judicial scrutiny of expert decisions, especially where the agency is otherwise subject to weak ex post control. Intervention by the court, however, may increase the potential for error, thereby militating against the credibility of the policy, and undermining the very rationale for delegating in the first place; in extremis the court may simply replace agency discretion with judicial discretion.
The exercise of judicial review brings to the fore the question of the relative legitimacy of the court and the authority under review. There are two possible sources of legitimacy. There is a distinction between a normativist perspective-the view that judicial review is primarily concerned with placing legal limits on governmental power-and functionalism-a broader conception of judicial review as a legal means for promoting and facilitating 'good government'. 15 According to the normativist position, the appropriate level of 10 Shapiro (n 6) 82. 11 With some notable exceptions, few attempts have been made to analyse the role of the court in controlling agencies from an agency theory perspective.  should be determined by the political legitimacy of judges vis-à-vis other decision-makers. This would mean ceteris paribus that the greater is the claim to democratic legitimacy of the agent, the greater will be the deference shown to it by the court. On the other hand, the functionalist perspective sees the appropriate degree of intensity as dependent on the institutional capacity of judges (ie, their relative expertise vis-à-vis the agent). 16 Following the functional approach one would expect judicial deference to increase with the expertise of the agent. As an example of the latter perspective, Shapiro posits: The more obvious it is that the judicial reviewer is second-guessing the administrative agency, and the more obvious it is that the guesses concern matters about which the agency knows far more than the judicial reviewer, the more the favourable legitimacy of the court is depleted. In the case of specialist tribunals, however, the expertise of the court is likely to be close to that of the administrative agency. 17 Simple narratives of judicial deference do not account readily for such factors as the availability of other ex post mechanisms of control and the importance of credible policy implementation. The court increasing intensity of review according to the distance between the agent and the principal (the 'democratic deficit') may be explicable from a normativist perspective, but a different (perhaps opposite) result would follow from a functionalist one. Both perspectives require some form of oversight to prevent errors and perhaps agency escape. Furthermore, where the court exercising the reviewing function is specialist in nature, less deference is required, from both perspectives.
The following points emerge. Delegation to an agency can be an effective mechanism for making a long-term, credible policy commitment. The greater is the credibility problem the greater will be the autonomy ceded to the agency, and the weaker will be mechanisms of ex post control. The courts should not be seen, however, simply as surrogates of the principal, as this characterization of the judicial role seriously underplays the autonomy enjoyed by the courts resulting from both constitutional principle and institutional separation. While the reviewing function can be characterized as an ex post control mechanism, it is not equivalent to one which is controlled by the principal. As a consequence, there is a danger of the principal's intentions being subverted. This danger is particularly acute where the court adopts an interventionist approach to judicial review thereby frustrating the principal's attempt to ensure that decisions are made by an expert body rather than by a generalist judge. This is the conundrum that the reviewing court faces.

III. INSTITUTIONAL SETTINGS FOR MERGER CONTROL AND THE ROLE OF THE COURTS
In this section, we describe briefly the institutions for merger control, placing the respective courts within those frameworks. We first explain the institutions involved in merger control and, at a more specific level, how the courts exercise judicial control over those institutions. We then explain at a general level the constitutional basis for, and underlying purposes of, judicial review in the three jurisdictions, which differ substantially. This is particularly the case for Germany, where the right to effective judicial control and the primacy given to the protection of individual rights has important implications for the levels of control exercised over administrative agencies. This sets the backdrop for more detailed comparison in Section IV of the case law and underlying principles of judicial review of merger decisions in the EU, Germany and the UK.

A. Merger Control Regimes
In all three jurisdictions, the merger control regime follows an administrative model rather than a court-based one, such as in the US. 18 All three systems adopt a two stage approach to merger investigations, with broadly equivalent timetables. In the UK, however, there is a bifurcation of authority between the two stages. The phase one investigation is conducted by the Office of Fair Trading (OFT) whereas the more in-depth investigation at phase two is conducted by the Competition Commission (CC however, this is to misunderstand the situation since the Commission is subject to important checks and balances in the exercise of its powers, in particular, judicial review by the GC. 22 In Germany, the Federal Cartel Office (Bundeskartellamt (BKartA)) has jurisdiction over mergers under the Act Against Restraints of Competition (ARC). 23 Where the merging parties (or other interested parties) make a complaint against a merger decision, in the first instance the complaint is filed with the BKartA (within one month of the decision). 24 Upon doing so the complaint is also automatically pending before the Oberlandesgericht Düsseldorf (OLG). The BKartA is still allowed to review and amend the challenged decision (although this is not a formal procedure).
The procedural and institutional framework underpinning the UK's merger regime underwent considerable reform with the enactment of the Enterprise Act 2002. Apart from exceptional cases, mergers are assessed by independent agencies against a strict effects-based test. In contrast to the EU and Germany, there is no compulsory notification of mergers, although transactions meeting the relevant jurisdictional thresholds may, subject to a limitation period, be subject to an ex post investigation. 25 Judicial review cases are heard generally by the Administrative Court but, uniquely for merger decisions, the Competition Appeal Tribunal (CAT) has jurisdiction to hear challenges to merger decisions of both the OFT and CC.
With the recent changes in the UK, there is now a limited role for political intervention across all three jurisdictions, and the merger agencies are, therefore, all functionally independent. Some important qualifications need to be made. In the EU, while DG COMP conducts the substantive assessment of the mergers, formally the final decision on whether to authorize or prohibit a merger is taken by the College of Commissioners. 26 Furthermore, the Commission is required to liaise with Member States' respective competition authorities through the Advisory Committee. 27 In limited circumstances, Member States may intervene in a clearance decision where their 'legitimate interests' are affected by the merger (notwithstanding the Commission's exclusive competence over mergers meeting the relevant turnover thresholds). 28 In the UK, the Secretary of State has limited powers to intervene in respect of 22 23 Except, of course, those with a Community dimension which fall within the exclusive competence of the EU Commission. The same applies in respect of the UK. 24 It is also possible to complain directly to the court within the deadline. 25 With the risk to the parties that the transaction may be unwound or, more probably, a remedies package be imposed subsequently as occurred in the Somerfield case (discussed below). 26 By convention, a phase one decision is taken by the Competition Commissioner. 27 ECMR, art 19. 28 ECMR, art 21(4). The non-exhaustive grounds for so intervening include public security, media plurality and prudential rules. the public interest, and where he does so, his decision is determinative (although informed by investigations conducted by the agencies). 29 In Germany, on the application of the parties, the Minister of Economics may allow a merger that was prohibited by the BKartA where the restraint of competition is outweighed by advantages to the economy as a whole or other overriding public interest grounds. 30 So at an institutional level, we see substantial convergence in the design of agencies, all of whom are generally independent and expert decision-makers. We now turn to the nature of judicial review in each jurisdiction.

B. Judicial Review in Constitutional Context
Judicial review in the EU is concerned primarily with the objective protection and observance of EU law, rather than protection of individual rights per se. 31 Article 19 TFEU, the central rule on legal protection, is framed in objective terms, providing that the court 'shall ensure that in the interpretation and application of the Treaties the law is observed.' The guiding principle of judicial review is the checking of the legality of acts of the European institutions, including the EU Commission. In addition to individuals with standing, actions can be brought by another Community institution or a Member State, underlining the Community Courts' role in promoting the legitimacy of EU policy-making, and the balance between the institutions. There is no distinction between constitutional and administrative review. While judicial review does provide for the effective protection of individual rights, the justification for doing so stems from the need to pursue the underlying goals of the Treaty.
The modern administrative law in Germany has been referred to as 'constitutional law in concrete form', 32  being first and foremost the purpose of judicial supervision of the executive. 33 The constitutional basis for the work of the courts and for all functions of the State is the Basic Law (Grundgesetz). It heavily influences decision-making and the application of law, laying down the guiding constitutional principles such as the catalogue of directly applicable Basic Rights (Grundrechte) and the rule of law (Rechtsstaatsprinzip). The principle of legality obliges the executive not to act against the law (primacy of law), and permits interference with individual rights only insofar as the law permits. Any individual that suffers from an unjustified intrusion into his basic rights by public power has the right to effective judicial protection. 34 It has been observed that German administrative law has had a significant influence on the development of European administrative law 35 on the one hand, while in respect of the theory and practice of control of judicial discretion, the German system diverges significantly from the other administrative law systems of the EU, including that of EU law. 36 Indeed, no other European country has principles governing discretion of this depth and complexity. 37 There are several reasons why German law employs such a strong model of judicial review in the control of discretion. 38 The right to judicial review, contained in article 19(4) of the Basic Law, has been interpreted to mandate expansive judicial intervention. In contrast to the EU and UK, the theory underpinning judicial supervision of the executive will admit there is only one correct answer in the exercise of discretion; it is precisely where there is room for two views that judicial review is mandated. 39 The delegation of discretionary authority is strictly controlled in Germany. First, as a general proposition, discretion is only recognized where it is expressly granted by statute. 40 This requires the use of permissive language; where such language is not used there is generally held to be only one correct 33 Schwarze European Administrative Law (n 31) 125; Nolte (n 32) 205; Y Arai-Takahashi, 'Discretion in German Administrative Law: Doctrinal Discourse Revisited' (2000) 6 European Public Law 1, 69-80, 70. 34 The same applies to domestic artificial persons to the extent that the nature of such rights permits (Article 19(3) of the Basic Law). 35 For a discussion of the influences of national administrative law on Community administrative law see J Schwarze, 'Developing Principles of European Administrative Law' [1993] Public Law 229. 36 Nolte (n 32) 196. 37 See generally Arai-Takahashi (n 33) and Schwarze European Administrative Law (n 31) 114-127. 38 Before the Second World War, in the positivist tradition there was a very strong scepticism in Germany over the efficacy of judicial review. The courts were merely concerned with ensuring that the strict requirements and bounds of the law were respected, without any pleas to substantive concepts such as fundamental rights or substantive justice. On the historical development of German judicial review, on which the paper draws, see Nolte (n 32) 198-205 and Schwarze European Administrative Law (n 31) 114-116. 39 Nolte (n 32) 208. 40 ibid 196. answer to a question, irrespective of how imprecise or ambiguous is the relevant legal norm. 41 In contrast to the EU and Germany, the UK has a relatively weak model of judicial review. Being unwritten, the UK Constitution is organized around the doctrine of the legislative supremacy of Parliament. 42 The courts have developed a body of law, centred around the doctrine of ultra vires, which is used to ensure that public bodies who exercise statutory powers do so in a way which is consistent with the enabling Act. This is commonly understood to be the central organizing principle of judicial review and it clearly resonates with principal-agent theory. In theory, when deciding a case, it is the intentions of the legislature that are determinative, there being no higher order law to which the courts can turn. 43

C. The Role of the Reviewing Courts
Judicial review of the decisions of Community institutions is carried out by the GC, with appeal on points of law only to the Court of Justice of the European Union (CJEU). 44 The legal basis for direct judicial review of an EU institution is article 263 TFEU-actions for annulment-which enumerates the grounds of review as being lack of competence, the infringement of an essential procedural requirement, the infringement of the Treaty or of any rule of law relating to its application and the misuse of powers.
In actions for annulment, the Community Courts review the legality of the institution's decision, but they do not have full jurisdiction on the merits of the case. 45 Unlike an appeal, the Community judges are neither competent to 41 ibid. Legal doctrine maintains a sharp distinction between three elements of administrative action. First, the interpretation and determination of Tatbestand (the constituent elements or definition of a provision). Second, the application of the interpreted legal norm to concrete facts (Subsumtion). Thirdly, the appraisal and determination of a certain legal effect (Rechtsfolge). In strict terms, the operation of administrative discretion is limited to the third element only, with the courts having full control over the first and second. For a full explanation, on which this article draws, see Arai-Takahashi (n 33). 42 As a consequence that the courts are incompetent generally to question the validity of an Act of the UK Parliament, subject to the limited exception where there is a conflict with EU law:  43 Although regard should be had to the Human Rights Act 1998 which provides that statutes should be construed so far as is possible in a manner consistent with Convention rights (s 3). 44 According to art 58 of the Statute of the Court, an appeal lies on the grounds of lack of competence of the General Court, a breach of procedure before it which adversely affects the interests of the appellant, or the infringement of Community law by the General Court. On appeal the CJEU's judgment may replace that of the GC, but in certain circumstances the CJEU may simply set aside the judgment and revert the case to the GC for reconsideration. 45 The Court has unlimited jurisdiction on the merits for fines (art 261 TFEU)-that is, it may cancel, reduce or even increase them-with the rationale that stricter scrutiny is desirable where sanctions are imposed. reassess the facts underlying a Commission decision, nor substitute their decision for that of the Commission. The review is based on error-the decision can only be annulled if the Commission committed an error of fact, law or assessment. 46 The effect of a GC annulment is to revert the case back to the Commission for a renewed decision. 47 In the instance of mergers, the concentration is subjected to reinvestigation (based on current market conditions). 48 The timetable starts again from the date of the judgment. 49 Both the GC and CJEU are generalist courts. Due to the number of competition cases which come before it, however, the GC could be viewed as a de facto specialist tribunal with increasing economic expertise, 50 particularly as it is configured in chambers, which in practice means that certain judges may develop specialized knowledge.
In Germany, if a merger decision violates procedural or substantive law 51 and, thus, the complainant's rights, the OLG is empowered to strike it down. Both the legal nature of the merger complaint procedure and the competence of the courts are somewhat unusual. 52 The decisions made by the BKartA correspond by their nature with administrative acts (Verwaltungsakte) and can be eliminated if the individual affected initiates an action for annulment (Anfechtungsklage). 53 Interestingly, the complaint procedure takes place before the OLG, a civil rather than an administrative court, including three specialist competition chambers, with exclusive jurisdiction over merger complaints. 54 Decisions of the OLG are appealable on points of law to the Federal Court of Justice (BGH) where all competition law cases are assigned to the cartel division (First Civil division).
The complaint procedure is not so much aimed at an objective control of accuracy, but serves the protection of individual rights. It is characterized by 46 The GC has full jurisdictional control over errors in the proof or accuracy of facts ( the investigation principle, unusual before civil courts but similar to the strict judicial review in administrative courts. The OLG reinvestigates the facts of the case, acting on its own initiative. 55 If the court finds a prohibition to be unlawful, it cannot remit the case to the BKartA or replace the original decision. Thus, the merger is permitted by default. If a clearance decision is set aside (in whole or in part), the time limit for the merger assessment begins to run anew from the time of the court's final ruling. 56 While the UK CAT is a specialist tribunal, the substantive grounds on which challenges can be made to merger decisions are expressly limited to the grounds of ordinary judicial review. 57 Judicial review in the UK is often characterized as essentially an interpretative process according to which judges are required to apply Parliament's will as it is expressed or, more often than not, implied in the words of the relevant statute. 58 The grounds of review can be summarized 59 as being: illegality (whether the public authority has acted within its jurisdiction and directed itself properly in law); fairness (whether the decision has been made in a fair way, having regard to the nature and seriousness of the decision vis-à-vis the individual concerned); and 'irrationality' (or 'Wednesbury unreasonableness') (whether a decision is so unreasonable that no public authority acting reasonably could have arrived at it). The latter ground, while concerned with merits, it is normally an extremely difficult ground of review for a claimant to make out, especially where the decision in question concerns matters which are paradigmatically for the executive. Reflecting the limited nature of the court's supervisory jurisdiction, when a decision is found by the courts to be unlawful, the decision is remitted to the original decision-maker.

IV. JUDICIAL REVIEW OF MERGER DECISIONS IN COMPARATIVE PERSPECTIVE 60
The case law analyses in this section are structured around three key questions. First, we investigate the extent to which the courts in the three 55 ARC, s 70(1). While the parties decide on the subject matter of the dispute the court is not bound by the wording of the complaint and may carry out all necessary investigations in order to establish the facts. 56 ARC s 40(6). 57 Enterprise Act 2002, s 120. 58 Since the courts are concerned with upholding the (express or implied) intentions of Parliament, judges are said not to draw upon principles or values exogenous to the legislation concerned. It is also the case that there are no fundamental individual rights or guarantees which can be upheld in the face of a conflicting Act of Parliament. 59 Council for Civil Service Unions v Minister for the Civil Service [1985] AC 374, 410-411 (Lord Diplock). The heads of review are brought within the doctrine of ultra vires by statutory implication: for example, in respect of irrationality or fairness it is implied that Parliament does not intend for powers delegated by it to be exercised in an irrational or unfair way. It is important to note that the grounds of review-themselves based in common law-should not be viewed as static. Rather, they vary over time and have tended to expand with increased judicial activism. 60 This section does not offer a complete analysis of all merger review cases. In comparison to both the EU and UK, the number of cases in Germany is relatively large. Since the introduction of jurisdictions show deference (if at all) to merger agencies in the interpretation of legal norms. Second, we investigate the extent to which determinations of fact are controlled by the reviewing courts. The third question involves the exercise of discretion, including economic assessment. At the outset, we note the conceptual problems in defining the boundaries between these three types of determinations. Paradigmatically discretion implies that agencies have a choice between alternative courses of action. Discretion may, however, be present in the interpretation of legal norms and in factual determinations. 61 We return to these issues in the discussion of the case law below.
A. Law: The Interpretation of (Indefinite) Legal Norms The application of rules may have a discretionary element; they have to be defined in respect of both meaning and relevance (the rule's applicability). 62 In this regard, Dworkin makes an important distinction between discretion in the strong and the weak sense. 63 In the present context, this distinction is highly relevant to the issue of agencies' interpretation of legal norms. If discretion exists in the strong sense, then the agency will be given some degree of deference in deciding how a statutory term (or rule) should be interpreted. Conversely, if discretion exists in the weak sense, then the courts will hold that there is one correct answer, to be determined ultimately by the court, to the question of the meaning of the legal norm (the 'rightanswer thesis'). 64 The distinction between rules and discretion is not straightforward, therefore, and we observe a divergence of approach across our jurisdictions.
In respect of questions of law, neither in the EU nor in Germany is there any deference shown to the merger agencies. According to EU law, there is no margin of appreciation left to the Commission in its interpretation of either In the strong sense, the decision-maker sets the standard(s) according to which discretion is exercised, whereas in the weak sense, the decision-maker is merely interpreting and applying a standard set by an external authority, for example, by the courts or the legislature. There is a further weak sense, where the decision maker's decision is regarded as final, and cannot be reviewed and reversed by any other official. For a critique see Galligan (n 61) 14-20. 64 Galligan posits that veracity of the strong/weak distinction depends in turn on the rightanswer thesis, which is difficult to maintain in the context of a mature and complex legal system. (Galligan (n 61) 16-17). the EU Treaty or secondary legislation (such as the ECMR). 65 Even in respect of highly specialized terms, such as the meaning of 'collective dominance' for the purposes of article 2 of the ECMR, the GC in Airtours specified in precise terms the criteria which the Commission must apply. 66 This is despite the fact that the concept of collective dominance-or coordinated effects-is based on game theoretic principles the interpretation of which, it might be argued, requires a high level of expertise.
In Germany, there is no discretion in the interpretation of legal provisions, even in respect of ambiguous legal terms. 67 The strict control of statutory terms is mandated by the (constitutional) right to effective judicial review, making the approach particularly resistant to change. 68 The rigour of the court's scrutiny requires the court to intervene in highly specialized areas, resulting in a shift of the locus of decision-making from the agency to the court. 69 In the UK, as with other common law systems most notably the US, the general approach has been to allow administrative agencies some discretion in the interpretation of statutory terms, including in competition cases. 70 In the South Yorkshire Transport case decided under antecedent merger legislation-the Fair Trading Act 1973-the House of Lords warned against the dangers of 'taking an inherently imprecise word, and by redefining it thrusting on it a spurious degree of precision'. 71 Similarly, in IBA Health v OFT, 72 the Court of Appeal chided the CAT for trying to 'rewrite' section 33 65 Vesterdorf (n 22) 12-15. As a leading text puts it: 'The Community Courts will normally treat the meaning of terms such as State aid, worker, services, goods, capital, agreement, and other such provisions as questions of law. Their general approach is simply to substitute judgment on these questions of law for that of the initial decision-maker. The CJEU or GC will lay down the meaning of the disputed term, and if the Commission interpretation is at variance with this then it will be annulled.' (P Craig and G de Burca, EU Law: Text, Cases and Materials (OUP, Oxford, 2008) 569). 66  The bright line distinction between the meaning of a legal norm and its application has been questioned. There are two core arguments. First, it has been recognized that to maintain that the interpretation of a legal norm will only yield one correct answer is often a 'fiction'; 75 the use of open-textured, ambiguous and incomplete statutory provisions in the delegation of administrative powers must leave open some 'margin of appraisal' on the part of the agency in the interpretation of a statute. While questions of law are normally for the courts, and the courts are the final arbiters of what the law is, an important distinction should be made between administrative law where the agency applies the law in the first instance, on the one hand, and civil or criminal law, on the other, where it is the court which is applying the law de novo. This leads to the second argument. Specialist agencies are often in a better position than the generalist courts to resolve ambiguities and interpret technical, legal terms. The interpretation of a statute in a way which promotes effective public policy may depend more on the expertise of the agency rather than the limited knowledge and modes of reasoning employed by the courts. 76 In comparing the approaches of the courts, regard should be had to the important differences between the UK's common law system with that of the EU and German codified systems. First, in respect of interpretation, the UK courts tend to adopt a literal approach, whereas the Community Courts tend to adopt a purposive/teleological approach. 77 Second, in this area, the constitutional setting of judicial review appears to be vital. In Germany, the control of administrative discretion mandated by the constitutional right to effective review, appears to produce the somewhat counter-intuitive result that the more complex a legal norm, the more it should be open to challenge. 78 73 In somewhat tortuous language, section 33 provides: 'The OFT shall . . . make a reference to the Commission if the OFT believes that it is or may be the case that . . . [the merger] may be expected to result in a substantial lessening of competition within any market or markets in the United Kingdom for goods or services.' (Emphasis added.) Section 22 makes provision for an identical duty but where the merger in question has been completed. While both are framed as duties, the language used necessarily involves an element of discretion. 74 77 This difference of approach in the interpretation of statutes tends to be linked to differences in the form of legislation; UK statutes tend to be more detailed and complex, whereas EU law tends to be written in more broad-brush terms (for a discussion see J Bridge, 'National Legal Tradition and Community Law: Legislative Drafting and Judicial Interpretation in England and the European Community' (1981) 19 Journal of Common Market Studies 4, 351). 78 As Nolte observes, the constitutional backdrop '. . .
[h]as contributed to the paradoxical result that the traditional continental distinction between administrative courts and ordinary courts, a distinction which originally symbolised that administrative action was only subject to a The use by the legislature of indefinite legal concepts can be seen as a mode of delegation whereby the agency, especially in complex or technical spheres, is required to clarify the 'normative programme' in context. 79 This appears to reflect the approach of the UK courts. One adverse consequence of severely limiting discretion, as in the EU and Germany, might be to force the legislature to enact more detailed and complex rules in the first place, but such complexity may in turn 'paralyse . . . the ability of the administration to ensure a proper balance of interests in individual cases'. 80 The need for expertise and specialist knowledge supplies a key justification for the delegation of discretionary powers. The credibility of policy, however, also calls for transparent and predictable rules. As recent changes in antitrust laws from formal to more effects-based, economic approaches typify, an impatience with juridified mediating rules has led to an increasing emphasis on goal-based, purposive decisions: what Galligan terms the pursuit of 'substantive rationality'. 81 The source of legitimacy in the exercise of discretionary powers stems in part from 'how well they realize specified goals, not whether [decisions] have been made by the impartial application of fixed rules'. 82 While merger law is an area one would associate with great complexity and variability-implying a highly discretionary approach-the underlying policy goal of maximizing efficiency also requires a framework of rules which are predictable and transparent. It appears that insofar as legal determinations are concerned, the courts in the three jurisdictions appear to attach different weight to these imperatives.

B. Determinations of Fact
While discretion appears to be limited to those instances where there is a choice for the decision-maker between alternative courses of action, it may also be true to speak of discretion being present in the determination of facts. The identification of relevant facts, and the interpretation of those facts in the application of the rule, necessarily involves some element of 'choice' on the part of the decision-maker. 83 Facts are susceptible to subjective interpretation given the ambiguities, and sometimes conflicting, nature of the evidence. Furthermore, discretion is exercised to the extent that the decision-maker may have to decide upon both the methods by which facts are ascertained, the weight to be attached to different types of evidence, and in deciding what level lesser form of judicial review, in Germany now has the opposite effect of encouraging specialised judges to conduct a very searching form of review' ((n 32) 205). 79 Arai-Takahashi (n 33) 77. 80 Schwarze (n 31) 276. Alternatively, there is evidence that increasing the complexity of rules may actually result in more discretion, not less. It has been shown empirically that increased complexity in rules may lead to more not less discretion (see Hawkins (n 62) 260 and the references cited therein). 81 Galligan (n 61) 76. 82 ibid. 83 Hawkins (n 62) 260.
of evidence is sufficient. 84 We discuss how the courts have resolved these issues in each of our jurisdictions. In respect of determinations of fact, review by the GC is said to be intense, apparently leaving 'no room for discretion on the part of the Commission'. 85 This is the case notwithstanding the prospective nature of the Commission's task in analysing (anti-) competitive effects. According to the Advocate General Tizzano, the issue for the court is to '. . . verify objectively and materially the accuracy of certain facts and the correctness of the conclusions drawn in order to establish whether certain known facts make it possible to prove the existence of other facts to be ascertained.' 86 On the other hand, where the review concerns 'complex economic assessments' by the Commission, '. . . review is necessarily more limited, since the [court] has to respect the broad discretion inherent in that kind of assessment and may not substitute its own point of view for that of the body which is institutionally responsible for making those assessments'. 87 Given this, there is an important distinction to make at the outset between what is a primary fact, on the one hand, and what is an assessment of, or inference drawn, from primary facts, on the other. 88 The simplicity of this distinction belies the complexity of determining the type of question being addressed. As Vesterdorf states: What is a fact and what falls within an appreciation of facts is . . . not always easy to discern. The issue must be decided on a case-by-case basis depending on the precise context. It seems to me, however, that whenever an issue involves a complex assessment which may lead two reasonable persons to disagree as to the conclusion to be drawn, we are not in the realm of pure fact, but in the realm of appreciation of facts. 89 In respect of the primary facts, the court must ascertain whether the evidence relied on is factually accurate, reliable and consistent, and also satisfy itself that the evidence is sufficient basis for the complex assessment. 90 It must also consider whether any possible omissions on the part of the Commission are capable of calling into question its finding(s). 91 In respect of the difference between primary facts and the 'complex economic assessments' of facts, there appears to be no general rule or definition 84  which has been devised by the court. Examples of questions which involve complex economic assessments may be said to include market definition, 92 the existence of barriers to entry or of dominance. 93 Other questions could fall into either category depending on the nature of the particular case. 94 So while it is an apparently clear distinction, in practice it is one which is likely to vary from case to case, not least determined both by the availability of evidence and situations where evidence of equal probative value is pulling in opposite directions. 95 In the UK, there is some conflicting case law over which is the correct test for a court to apply when reviewing determinations of fact. There is a consistent line of case law which would suggest that the Wednesbury test, and hence a high level of deference, is the operative test in this context. 96 Nevertheless, Carnworth LJ in IBA Health made clear that there is no general rule precluding a reviewing court from overturning a decision where it was clear that the original decision-maker's finding(s) of fact did not support that decision. 97 It was, in his view, unnecessary for the CAT 'to rely on some special dispensation from the ordinary principles of judicial review'. 98 These principles were sufficiently flexible to embrace the specific context in which it was operating, including specialized expertise of the Tribunal. 99 In his view, the question before the CAT-whether the evidence before the OFT was sufficient to dispel the uncertainties over the potential substantial lessening of competition 100 -was a question 'wholly suitable for evaluation by a court' involving 'no policy or political judgment . . . such as would be regarded as inappropriate for review . . .'. 101 Those authorities which were cited in support of the proposition that factual issues could only be overturned under the more restrictive Wednesbury test concerned a mix of both policy and factual determinations and were, therefore, distinguishable from the present case.  95 We return to the control of economic assessments and inferences from facts in more detail in the following section. 96  Subsequently in Unichem the CAT attempted to explain its approach to a review of the exercise of the section 33 duty to refer. It described the approach as being 'multi-layered': the first question was whether the OFT had 'properly evaluated the primary facts of the case', the second being whether, on the basis of those facts, the OFT's conclusion or belief that there was an insufficient likelihood of substantial lessening of competition was reasonable. 103 While being careful not to jettison the Wednesbury test, it observed that the reviewing function 'may properly be more intense than it would be if issues of policy or politics were involved'. 104 In sum the CAT was entitled 'acting in a supervisory rather than appellate capacity, to determine whether . . . conclusions are adequately supported by evidence . . .
[and] that all material factual considerations have been taken into account'. 105 In Somerfield, the CAT nevertheless followed the traditional Wednesbury approach in scrutinising a divestment package imposed by the CC. 106 It held that, within the confines of that test, the weight that a decision-maker attaches to a particular piece of evidence is for the decision-maker-and not for the court-unless the conclusion drawn from the evidence is unreasonable in the Wednesbury sense. 107 Given the restrictive nature of the test, and the need to determine remedial matters 'expeditiously', it held itself to be 'extremely reluctant to entertain applications . . . based on lack of or misinterpretation of evidence, unless the groundwork has been properly laid out' by the applicant, not least because otherwise the CC 'does not have a full opportunity to deal with the matter in its defence'. 108 As to the interpretation of evidence by the CC, the CAT held that where evidence was unclear or contradictory, in drawing inferences from it the CC was to be given a wide margin of discretion: As far as drawing the line is concerned, precisely where the line is to be drawn on an issue such as this is for the [CC] to evaluate: no doubt there will always be arguments in borderline cases. In our view it would need a strong case to show that the [CC] had manifestly drawn the line in the wrong place. 109 The CAT further held that the CC was entitled to look at all available evidence 'in the round' and the weight given to evidence was a matter for the CC, 110 subject to the proviso that the weight attached to any piece(s) of evidence was neither 'perverse' nor 'a manifest error of appreciation'. 111 So while we observe a move towards more intense review of factual determinations, this has been justified in two main ways. First, the new regime, with institutional separation of the merger authorities from political actors, has legitimated more intense review. This is consistent with agency theory: the new system removes an ex post control-namely the ability of the Secretary of State to depart from the agency's recommendations-justifying the imposition of a more intense ex post control in the form of a heightened standard of review. Second, the intensity of the ex post control is further justified by the CAT being an expert tribunal rather than a generalist court. Nevertheless, the CAT itself will mediate the intensity of review in recognition of the expertise of the authority under review, hence the more limited review for the CC as compared with the OFT.
In Germany, the position is quite different from that of both the EU and UK. Under section 70(1) ARC the OLG is required (as was its predecessor the Kammergericht (KG)) to reinvestigate the facts de novo and ex officio where a complaint is brought to it. 112 This duty to reinvestigate the facts is unique to the German system. However, despite the wording of section 70(1) it appears from the case law that the complaint court does not engage in a full review of facts. As has been clarified by the BGH on a number of occasions, the first comprehensive investigation into the facts must be carried out by the BKartA. 113 Furthermore, the ex officio examination of the complaint court is also constrained by the matter in dispute; only those facts relevant for the decision of the OLG are to be reviewed. 114 On that account, the OLG may rely on some of the factual findings of the BKartA. In Texaco/Zerssen the KG outlined how the task of investigating the facts is divided between the BKartA and the court and stressed that the first comprehensive investigation was to be done by the BKartA and will not be carried out by the court, 115 a practice which has been followed consistently. 116 Nevertheless, drawing the line between the court's duty to reinvestigate and its reliance on ascertained evidence has not been easy. 117 112 This is the position for administrative law generally. The requirement of re-interrogating the facts is said to be a mechanism of correcting for inequality of arms as between private individuals and better-informed public authorities (Arai-Takahashi (n 33) 71; Schwarze European Administrative Law (n 31) 124). 113 117 In Metro/Kaufhof the BGH reminded the KG of its reinvestigation duty criticizing it for relying on the BKartA's findings which did not support a conclusion regarding alternative sources of supply (BGH of 11 March 1986, KVR 2/85 (Metro/Kaufhof) WuW/E BGH 2231). The BGH pointed out that it would have been the role of the KG to investigate the structure of the demand In addition to the refinements of the de novo review of facts, the courts have also restricted the ex officio investigation duty insofar as the complainant needs to offer some substantiation to his allegations; simply rejecting the findings of the BKartA does not force a complete reinvestigation of facts. 118 The decision of the OLG not to reassess facts in the E.ON/Stadtwerke Eschwege merger was affirmed on appeal by the BGH. 119 The court stated that where a market data inquiry is carried out by the BKartA, the OLG can normally make use of the results, even if the complainant is denied access to underlying documentation and evidence. Unless precise objections regarding the concealed data have been made, there is no ex officio obligation of the court to double-check the reliability of the evidence. 120 One mechanism assisting the court in complying with its investigatory duty is the ability to call on the BKartA to gather further evidence on its behalf, another feature unique to the German system. 121 There are a number of cases where the court commissioned the BKartA with additional investigations, especially regarding markets and market shares. 122 This does raise due process concerns since in the complaint proceedings the BKartA's role is similar to that of a respondent. 123 However, as the BGH pointed out, 124 part of the rationale relates to the BKartA's superior expertise making it best suited to fulfil the task of factual investigation. 125 To guard itself against a possible bias on the part of the BKartA selecting and presenting facts to the court, the OLG acknowledged that evidence brought forward by the BKartA is not considered to have equivalent probative value to that resulting from the court's own investigation. 126 side in the relevant market even if the KG agreed with the factual assessment of the BKartA. In a later merger case the investigation of facts by the KG was found to be adequate (BGH of 07 In a number of cases, judges have referred to their own experience and knowledge in order to assess, for example, the substitution of products or consumer preferences. This appears to underline the fact that the court has full jurisdiction over facts. For example, in Texaco/Zerssen 127 the judges, considering the intensity and structure of competition in the market for light fuel oil, concluded, on the basis of their knowledge, that final consumers observe the market, compare prices and look for the best deal. The judges considered themselves being part of a relevant group of consumers and drew a parallel between their own experience (as final consumers) and the conduct of larger customers and bulk buyers. 128 In a later merger decision the BGH confirmed that a judge or member of the BKartA being among the relevant group of consumers can make the necessary appraisal on basis of his own life experience. 129 According to the BGH, the complaint judge, whose task is to assess the facts, is obliged to define the decisive market since the definition essentially depends on the (factual) circumstances in the market, a matter within the full control of the OLG. 130 By way of summary, in the EU review by the GC of factual determinations is intense, leaving little room for discretion on the part of the Commission. In justifying this approach, Vesterdorf argues that the GC was created in part for the purpose reviewing 'comprehensively and rigorously the factually complex decisions that the Commission adopts in the field of competition'. 131 There is nonetheless some room for discretion even in this area where the evidence pulls in different directions. In the UK, there appears to be some convergence, where the depoliticization of the regime has been used explicitly to justify more control over factual determinations. It is difficult, however, to marry up the subsequent case law, in particular the Somerfield decision. This might be due to the fact that the issue concerns merger remedies, where there is considerable discretion given to the merger authority (the CC) in its assessment of evidence-an explanation consistent with the judicial review of merger commitments in the EU cases. 132 127 KG of 02 July 1982, Kart 21/80, (Texaco/Zerssen), WuW/E OLG 2663. 128 ibid 2673. 129 The judges found that the BKartA was right to refer to the life experience of its members of staff because they belong to the relevant group of consumers while considering the exchangeability of niche science journals with general popular science journals. BGH of 22 September 1987, KVR 5/86, (Gruner+Jahr/Zeit II), WuW/E BGH 2433, 2437. The OLG has relied on the own personal experience of its judges in various decision and confirmed, that the BKartA was entitled to do so as well. See Part of the justification for a more intense review may be the specialist nature of the CAT, although regard should be had to the governing statute requiring that it applies the same grounds of review as the ordinary (generalist) administrative courts. Furthermore, context clearly matters. There is an importance distinction to be drawn between the OFT's role under phase I and the CC's role under phase II, in terms of the deference shown by the court. Again we note the resonance with our central hypothesis. With functional independence of the agency, the courts imply a need for heightened scrutiny of factual decisions, but the level of scrutiny is mediated by factors such as the depth of the investigation and the relative expertise of the agency.
The German system is the most interventionist, with the OLG having the duty to reinvestigate the facts de novo-although in practice this duty has been qualified with the requirement that the complainant offers some substantiation to its allegation that the factual determination is incorrect (in effect, a requirement that an error is disclosed by the complainant, close to the requirement of EU law). Nevertheless, the role of the court is an interventionist one, and some of the features of the system clearly distinguish it from the EU and especially the UK. For example, the court is required to substitute its own findings of fact on finding of a defect, rather than merely remitting the case back to authority. Likewise, the judges are entitled to draw upon their own experience and expertise.

C. The Control of Discretion
We noted in the previous section the difficulty of distinguishing between primary facts, on the one hand, and inferences drawn from facts (including economic assessments), on the other. We now turn to judicial scrutiny of the latter types of determination.
In the EU, whereas questions of law and fact can be challenged on the basis of mere error, complex economic assessments by the Commission can be overturned by the court only where there is a 'manifest error of appraisal'. 133 It is in the area of competition law that the standard of review has been most developed. The classic formulation can be found in Aalborg Portland, an article 81 EC (now article 101 TFEU) case: Examination by the Community judicature of complex economic assessments made by the Commission must necessarily be confined to verifying whether the rules on procedure and on the statement of reasons have been complied with, whether the facts have been accurately stated and whether there has been any manifest error of appraisal or misuse of powers. In the merger context, the language used has varied slightly, with the CJEU in the Kali and Salz case referring to a 'certain discretion, especially with respect to assessments of an economic nature'. 135 In consequence, when defining the rules on concentrations, the reviewing court must take into account the 'discretionary margin implicit in the provisions of an economic nature which form part of the rules on concentrations'. 136 Much has been written (which we do not rehearse here) on the recent landmark EU merger review cases and an apparently heightened standard of review relative to the manifest error test. 137 Far less has been said about the institutional reasons for respecting this discretionary margin in the first place. One rationale of this margin of discretion derives from the constitutional role of judicial review in the EU legal system and the core principle of institutional balance. Arguably this principle denotes a non-hierarchical relationship between the Commission and the Court. 138 As Advocate General Tizzano stated in Tetra Laval, '[t]he Court may not substitute its own point of view for that of the body which is institutionally responsible for making those assessments'. 139 A second and related rationale for this margin of discretion is that DG COMP has greater economic expertise than the Community courts. The GC had previously explicitly recognized in Petrolessence that it should not substitute its own assessment for that of the Commission. 140 Likewise, in that case, the GC reiterated the need for the claimant to demonstrate a manifest error. 141 The trio of annulled prohibition decisions in 2002 in Airtours, 142 Schneider 143 and Tetra Laval, 144 do appear to demonstrate the GC's willingness to more closely scrutinize the Commission's treatment of economic evidence, particularly when it uses novel economics theories to reach its conclusions. 145 In Tetra Laval the CJEU appeared to approve of this approach, being careful to distinguish between the complex economic assessment itself, over which there is a margin of discretion, and the Commission's 'interpretation' of evidence of an 'economic nature': Not only must the Community courts, inter alia, establish whether the evidence relied on is factually accurate, reliable and consistent but also whether the evidence contains all the information which must be taken into account in order to assess a complex situation and whether it is capable of substantiating the conclusions drawn from it. 146 Following that judgment, there appears to be a heightened standard of review in cases such as MCI, 147 GE, 148 and Impala 149 (although the latter has been curtailed by the subsequent Sony BMG appeal in the CJEU). 150 The GC's reasoning in Tetra Laval led the Commission to protest that the test was no longer whether the applicant had established that the Commission had committed a manifest error of assessment, but whether it had cast sufficient doubts on the convincing nature of Commission's case. 151 In Schneider, the court considered that it could uphold a Commission decision even if it was incomplete, as long as it enabled the GC to conclude that the concentration was incompatible with the common market, which appears very close to substituting its decision for that of the Commission. 152 Most recently, in Sun Chemical Group, the GC engaged in some degree of economic assessment itself in deciding whether the Commission had followed its horizontal merger guidelines. 153 We note, therefore, an increased intensity of review in relation to economic assessments made by the Commission. In one sense this may be consistent with our hypothesis-that the increasing discretionary nature of merger lawwith a move from a formal to an effects-based approach-increases the need for the court to more intensively scrutinize the Commission's decisions. Two points are of particular note. First, the GC itself is arguably a more expert body than the CJEU-the latter espouses the 'manifest error' standard and it is more concerned with the systemic coherence of EU law as a whole, aligning it with other policy areas. Second, it is also the case that the standards which structure the way in which the decision is made, while being extra-legal, are derived from economics and are therefore capable of objective justification. 154 As a general proposition, discretionary authority is strictly controlled in Germany. In scrutinizing the exercise of discretion-once the legal norm(s) and facts have been established-the courts apply a proportionality test which ensures that the individual affected is not placed under an excessive burden. 155 In applying this test, the agency is given a certain margin of discretion (Ermessensspielraum) which varies according to the context of the decision in question, from deferential to meticulous and rigorous scrutiny. 156 This is the only element of a discretionary decision where the court will entertain the possibility of a range of (legal) outcomes.
In theory the BKartA, as an administrative agency, enjoys a margin of discretion (whenever the law provides for it), and prospective merger analyses necessarily involve some element of judgement. The words of the substantive test, however, appear to leave little room for discretionary judgement. 157 It is only where the undertakings concerned offer commitments, which would allow a clearance subject to conditions, that the BKartA can exercise discretion on the merits. 158 In cases where discretionary power is given to the BKartA, section 71(5)1 ARC states that: The decision shall also be inadmissible or unfounded if the [BKartA] has improperly exercised its discretionary powers, in particular if it has exceeded the statutory limits of its discretionary powers or if it has exercised its discretion in a manner violating the purpose and intent of this Act.[. . .] 159 154 Although the freedom of action of the agency is constrained by the need to produce a convincing theory of harm if the merger is to be prohibited, economics is not an exact science to the extent that the choice of expert may well have a bearing on the choice being made. 155 Arai-Takahashi (n 33) 74; Nolte (n 32) 201-202. 156 Arai-Takahashi ibid. According to Nolte, the proportionality test is applied in an overextensive manner by the courts ((n 32) 202 and the cases cited therein). The exercise of discretion may also be controlled on the grounds of failure to act, excess, or abuse of discretion (Arai-Takahashi (n 33) 74-75; Schwarze (n 31) 277). 157 Section 36(1) ARC provides: 'A concentration which is expected to create or strengthen a dominant position shall be prohibited by the Bundeskartellamt unless the undertakings concerned prove that the concentration will also lead to improvements of the conditions of competition and that these improvements will outweigh the disadvantages of dominance.' 158 Immenga and Velken, 'section 40 para 40' in Immenga and Mestmäcker (n 52). 159 Section 71(5)2 ARC limits judicial control by stating that the appraisal of the general economic situation and trends shall not become subject to judicial review. This is addressed to macro-economic questions which fall to the government to determine. Otherwise, the court would interfere with political decision making and constrain the scope for governmental evaluation (Heinrich Halbey, 'Kartellbehörder und richterliche Kontrolle-Zur Auslegung des · 70 Abs. 4 GWB' (1968) Wettbewerb in Recht und Praxis, 349, 352).
The correct reading of this section is contentious. Some identify a clarification of the common administrative law doctrine whereby the courts will only control whether the authority has 'overstepped the boundaries' of its discretion. 160 If the administration has the legal power to choose one measure out of a number of distinct but legally permissible alternatives, it is not for the court to overrule this decision simply because another alternative would be superior from the judge's point of view. 161 However, according to others, the section's effect is to give the OLG full jurisdiction over decisions of the BKartA. 162 Accordingly and relying on this provision, the courts have gone beyond administrative law doctrine and thoroughly scrutinized merger decisions including the BKartA's substantive assessments from the beginning of merger control in 1973. 163 That includes the establishing of legal norms-the so-called margin of appraisal for indefinite legal concepts-and 'real' discretion regarding the legal consequences. 164 By way of example, in one of the first merger decisions, the KG comprehensively reviewed the assessment of facts (albeit incorrectly in substance as was ruled later by the BGH 165 ) replacing the BKartA's market share computation with its own calculations. 166 A margin of appreciation for the BKartA, especially concerning economic questions such as definition of markets, apparently did not and does not exist. 167 It has been argued that the special administrative procedure under the ARC aims at a thorough review of decisions in order to counterbalance the wide discretion stemming from imprecise terms in the statutory provisions. 168 This is notwithstanding the fact that the court also strictly controls the interpretation of ambiguous legal norms.
According to the orthodox theory, in the UK judicial review is not concerned with the merits of a decision. Indeed, were this otherwise it is often said that a court would be acting unconstitutionally by substituting its view for that of the public body (a supreme) Parliament chose to make the decision. 169 Furthermore, in respect of economic regulation (including competition law), the courts have traditionally shown agencies a high level of deference. 170 As was discussed previously, the approach in the merger context has shifted significantly, partly as a result of the institutional and procedural reforms under the Enterprise Act 2002. While section 120(4) of that Act specifically limits the jurisdiction of the CAT to that of judicial review, 171 in the IBA Health case the CAT was attempting to rely on its status as a specialist tribunal as permitting it to depart from the standard Wednesbury test governing the exercise of discretion, 172 stating that the test was one simply of (objective) 'reasonableness'. 173 While the Court of Appeal did not approve of its attempt to rework the Wednesbury formulation, 174 it held that this was not the correct test for factual determinations, in respect of which the court was entitled to ask whether they were 'reasonable'. 175 In other words, with respect to questions of fact, the court's jurisdiction was not circumscribed by the Wednesbury test. Irrespective of the use of subjective language, the absence of a political or policy dimension rendered the question as being one of fact and, in regard to such questions, there 'is no doubt that the [CAT] is entitled to enquire whether there was adequate material to support that conclusion . . . .' 176 The real significance of the case is that the Court of Appeal read down the application of the Wednesbury test by defining 'discretion' in narrow terms, ie, applying only to decisions with a political or policy dimension. There followed some confusion in subsequent cases over the nature of discretion. For example in Unichem v OFT the CAT held that for a decision without a policy or political dimension, the use of the word discretion with its 'connotation of policy' was not appropriate; 177 the 'more correct concept is one of a margin of judgment'. 178 Subsequent case law demonstrates that the level and type of scrutiny depends very much on institutional factors and the need to ensure that the court's reviewing function does not undermine the implementation of the policy. The CAT had an opportunity to revisit the appropriate standard of review in Somerfield v Competition Commission. 179 In this case, the review was against a decision of the CC, rather than the OFT, and concerned the imposition of merger remedies. 180 Looking to the relevant statutory provisions under which the Commission was operating, the CAT concluded that it was entitled to a 'clear margin of appreciation' in deciding what action was 'reasonable and practicable' for remedying the substantial lessening of competition. 181 It also stressed the 'considerable experience' that Commission members had 'particularly in business and commercial matters' and its entitlement 'to bring that experience to bear when reaching judgments on particular matters'. 182 Similarly, in Stericycle v Competition Commission 183 the CAT were faced with a challenge to the CC's decision to require the appointment of a hold separate manager as an interim measure during the course of its inquiry. 184 The powers were expressed in broad terms allowing for the intervention where actions of the firm 'might prejudice' the outcome of the reference, signifying a 'relatively low threshold of expectation' that the ultimate remedies might be impeded. 185 The Commission was entitled to intervene, therefore, where it considered there was 'at least some risk of that happening'. 186 While the CC must exercise its interim powers 'reasonably and proportionately' it did enjoy 'a considerable margin of appreciation'. 187 The task for the Tribunal was whether the appointment of a hold separate manager 'was within the range of decisions open to it as a reasonable decision-maker'. 188 This is a well known reworking of the Wednesbury test. Furthermore, the fact that alternative (perhaps less restrictive) remedies were available to the CC did not render the decision unreasonable. 189 This would appear to distinguish the test from that of proportionality.
179 Somerfield (n 106). 180 The case concerned the terms of a divestiture remedy which the CC imposed upon Somerfield after it had acquired a number of stores from a competitor supermarket chain. In particular, Somerfield challenged the CC's decision requiring the divestment of acquired rather than pre-existing stores, arguing that both would have the effect of restoring the status quo ante.  [149]. 189 ibid.
In the area of discretion, in the strict sense of a choice between alternatives, we note evidence of a strong convergence of the UK system towards that of the EU. The case law is somewhat contradictory. Initially, the Wednesbury test was limited to those decisions with connotations of policy. The change of approach has been justified explicitly by the procedural reforms made under the Enterprise Act 2002. As Carnworth LJ noted, the Act has led to the cessation largely of the role of the Secretary of State and the abandonment of the 'public interest' test in favour of an explicit competition-based test. 190 This judicial recognition of the depoliticization of merger control, with the consequence it has for an increase in the judicial scrutiny of merger control, is the most significant aspect of the case, at least from our point of view. Whereas under the previous regime, the decisions in merger cases were taken ultimately by a politician, now the decisions of the independent competition agencies are determinative. This justifies the court tightening the ex post control it has over the authorities, a result consistent with agency theory. At the EU level we see evidence of an increased level of scrutiny in merger cases. While there has been no comparable shift in the locus of decision-making (as in the UK), one argument that has been used to explain the increased levels of scrutiny of the court has been the need for credible decision-making, especially in the light of a move towards a more effects-based analysis. The German system appears to be the most stable and is the one where the courts have the most control over merger decisions. This is more likely explained by the constitutional foundations of judicial review in that jurisdiction, in particular, the central importance of effective judicial protection of private rights.

V. CONCLUSIONS
The central motivation for this article is the apparent increased intensity in judicial review of merger decisions, despite the fact that this policy area has become more complex, requiring more (particularly economic) expertise. We would therefore expect more, not less, deference to be shown by the courts to agencies. Simple accounts of juridification fail to fully explain this phenomenon. 191 At least for the UK and the EU, we consider that the increased scrutiny is better explained by agency theory.
The apparently counter-intuitive increased level of scrutiny in the EU and the UK is not without a credible explanation, taking into account the particular need for ex post controls of independent agencies in order to promote the credibility of policy and decision-making. For the EU we would associate ourselves with the observations of Gerber who claims that the relationship between the European Commission and the EU Courts has shifted over time. 192 He traces an apparent shift from an integration imperative during the foundational period of the Community, where the Court is characterized as 'saviour' of the European Commission, to a more recent critical relationship as the system has matured, reflecting a need for increased credibility and legitimacy. The recent landmark judgments of the EU Courts evidencing a heightened standard of review seem consistent with this narrative.
What of merger review at the national level? In national systems, a similar trend can be observed. Schwarze notes an identifiable Union-wide trend towards common standards of modern administrative law among the EU Member States. 193 For example, depoliticizing merger policy by removing, or at least limiting, the possibility of ministerial intervention and approval heightens the need for alternative modes of accountability, with the courts playing an increasing role. The recent approach of the UK CAT and Court of Appeal to judicial review seems consistent with this observation. There appears to be less deference displayed than was the case under the previous system, in part justified by the depoliticization of the regime, which has the curious result, according to the appellate court, of shifting the fact-discretion distinction. In the EU, it is well known that the Commission is granted a 'margin of appreciation' in complex economic assessment, but it has been claimed, not least by the Commission, that this margin has been eroded in recent times. 194 Likewise, in the UK, the Wednesbury test for the standard of judicial review has largely been side-lined. While there will always be tensions between the need for legal certainty and the need for flexibility and expertise in the implementation of policy, this trade-off may be struck differently in jurisdictions such as Germany, where there is a strong tradition of the legal control of power in administrative law generally, and in merger law in particular. In practice, though, the courts do show some respect for the expertise of the agency by commissioning it with further investigations.
Differing constitutional bases are also highly relevant. This is particularly so in the case of Germany because of the explicit significance, at a constitutional level, of effective judicial protection of individual rights. In a jurisdiction such as the UK, where the role of the review court is characterized as 192  being one primarily to ensure that the intentions of the legislature are complied with, we would expect to see more deference. This is certainly the case. For example, in the UK, partly flowing from difference in the literal approach to statutory interpretation, some level of deference is shown in the interpretation of legal norms, although this depends very much upon the decisional context. With respect to the important distinction between weak and strong discretion, the UK's system appears to grant the agency some discretion over the interpretation of legal norms. In contrast, the EU Courts are increasingly willing to prescribe and specify in detail complex legal norms, and in Germany the approach is much the same. 195 Courts are often omitted from institutional analysis, despite the fact that their rulings can have important implications for the success, or otherwise, of a policy. Agency theory offers a convincing framework given the particular need for ex post controls of independent agencies in order to promote the credibility of policy and decision-making. This is not, however, without tensions. While increased judicial supervision can promote credibility, if taken too far it may actually undermine the policy by increasing the scope for errors, especially where a generalist court substitutes its decisions for those of an expert agency, or controls discretion to such an extent that 'substantive rationality' is undermined.
In addition, standard principal agent models are too simplistic to account for the role of the court. Given the importance of the constitutional context in which courts operate, in addition to principles of institutional separation and the distinct forms of legitimacy that the judiciary enjoys (as compared with the executive and the legislature), it would be wrong to characterize the courts as merely the surrogate of the principal (exercising ex post control on its behalf). A more nuanced approach is required, as this article has sought to demonstrate.