Some Thoughts about Proof in Competition Cases
Delivered virtually at the 18th Annual Competition Law and Economics Workshop
1While these thoughts are not random, they are unconnected to any particular factual circumstance.
2Numerous provisions central to the operations of the Competition and Consumer Act 2010 (Cth) (CCA) involve an action or proposed action, the causal connector “is likely to”, and the proscribed effect “of substantially lessening competition”, including:
(1) s 45: if a provision of the proposed contract, arrangement or understanding has the purpose, or would have or be likely to have the effect, of substantially lessening competition;
(2) s 46: a corporation that has a substantial degree of power in a market must not engage in conduct that has the purpose, or has or is likely to have the effect, of substantially lessening competition;
(3) s 47: the practice of exclusive dealing has the purpose, or has or is likely to have the effect, of substantially lessening competition; and
(4) s 50: a corporation must not directly or indirectly acquire shares or assets if the acquisition would have the effect, or be likely to have the effect, of substantially lessening competition in any market.
3The causal connector “likely to” and proscribed effect “substantially lessen competition” define the parameters of proof required to establish the existence or non-existence of a contravention of the statutory provisions.
4Requirements of proof involve policy choices of the legislature and the common law. The common law generally recognises two standards of proof, the criminal (beyond reasonable doubt) and the civil (on the balance of probabilities). Statutory provisions are able to reflect more nuanced policy choices. The sliding scale from more onerous to less onerous proof requirements includes:
(1) beyond reasonable doubt;
(2) on the balance of probabilities;
(3) a real chance or a real possibility;
(4) a possibility;
(5) a reasonable hypothesis;
(6) a rebuttable presumption;
(7) a non-rebuttable presumption;
(8) shifting the onus of proof; and
(9) reversing the onus of proof.
5“Likely” in the law may mean “more probable than not” or a “real chance”. In competition law in Australia the meaning of a “real chance” has been adopted. In my sliding scale above this is level 3 proof.
6This approach to “likely” is reflected in the definition of “likely” which operates in respect of the principal cartel provision, s 45AD which operates if the provision has the purpose, or has or is likely to have the effect, of directly or indirectly fixing, controlling or maintaining prices. Section 45AB provides that likely includes a possibility that is not remote. Where the formulation fits in my sliding scale remains to be seen.
7Because the statutory causes of action are civil they operate by reference to the civil standard of proof so what is currently required is that the court must conclude that it is more probable than not that the conduct in issue involves a real chance of the proscribed effect.
8A “real chance” is not evaluated on a numerical basis. It involves a qualitative assessment:
(1) a “mere possibility”, no matter how apparently plausible, is insufficient: Australian Gas Light Company v Australian Competition & Consumer Commission (No 3)  FCA 1525; (2003) 137 FCR 317 (AGL) at ;
(2) mere speculation or theory is insufficient: AGL at ;
(3) the assessment must be “commercially relevant or meaningful”: AGL at ;
(4) the test is not one of probability, a real chance is necessarily something less than a more probable than not chance; and
(5) the chance cannot be a real chance if it is merely trivial, fanciful or speculative.
9Recent cases show that adducing evidence sufficient to meet this standard of proof is not straightforward.
10Vodafone Hutchison Australia Pty Limited v Australian Competition and Consumer Commission  FCA 117 involved Vodafone acquiring all the shares in TPG. The ACCC refused merger clearance. Vodafone sought a declaration that the merger would not contravene s 50 of the CCA.
11The ACCC contended that the future state of competition without the merger was one where it was likely or there was a real chance that TPG would roll-out a mobile network, and would focus primarily on winning new customers through its aggressive pricing.
12I will make a few observations here:
(1) by agreement of the parties Middleton J adopted the view of Yates J in Australian Competition and Consumer Commission v Metcash Trading Ltd  FCAFC 151; (2011) 198 FCR 297 and that of Beach J in Australian Competition and Consumer Commission v Pacific National Pty Limited (No 2)  FCA 669 to that of Buchanan J in Metcash. Buchanan J had said that the real chance test should not be applied to s 50, as it departed from the ordinary civil standard of proof of on the balance of probabilities (at ). Yates J said that he preferred the view that a single-limbed test should be applied, “involving one evaluative judgment”, with the “real chance” standard applying “to determine the existence and interrelationship of all of those facts, matters and circumstances” which defined “the future state of affairs” (at -);
(2) the real chance test required a comparison of the future state of competition with and without the merger; and
(3) while only one future state is called counter-factual (the state not reflecting a continuation of the status quo), both future states are hypothetical because neither has yet occurred.
13Vodafone exposed that timing is everything. In Vodafone in 2017 there was a real chance that TPG would roll out a retail mobile service. But the case was heard in 2019. By that time the Court was satisfied there was no longer any such chance.
14Middleton J said the real chance was to be assessed by reference to the next five years by reference to the “circumstances that can reasonably and sensibly be predicted based upon the evidence before the Court”: . That is, the concept of a real chance did not permit the hypothetical futures to be populated by circumstances that were mere possibilities – the future was to be populated by circumstances that can reasonably and sensibly be predicted based upon the evidence.
15Middleton J stressed that in populating the future commercial entities are assumed to be commercially rational: .
16Middleton J said “[a]ll courts in the course of their fact finding function must reach a sufficient level of persuasion depending on the nature of the fact finding they are undertaking”: . This reflects the requirement for rationality – some facts are easier to predict than others.
17Middleton J accepted that TPG had a strong commitment to commence and continue a roll-out of a mobile network from 2016: . However, the limited roll-out TPG undertook involved unanticipated technical difficulties and market conditions changed between 2017, when the limited roll-out started, and 2019. Relevantly:
(1) the Commonwealth Government imposed the Security Guidance relating to Huawei;
(2) an alternative technical solution that would provide a 5G upgrade path while using TPG’s Huawei 4G network plan was required;
(3) such a solution involved numerous business risks;
(4) the board decided to cease any further roll-out; and
(5) since the time TPG had ceased its roll-out no other vendor has emerged with a solution that would allow TPG to upgrade to 5G.
18Other important considerations included:
(1) the required level of capital investment which could not be funded without further debt or equity;
(2) the inability of the TPG business to support further levels of debt;
(3) the unlikelihood of any further capital raising being an available option;
(4) the lack of any current business case for a roll-out; and
(5) the shift that had occurred so TPG would no longer be a first mover or a disrupter in the 5G market.
19To reprise aspects of the context:
(1) TPG decided to roll-out a mobile network in April 2017;
(2) until November 2018 TPG proceeded on the basis it could continue its roll-out despite the Security Guidance;
(3) the board resolved to cease the roll-out in January 2019;
(4) the ACCC’s decision to refuse merger clearance was on 8 May 2019; and
(5) the hearing was in September-October 2019.
20It is apparent that had the timing been different the Court would have had no difficulty concluding that there was a real chance, even a probability, of TPG rolling-out a mobile network and by so doing introduce substantial competition into that market. The change in circumstances occurred over a compressed period of time – in effect, November 2018 to January 2019.
21This exposes how rapidly commercial circumstances can change, altering what once appeared clear and fairly certain one month into an extreme unlikelihood by the following month.
22However, the mere fact that things can change rapidly and dramatically does not mean that any particular kind of change, inherently, is a real chance. This is because while a fact of a real chance of change can always be accepted, the directionality or substance of the change necessarily requires rational inferences from evidence. And an inference, even of a real chance, requires a sense of positive persuasion from the evidence.
23Australian Competition and Consumer Commission v Pacific National Pty Limited  FCAFC 77 involved the sale of a rail terminal referred to as the ART. The ACCC contended that the sale would contravene s 50. The basis for this contention was the proposition that the acquisition of the ART by Pacific National would be likely to have the effect of substantially lessening competition in markets for certain interstate rail linehaul services. The primary judge would have found a contravention of s 50 but for the provision of an undertaking by Pacific National relating to the future conduct of the ART.
24Middleton and O’Bryan JJ adopted the traditional approach of the common to the meaning of “likely” at . Accordingly, their Honours said that while there are strong arguments based on the statutory text for “likely” to mean “probable”, “likely” in the context of the competition legislation had been construed to mean “real chance” for 40 years more or less consistently without any suggestion of widespread inconvenience, so the common law would not respond by imposing any change in meaning irrespective of subjective views of individual judges.
25Middleton and O’Bryan JJ concluded that the primary judge erred in concluding that, but for the undertaking, there would be a contravention of s 50. They also considered that the Court could not accept an undertaking as to future conduct and then include that undertaking to populate the hypothetical future circumstances. Rather, the power to accept an undertaking is enlivened once a contravention of s 50 is found, and the Court may then accept an undertaking in lieu of granting injunctive relief as a remedy for the contravention.
26Their Honours noted the primary judge’s rejection of the notion that one real chance might be built on another real chance and so on, leading to an overall evaluation of a real chance: . This reinforces that the evaluation of the various building blocks by which the hypothetical world is constituted necessarily calls for the application of commercial rationality having regard to the nature and character of the fact in issue.
27Likelihoods are required to be evaluated in a common-sense commercial context.
28In Pacific National Middleton and Yates JJ put the same proposition this way at  in agreeing with the primary judge:
(a) the application of s 50 requires a single evaluative judgment (at );
(b) it is a distraction (and, we would add, wrong) to ask what standards of proof apply to the primary facts which will involve predictions about the future (at );
(c) however, the degree of likelihood of any particular future fact existing or arising will be relevant to the assessment of the likely effect on competition of the acquisition (at ).
29The significance of proposition (c) should not be under-estimated. It was reiterated at  in these terms:
… [the] assessment is to be made on the basis of an overall evaluation of the evidence, taking account of the significance of the predicted facts and circumstances to competition and the likelihood of such facts and circumstances occurring in the future.
30In practical terms this means that if the party with the burden of proof is seeking to populate the hypothetical future with a fact about the potential for a commercial entity to act or to be able to act in a particular way then the higher the inherent or common-sense commercial barriers to the entity in so acting or being able to act, the more cogent the evidence will be required to be to enable an ultimate conclusion of a real chance of a substantial lessening of competition.
31As Pacific National at  demonstrates, the fact that the primary judge “could not rule out” another entrant into the market emerging involved error – not because it reversed the onus of proof but because the incapacity to rule something out, inherently, involves a mere possibility and not a positive persuasion required to enable any such inference to be drawn. This distinction, between negative and positive findings, was exposed in the reasoning of Middleton and Yates JJ at  where they said:
Ordinarily, there is a difference between a negative finding that a fact cannot be ruled out and a positive finding of the existence of the fact. The negative finding conveys that the fact is a possibility. His Honour’s conclusion at  (repeating what was said at ) appears to be a finding that, if the acquisition does not proceed, there is a possibility of a real chance of new entry in 5 years. It is not a positive finding that there is a real chance of new entry in 5 years.
32Their Honours rejected the ACCC’s proposition that the evidence supported the positive finding of a real chance of another entrant within five years, referring at  to these facts:
(1) barriers to entry are high;
(2) substantial capital investment would be required;
(3) significant lead time for the investment would be required; and
(4) the investment time horizon would be long-term.
33In other words, the nature of the fact to be proved – a real chance of a new entrant into this market – required cogency of proof given the commercial reality of what would be required for a prospective entrant to persuade itself that entry would be worthwhile.
34To emphasise the point yet further that it is the particular fact sought to be proved that is determinative of the kind of proof required, Middleton and Yates JJ said at :
It should be emphasised that, in a given case, findings about the prospect of new entry may be capable of being made without identifying a specific potential entrant or direct evidence of an intention to enter. Such a finding could be supported by evidence concerning the nature of the market in question, the nature of the barriers to entry and the history of entry to and exit from the market…
35There are some similarities between the outcomes in Vodafone, Pacific National and in the case I recently decided, Australian Competition and Consumer Commission v NSW Ports Operations Hold Co Pty Ltd  FCA 720, albeit recognising that the latter is subject to appeal. Insofar as the competition issues are concerned, the regulator failed in all three cases on the basis of proof. In each case, there was at least one essential conclusion – that without the conduct in issue there was a real chance of an asserted future event or circumstance occurring or existing but with the conduct in issue that real chance was eliminated.
36Proof of the future, no matter what the standard of proof to be applied, presents distinct challenges. This is because the exercise is predictive but must be based on evidence – as noted, while all rational potential futures are possible, the requirement of a real chance or real possibility as opposed to a possibility requires the Court to be able to reach a state of positive satisfaction that the asserted future circumstance is a real and not a mere theoretical chance.
37It must also be recognised that the kind of cases likely to reach the Court are hard cases. They are cases where the issues are inherently contestable. They are likely to involve the very kinds of circumstances where evidence will need a certain degree of cogency to enable such a state of satisfaction to be reached – such as high barriers to entry, significant capital investment requirements for entry, significant lead times, and significant time horizons for returns on investments. In other words, of their nature, the subject-matter of the real chance tends to involve the kinds of decision any corporate entity would not take lightly. A market participant also probably has the capacity and information to replicate that kind of sophisticated commercial decision-making exercise in the context of litigation, and cases show that market participants have been willing to do so.
38For statutory causes of action, there is always scope for legitimate policy debate about what must be proved, the standard of proof to be applied, and the appropriateness of provisions facilitating proof.
39Whatever policy choice is made, the fundamental judicial obligation remains fidelity to the law as in force and in the terms in which it is expressed. The court’s focus on the reasonably predictable future “with and without” the conduct in issue represents a faithful application of the current statutory provisions.
40As noted, in different contexts, issues of proof of future and thus, by definition, uncertain events have been resolved in different ways.
41For common law causes of action in contract and tort which sound in compensatory damages, the courts developed doctrines to ameliorate the all or nothing effect of the common law in certain cases. The all or nothing effect of the common law is that when a court has found that, on the balance of probabilities, a fact exists or would have existed, that fact is treated as a 100% certainty. Where the contractual or tortious breach involves a loss of opportunity and the court can satisfy itself that the prospects of obtaining otherwise would have been real and not merely speculative then the requirement of causation is satisfied and the court assesses damages by reference to the degree of likelihood of the opportunity having been obtained be it greater or less than 50%. By this means: (a) the defendant in breach does not have to compensate 100% for a lost opportunity the plaintiff had only, for example, an 80% or 20% chance of securing, and (b) the harmed plaintiff does not obtain 0% for a lost opportunity that the plaintiff cannot prove on the balance of probabilities (> 50%) the plaintiff would otherwise have secured.
42The attribution of a percentage likelihood of the opportunity having been secured is undertaken recognising what, post Donald Rumsfeld, are routinely described as known unknowns rather than unknown unknowns. In competition law there is usually an abundance of known unknowns. Given the nature of the relief involved, however, a percentage likelihood approach to uncertainty is not feasible.
43In consumer law where there is a representation as to a future matter the person is taken not to have had reasonable grounds for making the representation and the representation is therefore taken to be misleading unless the representor adduces evidence that the person had reasonable grounds for making the representation: s 4 of the ACL.
44In the context of veterans’ entitlements there is a reverse onus of proof on the Military Rehabilitation and Compensation Commission to accept a claim for compensation unless the Commission is satisfied beyond reasonable doubt that the injury, disease or death are not related to military service: s 335 of the Military Rehabilitation and Compensation Act 2004 (Cth).
45Other examples of reversals of the onus of proof and rebuttable presumptions facilitating proof are apparent in a wide range of civil law contexts including: (a) adverse action by employers where the employer must prove the action was not taken for a prohibited reason: s 361 of the Fair Work Act 2009 (Cth), and (b) indirect disability discrimination where the burden of proving the requirement or condition is reasonable lies on the person who requires, or proposes to require, the person with the disability to comply with the requirement or condition: s 6(4) of the Disability Discrimination Act 1992 (Cth).
46In the context of environmental law many legislatures around in Australia have adopted a test for controlling actions of “likely to significantly effect/have a significant impact on the environment”. Given the protective and remedial nature of environmental legislation, a quality shared by competition legislation, in this context “likely” has also been interpreted to mean a real or not remote chance or possibility. In Australia and around the world most environmental legislation also gives effect to the precautionary principle. To take the Environment Protection and Biodiversity Conservation Act 1999 (Cth) (EPBC Act) as an example, the defined principles of ecologically sustainable development (s 3A) include the precautionary principle that:
if there are threats of serious or irreversible environmental damage, lack of full scientific certainty should not be used as a reason for postponing measures to prevent environmental degradation
47In making decisions authorised under the EPBC Act, decision-makers are required to take into consideration the precautionary principle.
48The precautionary principle is triggered by the threat or serious or irreversible environmental damage. If that threat exists (that is, serious or irreversible environmental damage is one of a range of possible future outcomes depending on how events unfold) then measures designed to prevent the threat eventuating are required despite both the threat and the unfolding of events being uncertain.
49It is not difficult to conceptualise a capacity for debate about a similar principle in competition law, particularly given the rapidity of development of some markets outstripping the pace of policy and legislative responses.
50The Cremer Report (European Commission, Competition Policy for the Digital Era, report prepared by Cremer J et al. (European Commission, Brussels, 2019)) said at p 4 that:
In particular, in the context of highly concentrated markets characterised by strong network effects and high barriers to entry (i.e. not easily corrected by markets themselves), one may want to err on the side of disallowing potentially anticompetitive conducts, and impose on the incumbent the burden of proof for showing the pro-competitiveness of its conduct.
51In the language of economists, this involves a policy preference for false positives (that is, regulation where no regulation is necessary) over false negatives (that is, no regulation where regulation is necessary). The proposed pre-condition is highly concentrated markets characterised by strong network effects and high barriers to entry because, as indicated, these kinds of markets are less likely to correct themselves over time.
52The European willingness to adopt a risk averse policy for regulation of certain kinds of markets has been subject to criticism, particularly in the United States.
53In an article analysing the tension between competition and innovation in markets, digital markets in particular, Aurelien Portuese (“Precautionary Antitrust: The Changing Nature of Competition Law in the Digital Era” (2022) The Journal of Law, Economics & Policy (forthcoming)), proposed that:
(1) the approach in the Cremer Report – that digital markets change the error-cost dynamic of regulatory agencies (meaning there will be a greater tendency to over-estimate the costs of excessive regulation and under-estimate the costs of non-intervention) – is unsatisfactory;
(2) it is unsatisfactory in part because it would involve “reneging on fundamental legal principles that constitute the rule of law and ensure legal certainty, such as the unreversed burden of proof” (p 15);
(3) shifting from one kind of error (the false positive of over-estimating the costs of excessive regulation) to another kind of error (the false negative of under-estimating the costs of non-intervention) is ethically and inherently unconvincing, given the objective of avoiding or minimising error (p 16);
(4) the precautionary principle in an anti-trust context is excessively risk averse and harms innovation (p 29);
(5) another tool for analysis is the innovation principle – that decision-makers should be required to take into consideration the impact of the action on innovation (p 32), including consideration of a full range of alternatives rather than only a comparison of the action with the status quo (pp 34-35); and
(6) there is a need to overcome precautionary anti-trust with innovation-based anti-trust (p 69).
54I will confine myself to these observations:
(1) the precautionary principle does not operate at all unless there is proof of a risk of serious or irreversible harm;
(2) the precautionary principle does not involve any formal legal reversal of the onus of proof – but in practice a proponent may well become subject to an evidentiary burden to disprove or mitigate a risk of serious or irreversible harm; and
(3) as noted, legislative reversal and provisions facilitating satisfaction of the onus of proof are not uncommon in response to harm avoidance legislation – in particular, in the context of statutory regulation such provisions could not be said to conflict with the rule of law.
55A US version of the precautionary principle is also being reinvigorated in the US as a result of a renewed focus on the “incipiency doctrine” based on a textual approach to the 1914 Clayton Act which extended the Sherman Act to acquisitions that may substantially lessen competition or tend to create a monopoly.
56In an article “The Merger Incipiency Doctrine and the Importance of ‘Redundant’ Competitors” (2018) Wisconsin Law Review 783, Peter C Carstensen and Robert H Lande considered that the attenuation of the incipiency doctrine had resulted in mergers which increased prices and decreased innovation (p 784). They advocate for aggressive implementation of the doctrine (p 784). They postulated that this was necessary as:
(1) there is a need for protective “redundancy” in the number of competitors in a market (p 787) as: (a) underestimating this minimum number if likely to result in medium to long-term harm, and (b) the number of competitors can be reduced by normal competitive processes and abnormal events, and (c) empirical evidence after the event shows that large mergers do not provide significant efficiency and innovation benefits (p 787); and
(2) the incipiency doctrine, in practice, means ensuring there is some protective redundancy in the number of competitors in the market (p 788).
57Carstensen and Lande proposed a rebuttable presumption against significant mergers in any moderately concentrated market. The presumption would be rebutted by evidence that the merger would not have anti-competitive effects (p 788). The authors identified five aspects of the doctrine as articulated in US case law:
(1) the incipiency doctrine prohibits even small decreases in competition: p 792;
(2) a merger should be blocked because if it could cause an industry trend or wave toward mergers: p 793;
(3) a lower probability of proof of harm will suffice for a violation of the Clayton Act than that required for a violation of the Sherman Act: p 794;
(4) the Clayton Act should look further into the future for possible harm: p 794; and
(5) merger enforcement should err on the side of over-enforcement: p 795.
58Carstensen and Lande considered that this rebuttable presumption was justified as:
(1) their analysis showed that evidence is very strong that it is hard to predict accurately how markets will function and how competitive they will remain: p 826;
(2) an unduly fragile market is more vulnerable to a variety of vagaries: p 831;
(3) conduct-oriented remedies once a firm has substantial market power are problematic: p 837;
(4) retaining a market with resilience redundancy in competition terms is more effective than attempting an after-the-event remedy: pp 841-842; and
(5) the costs of resilience redundancy in a market is minor compared to the potential harm to competition in the medium and long-term: p 844.
59Richard M Steuer (“Incipiency” (2019) 31(2) Loyola Consumer Law Review 155) also proposed that the incipiency doctrine needed to be revitalised in the US. Reviewing the history of judicial decisions Steuer observed how the concepts of “may” and “tend to” in the Clayton Act had become assimilated to the approach in the Sherman Act requiring proof of actual probable substantial harm to competition. Steuer proposed that the reinvigoration of the doctrine would not be achieved by forms of tinkering so that: (a) changing the requirement of “substantially” to “materially” was unlikely to lead to any change in approach by courts as it would not change the degree of probability required: pp 171-172, (b) a more meaningful change would be to replace the concept of lessening competition with “threatening” competition, as this would emphasise the long-time horizon for the analysis and change the requirement from a likelihood to a risk: p 172, and (c) a requirement of reasonable foreseeability of the threat could also be added to ensure that mere theoretical possibilities would be insufficient to trigger the legislation: p 173.
60Herbert J Hovenkamp (“Prophylactic Merger Policy” (2018) 70 Hastings Law Journal 43), however, proposed that the incipiency doctrine should not be applied in two cases: (a) when we are unable to predict with sufficient confidence that a certain anticompetitive outcome will occur and that it can be attributed to the merger: p 49, and (b) when the feared post-merger anticompetitive conduct is readily remedied by the antitrust laws if it should occur: pp 49-50. These propositions were derived from the view that “[m]ost mergers are lawful because they are thought to generate cost savings from economies of scale, integration, elimination of market transactions, or some other efficiency” with the consequence that “we do not want to condemn a merger based on mere speculation that it might lead to some anticompetitive outcome”: p 50.
61As I have said, determining the right policy balance is a matter for others. What I can say is:
(1) public policy, whatever its content, is capable of being reflected in statutory provisions;
(2) in civil law, standards and facilitation of proof specified in statutes reflect these policy choices;
(3) a fundamental concern of the rule of law is not the particular policy choice which is a matter for the legislature, but the degree of certainty, predictability, non-arbitrariness and transparency of the legal implementation of the policy choice;
(4) another fundamental aspect of the rule of law is the judicial obligation of fidelity to the statutory text construed in context; and
(5) the precise words used in the legislation to convey the operation of the chosen policy matter. The judicial duty of fidelity to the statutory text is deeply ingrained in our legal tradition. While the result of an exercise of statutory interpretation is inherently contestable given the nature of language, no one should doubt that courts pay scrupulous and painstaking attention to text in context.
62To return to the current statutory provisions, the one further observation I wish to make is this. In considering modes of proof, the importance of theoretical economic, as opposed to econometric, evidence should not be under-estimated. In the context of competition litigation economic modelling exercises divorced from a sound economic theoretical base are easily destroyed; even thoughtful and theoretically sound modelling exercises are open to extensive debate.
63The degree of focus apparent in evidence on economic modelling may be at the expense of critical evidence of economic theory. Evidence of economic theory (particularly if supported by academic literature and analysis involving the kinds of markets in issue) has explanatory power – it can shape and inform the rest of the evidence, it can inform the nature of the modelling which will be useful, it can explain outcomes and apparent anomalies, and it can assist in gauging the robustness of commercial evidence from market participants.
64Economic theory also has real predictive power. Evidence of what economic theory would predict in the circumstances, academic analysis of the fit between economic theory and the real world, and how those considerations relate to the nature, inputs and outputs of any modelling evidence is different from the mere presentation of the opinion of one or other economist. I am not suggesting that evidence of economic theory alone can suffice. But given the asymmetry of information that the regulator faces in comparison to market participants and the inherent contestability of economic modelling exercises, sound evidence of what economic theory would predict in the circumstances and why it would do so, even and perhaps especially if subject to fierce contest, has explanatory and predictive power that should not be under-estimated by anyone involved in competition litigation.