Fault Lines in Competition Policy
Competition rules can take a variety of forms, each displaying varying degrees of receptiveness to the input of economists. These legal taxonomies vary in detail across legal systems. Despite the differences, the following broad distinctions hold true:
a) Liability-creating as opposed to liability-denying rules.1
b) Rules in which economic evidence or analysis is constrained only by broadly-framed behavioural prohibitions of undesired outcomes, as against rules which forbid particular conduct by describing it in detail and in which economic analysis plays no part in establishing liability in a given case.2 The former are usually described as rules of reason, the latter as per se rules, an over-simplified dichotomy which is not always as rigid as this classification would suggest.3
c) Heads of liability that depend on proof of actual or likely anti-competitive intent versus those that work by way of assessing anti-competitive outcomes.4
d) Rules that are evidentiary or remedial in their effect as distinct from those that are substantive.
e) Rules that strike at collective misconduct versus those that focus on unilateral acts or omissions.5
All of these distinctions affect the design and legal relevance of economic analysis in different ways. None of them, however, distinguishes between particular economic activities or forms of investment, or between action and inaction.
The line between fact and law (never as clear-cut as common lawyers would like to think) is more blurred in competition cases than in most other areas of law. The line can become fuzzier still when common law and civilian modes of reasoning meet (or fail to meet). Thus, common lawyers, when perusing the General Court’s decision in Microsoft Corporation v European Commission (‘Euro Microsoft’)6 find it difficult to discern the rule-making wood for the myriad factual trees, where civilians (because they tend to look askance at too overt a usurpation of the legislator’s role) would see the primacy of fact-finding over rule-making displayed in the General Court’s decision as perfectly normal. This blurring of fact and law, while visible in judicial decisions on both sides of the Atlantic, tends to assume different forms in Europe and the United States. In the United States it may be seen in the use of judge-made presumptions that obscure the invisible privileging of one economic viewpoint over another. In Europe there is a tendency on the part of the courts to describe the need to interfere with private law rights as being in some sense extraordinary, a regrettable but rare necessity to be resorted to only when an unlikely combination of seldom occurring facts is present in the case before them. Both techniques have the capacity to divert attention from the underlying economic issues. Both therefore have a direct bearing on how these jurisdictions treat refusals to license.
When courts describe the likelihood of intervention as being ‘rare’ or a case as ‘exceptional’,7 they are, on the face of it, doing no more than predicting factual outcomes on a quasi statistical but wholly unscientific basis. Predictions can too readily morph into rules, however, when used to explain why the court’s approach to refusals to license intellectual property diverges from its approach to other forms of property. Nothing remarkable in this, it might be thought, just the normal mechanism by which judges make law (relatively overtly in common law legal systems these days, more covertly in civil law ones). When used in a series of cases over a short time frame, however, the technique can bestow unwarranted legal significance on the facts of the first case in which the exceptionality principle happens to be invoked. While the presence or absence of these facts may or may not have economic significance, that is not something that can be debated in later judicial decisions. The first case off the block can thus, purely fortuitously, put the economics into a legal straitjacket.
In competition law, as in other legal contexts, confusion often arises when courts do not adequately distinguish between a constitutive element of a legal rule and the way in which the existence of that particular element is to be proven on the facts. Given the seeming intractability of much economic evidence, judges are sometimes tempted to remove the particular matter at issue from the rule of reason and treat it not as a question of fact but as an unformulated question of law, employing for this purpose various ‘presumptions’ as a kind of verbal shorthand for dispensing with the need to identify the process of legal reasoning by which they have arrived at their conclusions.8 Thus presumptions are often invoked by lower courts in the United States in antitrust cases to provide an exit from the impasse of indeterminate or contested economic theory. It is not always clear, however, what judges imagine themselves to be doing in these cases. The language of presumption can be (and in competition cases often is) applied in very different ways. Historically, the use of presumptions has been a way of ‘softening’ a per se rule that has proved overly interventionist. As such it may be seen as an intermediate stage in the transition to a full rule of reason. Once the case-by-case approach is in place, however, courts will sometimes be daunted by the complexity of the factual inquiry with which they have landed themselves and seek to revive the presumption again, but this time as one of legality rather than one of illegality. Whether the presumption is one of virtue or of vice, its function in these cases is to paper over cracks in the economics.
The presumption of competitive virtue suggested by courts in relation to some kinds of transactions9 needs to be set against the general move away from presumptions of anti-competitive harm when assessing liability. It is not obvious that one form of presumption is better than the other. These presumptions of virtue seem to rest on another presupposition (often unstated): a presumption against any regulatory intervention at all. This makes them a stalking horse for the assumption that type one errors are in all circumstances to be avoided. If framed as remedial recognition of the law of unintended consequences or a general warning against drawing unwarranted evidentiary conclusions from particular types of economic activity, the presumption of virtue is harmless, if misleading. When presented as a substantive shield, however, it gives rise to two further questions: first, when is it to be applied; and, secondly, if it is, why should it operate more strongly when intellectual property is involved than in other cases?10
A frequent source of confusion in these cases is that it is not always clear whether judges are talking about the burden of proving facts11 or assessing exante a party’s success or failure in persuading a court to adopt its submissions as to the content or applicability of a legal rule. This is wrong-headed. Courts and regulators must be assumed to know and accurately apply the law. There is (or should be) no such thing as the onus of proving the applicability or content of a legal rule.12 That reasonable minds may differ as to the application of the rule, or that there is such a thing as regulatory leeway in its interpretation when it comes to be considered by an appellate or reviewing body, in no way detracts from this principle.13
Useful guidance may be found here from the way in which judges in some jurisdictions have put presumptions to work in tax cases when required to distinguish between expenditure on account of revenue and outgoings of a capital nature. Here, courts have by and large insisted that any presumption emerges a posteriori from proven facts rather than laid down in futuro. Used in this way, the term ‘presumption’ is merely a way of tracing the manner in which the decision has been arrived at and is not necessarily of assistance in other circumstances.14 A clear distinction is thus made between judicial observations that the presence or absence of a particular fact or factor resulted in the case being placed in a particular legal category, and the very different situation obtaining where that presence or absence is presumed without investigation or factual inquiry.
The real difficulty with presumptions comes when they are used not as a short-cut for establishing what economists can agree on but in order to mask the lack of consensus among them. When used in this way the shorthand of presumptive liability or non-liability ensures that the party bearing the onus of proof will lose.15 Given the indeterminate nature of much of industrial economics, this means that policy is being upended by process, all the worse for being done in an unexplored way. A more rational basis on which to construct a presumption would be to ask who is most likely to have the information necessary to prove the anti-competitive harm under investigation16 (a mode of allocating the evidentiary burden and the onus of proof on a particular issue that can be seen at work in the General Court’s decision in Microsoft17). Viewed in this way, the presumption becomes a kind of regulatory version of private law’s res ipsa loquitur. Such an approach would be more productive than simply treating presumptions as a kind of halfway house between economics as fact and economics as law, with no empirical justification as to why this particular stopping point has been chosen or why the journey has been embarked upon in the first place.
Jurisdictions differ as to the freedom with which they allow courts to create (and withdraw) presumptions (be they presumptions of vice or virtue) in competition cases. In the United States presumptions are almost entirely judge-made.18 In Australia and New Zealand the existence or absence of presumptions is generally seen as the province of the legislature, while in the European Union presumptions tend to take the form of minatory observations that departures from previously expounded case law are to be regarded as rare.19
In sum, the authors are not attempting here to point courts towards the right economic answer but rather to show that disputes among economists cannot be bridged simply by manipulating the onus of proof.20
Issues of market definition and market power and the role of product substitut-ability loom large in the refusal to license debate. Unresolved doctrinal disputes over these core concepts prove singularly troublesome when the boundaries of rights and markets are conflated, or when right holders project (or fail to project) market power across those boundaries. Equally problematic are attempts to rank markets in terms of modernity or technical sophistication when justifying (or decrying) regulatory intervention.
Markets are the fundamental construct around which competition law is built.21 Potentially anti-competitive outcomes are not assessed in the abstract but in terms of particular products or regions. Product markets are groupings of outputs (demand side) or inputs (supply side) that are substitutable for each other. Geographic markets assume an invisible boundary across which it will not pay to move or provide competing goods or services. Limitations on the jurisdictional reach of individual competition regimes (all have some) will turn these invisible lines into real borders, setting purely legal limits to market definition which ignore both the reality of free trade and the distance-defying possibilities of cyberspace. All of this can sometimes look rather like the process whereby the reach of intellectual property rights is set by limitations on protectable subject matter, territorial allocation of licences and the principle of national treatment. The resemblance is purely superficial, however. Delimiting rights is a question of law, defining markets is a question of fact. Confusing the two distorts the debate. Rights and markets coincide only by accident.22 From this two things follow. Right owners and licensees who step outside the boundaries of their right should not, from that fact alone, be presumed to be acting anti-competitively. Conversely, staying religiously within those confines cannot be assumed to confer immunity from regulatory intervention. (Neither should we ignore the often circular nature of the interrelationship between market definition and market power.23)
Any economics-based analysis of the role of intellectual property in competition policy has to take account of substitutability. On the demand side, this means that regard must be had to whether ordinary goods or services unprotected by intellectual property are as attractive to consumers as those that are so protected. On the supply side, the issue is whether competitors can invent or create around the intellectual property right within a reasonable time frame without infringing it. Hypotheticals in these cases tend to take one of the following forms:
a) Would there be substitutable alternatives even if the intellectual property right in question had never existed?
b) Is it the intellectual property right that is preventing actual or potential substitution?
Both are, of course, different ways of putting the same question, and both may be resolved by pointing to real evidence of past substitution or, more problematically, speculation as to future substitution using the SSNIP24 test and the like. What makes this particularly difficult in the case of products protected by intellectual property is the time frame. Intellectual property can delay substitution as well as preventing it outright, but constructing a model of the latter will always be easier than arriving at a formula capable of pinpointing exactly when during the life of the right, substitution becomes probable.
Compounding the confusion between right and market is the fact that more than one market and more than one set of rights may be under the regulatory microscope in a given case. Take, for example, the classic spare parts or after-sales service case, in which component or service A goes into the making or supporting of product B. The issue here is whether there are separate markets A and B or a single market AB, because consumers build the costs of servicing or spare parts into their initial cost calculation. A and B may or may not be covered by intellectual property rights, and those rights may be either wider or narrower than either market. (B may be the subject of a patent, A, a design right and AB, a trade mark.) Allegations of anti-competitive behaviour in such cases usually take the form of tying product A to product B. When this happens, either the tie itself or the refusal to allow competitors to bypass the tie may become the subject of competition proceedings. This can cause problems when the tie and the refusal are covered by different rules, and where the application of those rules varies according to whether an intellectual property right is involved or not. The tying claim may be defended by the argument that there is only a single market AB. That defence may, in its turn, be sidestepped by the argument that competitors are being locked out of the market for the now composite product by refusal to license intellectual property rights protecting either the part or the whole.
A somewhat different state of affairs arises when a formula or a piece of information (again possibly covered by an intellectual property right, or possibly not) is being used in market A to produce a good or service X, but is also capable of being utilised in market B to produce a different good or service Y. In such a case, A is variously described as the ‘upstream’ or primary market and B as the ‘downstream’ or secondary market. Market B may be actual (where the technology or process is licensed to some downstream market players but not others) or hypothetical (where, for example, the right holder is vertically integrated and feels no need to license to anyone but a potential demand for a licence does exist). In such cases, it may not always be clear whether this hypothetical market is in the upstream product or process, or the rights that protect it. There is nothing inherently implausible about a market for rights, but competition cases are rarely argued that way. Complicating matters further is the dubious concept of a submarket. In some jurisdictions, most notably the United States, courts sometimes treat these as markets in their own right.25 In other jurisdictions, a submarket merely pinpoints the source of market power within a market.26 Either approach can be misleading if the ‘submarket’ and the intellectual property right are given the same borders.
Multiple markets and multiple rights pose two quite distinct kinds of problem for courts and regulators. The first arises when market power is claimed to have been leveraged from one market to another. The second occurs when it is sought to stand this argument on its head by arguing that only when there is leveraging can a competition problem involving intellectual property arise. The first problem is a debate among economists. The second is an artificial truncation of that debate imposed by lawyers.
(a) The leveraging debate
The form of words used in competition legislation to strike down actual or attempted abuses of market power is open-textured enough to embrace at least the theoretical possibility that market power can be projected from one market into another. Sometimes, as in the Australian and New Zealand statutes, the leveraging door is expressly left open. More usually the spectre of anticompetitive leveraging can only be inferred sub silentio from the legislation’s failure to locate either the prescribed behaviour, or its intended or actual consequences within the confines of a particular market (jurisdictional boundaries apart). Whether and when cross-market leverage is sustainable or even theoretically possible is a subject on which economists differ. Chicago School theorists in particular are sceptical that vertical leverage can ever make economic sense. Why, they ask, would a firm that was dominant in upstream market A not simply extract monopoly profits in that market rather than try the chancier exercise of projecting its power into downstream market B where, as they surmise, it can extract additional profits only at the cost of losing them in market A? On this view of things, there is only one monopoly profit to be had, and it will be taken where it is most easily earned. Indeed some commentators would go further by positing that leveraging is not only pointless but will also decrease the dominant firm’s own profits overall if it engages in it.27 While generally raised in the context of tying, bundling and vertical mergers, the single monopoly profit theory may also be invoked in the context of refusals to deal or license.
Other analysts, while conceding that such self-destructive outcomes can occur when dominant firms attempt to leverage in this way, refuse to extrapolate from possibility to universal rule. There are, they say, some circumstances in which vertical leveraging of market power is perfectly rational.28 This is especially likely to be the case when the downstream market is itself not competitive for some reason.29 It may also make sense when (as often occurs where intellectual property rights are involved) the upstream market is proportionally small relative to the size of the downstream market.30 These possibilities should at least be investigated, they suggest, not simply brushed aside with a priori assumptions about perfect competition or costless entry and exit. (Indeed for some post-Chicago thinkers, leveraging can be the mechanism whereby strategic barriers are erected in the first place.31)
Other critics point out that the single monopoly profit theory assumes too easily that power in the upstream market is impregnable. If it is in fact vulnerable then firms active in market B could try to get a foothold in market A, thus giving a firm dominant in the latter market every reason to frighten them off while its power lasts, or better still bluff would-be entrants out of both markets.32 The theory is similarly flawed, such critics say, where the withheld input has dual uses33 or the dominant firm is trying to evade regulation in the upstream market.34 (An example of the latter relevant to refusals to license would be where a dominant copyright owner was seeking to avoid the application of the fair use rules, or to resile from open-source protocols to which it had previously agreed.) To focus solely on the profit forgone and ignore the wish to see off competing technologies is to address only half the problem.35
Whatever the outcome of these debates,36 two points need to be made concerning them. The first is that courts in most jurisdictions accept that leverage can occur, and nowhere has it been definitively ruled that it cannot.37 The second is that the real problem with the single monopoly profit theory is an evidentiary one. That is to say, even if one accepts that a rational monopolist will take the best and easiest route to profit, how is one to assess what is meant by profit in these cases? Bundling, tying and even refusals to deal can both increase or reduce price, or limit or expand choice depending on the circumstances. In such situations the temptation to avoid this difficulty by resorting to all or nothing leveraging theory becomes correspondingly greater as the available evidence on cost or quality becomes exiguous and ambiguous. Such temptation should be resisted. Evidential difficulties are part and parcel of the case-by-case analysis through which any rule of reason has to operate.38
The leverage debate becomes even more complicated when an intellectual property right enters the picture. It is not always accurate, for example, to assume (as Chicago thinkers do when they postulate a ‘single profit’ approach to leveraging) that the dominant firm will always be operating directly in the upstream market. Here too, it may be dealing through licensees to capture the supposed single profit. In such cases expanding output in the upstream market would detract from the margin to be extracted by artificially restricting the supply of licences.39 One should be wary of the fallacy of requiring a showing of leverage (and with it the presence of two markets) in refusals to license cases. This can encourage forensic gaming, by tempting parties to treat separate products as one for litigation purposes.40
(b) Leveraging as a condition precedent for regulatory intervention
A very different stance on leveraging is evidenced in later European cases on refusals to license.41 Here both the existence (actual or hypothetical) of two markets and actual leveraging or attempted leveraging across them have to be demonstrated before the court or regulator will even begin to look behind the refusal. Not only is leveraging rational in this scenario, its exercise is a mandatory fetter on what the court or regulator can decide. This self-denying ordinance takes two forms. The first is jurisdictional, so that the regulator’s scrutiny stops at the boundaries of the right and the right owner’s powers are not interfered with (for example, the existence/exercise dichotomy that is a feature of much Article 101 and 102 TFEU jurisprudence42). The second response (and now the prevailing one under the exceptional circumstances test applied to these cases under Article 102 TFEU) is remedial. The refusal may be contested in exceptional circumstances, but any order made should not extend to allowing competitors to compete directly with the right holder in the upstream market, that is, the intellectual property right should be bypassed only to the extent necessary to ensure effective competition in the downstream market and in the downstream market alone.43
Both of these approaches are manifestations of the scope of grant theories. Empirical evidence to support them is lacking. They are purely black-letter attempts to limit what courts and regulators can do to interfere with intellectual property rights. They have no equivalents in the sphere of tangible property.
Another of the issues dividing law and economics scholars is whether the likelihood of acquiring or misusing market power is present to a greater or lesser degree in particular kinds of markets. The industries most commonly singled out for this bending (or even, it is sometimes suggested, abrogation) of rules previously regarded as being of general application tend to be those associated with the tendentiously named ‘new’, ‘innovative’ or ‘emerging’ economy44 often built around the development and application of various kinds of digital technology. Interest in the supposed ‘specialness’ of these industries has in part been driven by the positions taken by Microsoft and its opponents in competition cases on both sides of the Atlantic over the last two decades.
How different in fact are these markets from other markets, and what implications might such a difference have for competition enforcement? The first of these questions is easy enough to answer. High-technology industries do have some distinguishing features. Choice, market differentiation and product function are apt to be more important than price in such markets,45 and substitutability therefore correspondingly harder to pin down.46 They are also often characterised by network effects,47 whereby the value that individual consumers place on adopting or subscribing to a particular product or service increases commensurately with the number of consumers who adopt or subscribe to it. Network effects mean relatively high switching costs that can sometimes discourage consumers from embracing competing technology that they might in other circumstances find more technically useful and/or cost-effective.48 The resultant tipping of the market in favour of the dominant firm can be both swift and total. This in turn can lead to a high degree of path dependence, under which inefficient products cannot easily be displaced by more efficient ones. (Particularly so, when a dominant player’s standards have become de facto industry standards.49)
Putting aside for the moment problems of proving that any of these things has occurred in a particular case (and such problems can be formidable50), how should competition policy react to these where they can be shown to exist? It is at this point that the economic consensus starts to fray, and fray badly. Three possible responses may be found in the literature (and to a much more limited extent in the decisions of those few courts and regulators who have had occasion to consider the problem). There are those who believe that high-technology markets are more likely to self-correct (and self-correct more quickly) than low-technology ones, thus both requiring less regulatory intervention in the first place and making such intervention more problematical if resorted to. Buttressing these views is the notion that market power is only a temporary phenomenon in these industries if the system itself can be bypassed.51 Often accompanying these theories is the idea of an ‘innovation market’ that dwells on competition between clusters of products grouped in rival systems rather than competition within a single product market.52 Capturing the latter confers no market power. Holders of these views say path dependence can occur irrespective of any anti-competitive conduct on anyone’s part. They also point out that awareness on the part of monopolists that loss of market share can be total if consumers do switch to another system in the future will act as a moderating influence on the pursuit of monopoly profits in the present.53 Competition exists, these critics say, but it is for the market not in the market.54
Opponents of this way of thinking argue that high-technology markets need more regulation not less, because mistakes made now will not easily be undone later. These technoskeptics doubt the capacity of such markets to self-correct at all, much less self-correct quickly. They would also say, if an industry is truly dynamic, short-term and anti-competitive glitches may matter more.55 Indeed the short nature of the advantage gained may provide dominant firms with an incentive to capture the upstream market before it tips.56 In such cases competition regulators would be wise to focus on the state of affairs before the tipping innovation is brought to the market, rather than on post-innovation competition. Whichever of these competing viewpoints eventually succeeds in imposing itself on judges or regulators, it needs to be emphasised that neither of them depends on whether the technology in question is shored up by an intellectual property right. Nor are they predicated on a need to distinguish between refusals to deal and proactive tying or bundling.
Both sides in this debate ask valid and important questions of their opponents. Those questions cannot, however, be definitely answered in the absence of empirical evidence. Lacking such evidence, neither side can convincingly make a case that in technology-saturated industries the activities of dominant firms are presumptively good or bad in competition terms. Nor, indeed, could any such presumption be imposed in the absence of any agreed definition of what constitutes a new or emerging market. Nothing about these markets requires the abandonment of the ordinary techniques of antitrust analysis. Uncertainty and imprecision are a necessary, if regrettable, part of the rule of reason whether applied to the ‘old’ or the ‘new’ economy.57
There are, however, two aspects of technology-intensive industries which do merit further discussion. These are the opportunities for gaming provided by standard-setting activities, and attempts to regulate technological interoperability via intellectual property law thereby displacing the jurisdiction of competition authorities.
It has been suggested that when an intellectual property right is allowed to act as a gateway to implementing an industry standard (allowing interoperability and compatibility between products from different suppliers58), its owners have been handed irreversible market power by the mere fact of their possession of that right. In the view of some commentators,59 this means that once a particular technology is absorbed into a standard, competition from outside the standard can no longer happen. Others contend that such assumptions are based on static, simplistic, stand-alone models of how the interlocking intellectual property rights that together commonly comprise a standard interact with each other.60 Moreover, the counter-argument continues, to make such an assumption about standard capture is to ignore the dynamic nature of standard setting in high-technology industries, where one standard tends to follows fast on the heels of another so that different intellectual property owners compete with each other for inclusion.61
Problems relating to standard setting and standard capture are not confined to discordant economics. Excessive compartmentalisation within individual competition regimes can also cause difficulties. A standard may act as a collective monopsony.62 The process of competitors collaborating to forge a common technological platform may provide opportunity for collusion. Anti-competitive behaviour can also take place in relation to both setting and implementing the standard. A party may refuse to make intellectual property it already owns available to the standard setters, or may act as a ‘patent troll’, deliberately setting out to acquire patents or copyrights that are essential to the standard’s functioning. Capture may be facilitated by a captor concealing or misrepresenting the existence or extent of its intellectual property rights tied up in the standard.63 Once captured, the captor may refuse to license, license selectively or license on non-RAND64 terms. If too rigid a regulatory line is drawn between these activities, outcomes could be perverse and gaming encouraged. Privileging either pre- or post-standard setting refusals to license, while punishing the act of acquisition or post-capture royalty setting, will only encourage parties to steer for the artificially safe harbours created by such privileges. Privileging unilateral bad behaviour over collective misbehaviour will have a similar effect.
There is no dispute that the ability of various software products in a system mutually to exchange information and interoperate with each other promotes incremental innovation, as well as exposing consumers/end-users to a wider choice of products. Indeed, in all our selected jurisdictions, copyright laws foster interoperability, either by allowing it under wide fair-dealing rules (as in the United States66) or by providing for a specific reverse engineering exception to owners’ exclusive rights.67 As the reverse engineering exception commonly operates, firm A may design a new product compatible with an existing product controlled by software owned by B, by first copying B’s program in object or machine code form in order to understand how it works and what is required to allow it to interact with other software. A will then not infringe B’s copyright, provided it does not use a substantial amount of B’s program in its own new, compatible product.
There are, however, three reasons why the reverse engineering exception under copyright law may not go very far to promote interoperability. First, some kinds of computer programs may enjoy dual protection under both patent and copyright law in some countries.68 (Certainly, the quid pro quo for patent protection is full public disclosure of the ideas underlying patented products or processes, but registered provisional or final patent specifications may be deficient or even misleading by not providing crucial information.) Secondly, copyright holders sometimes use technological protection measures (TPMs) to lock up their programs so that they cannot be lawfully reverse engineered without breaking the lock to access them.69 Thirdly, would-be second-comers sometimes require access to more information than just the object code owned by the first-comer in order to design an interoperable software product. The partial and incomplete allowance for interoperability under some intellectual property statutes should not be able to displace the jurisdiction of competition authorities.70
Debates between competing schools of economics often centre on the place of efficiency and consumer welfare in setting competition policy. For enforcement agencies and policy makers, these intra-mural disputes within the economics profession pose practical as well as theoretical questions: How can economists’ definitions of efficiency and consumer welfare be converted into legal terms of art? Are these concepts appropriate surrogates for the existence or absence of competition? Should they be exculpatory or inculpatory in their effect? Are they regulatory goals in their own right? Not surprisingly, different jurisdictions give different answers to these questions, when they bother to answer them at all.
In none of the jurisdictions surveyed in this book do courts unequivocally embrace efficiency as the sole goal of competition policy. Sometimes this is because the relevant legislation lays down broad statutory objectives that by implication exclude efficiency. In other cases, the statute recognises the importance of efficiency as a concept but grants it only a limited ex post facto role after anti-competitive purpose or outcome has been proven by other means (the so-called ‘efficiency’ defence in Canadian law and administrative authorisation of otherwise unlawful practices in New Zealand). Courts have generally been more reticent than regulators in giving efficiency a role in the absence of a legislative prompt in Europe and the United States.71
If economic efficiency were indeed to become the core concept of competition law, it would be a remarkably elusive one both in theoretical and empirical terms. Market behaviour is efficient in the broadest sense when it allows firms to achieve more or better outputs with the same or cheaper inputs.72 A definition so wide, while it might attract universal approbation, is unlikely to be immediately useful, especially in the cut and thrust of antitrust litigation. Economists have therefore sought to put analytical flesh on these bare definitional bones by positing three types of economic efficiency: allocative, productive and innovative.
Allocative efficiency exists where goods and services find their way to those consumers who value them most, as evidenced by a willingness to pay more or forgo other forms of consumption. Innovative efficiency is achieved when new products and technologies are spread throughout the economy in wealth-creating ways.73 Productive efficiency is found whenever goods and services are made available using the most cost-effective inputs and processes available under current technology. While each of the trinity has its own sect of economic believers, courts and competition regulators wisely refuse to elevate any one form of efficiency over the others. Of the three, only allocative and innovative efficiency can easily be brought to bear on refusals to license situations, and even these two do not point unerringly in a single direction.
While it might be thought that innovative efficiency should have automatic primacy when intellectual property rights enter the competition equation, this is far from being the case. Innovative efficiency is loved by all but measured by few. Indeed, it is all but unmeasurable. Thus, while one might not agree with the Chicago view that allocative efficiency is the one that matters most, it is a perspective likely to win by default in a measurement vacuum. Innovative efficiency is, on this view, a by-product rather than a catalyst. Get allocative efficiency right and innovative efficiency will follow. Innovation and technological progress are best assured by marshalling capital and encouraging it to invest in innovation.74 Broad intellectual property rights do this by pulling investment into as many derivative niches as a particular intellectual property right will stand. They are therefore to be applauded rather than feared or decried.75 Such an approach is far too simplistic, say others. Not all innovation can be assumed to be efficient. Change and novelty are not economic objectives in themselves. Efficiency gains cannot simply be equated with the breadth and longevity of a particular piece of intellectual property, still less with the owner’s assessment of its value or the cost of developing the technology which it protects. Indeed, if investment displacement theory is to be believed, over-extended intellectual property rights are themselves allocatively inefficient if they attract investors who would otherwise put their money elsewhere. Similarly, the encouraging of investment by first-comers at the expense of investment in downstream innovation and creativity is not easy to justify on the grounds of either innovative or allocative efficiency.
‘Consumer welfare’ has been described as the ‘most abused term in modern economic analysis’,76 and yet few would deny its centrality to the setting and implementation of an economically rational competition policy. To a non-economist, consumer welfare might seem to be useful shorthand for ‘anything that makes consumers better off’. For economists, however, there are three other, more targeted concepts wrapped up in this somewhat anodyne phrase: consumer surplus, producer surplus and total surplus77 (the last being usually defined as the sum of the previous two). Of the three, total surplus would seem on the face of it to be more important in setting regulatory goals than benefits adhering to consumers or producers alone. Unfortunately for the internal coherence of competition regimes, total surplus is not a banner under which economists find it easy to rally.
On this issue, economists divide into two broad camps: those who would treat gains to consumers as incidental side-effects of pursuing efficiency objectives (hoped-for outcomes but not necessary to the policy); and those for whom the fostering of consumer surpluses is the end to which competition law should work and which would thus become frankly redistributive in its effect.78 Members of the first camp justify reference to the term ‘consumer welfare’ by using a mixture of inductive and deductive reasoning. The logical chain79 unfolds as follows: All producers are also consumers of something. Benefits to producers are therefore beneficial to some consumers. End consumer A is not inherently more worthy than intermediate consumer B (or as they would put it, a dollar in the hand of a producer is no less important to the economy than a dollar in the hand of a consumer). Even if this were not the case, consumers can still benefit collaterally because they are part of the wider economy in which producer profits are spent (the ‘rising tide lifts all boats’ argument). Efficiencies which are captured by producers will therefore benefit end-users even if the price they pay for what the producers produce does not fall. Consumers also benefit when producers reinvest their profits in new or improved goods or services. Both productive and innovative efficiency are thus served.
Economists in the other camp would say that there is too much ‘might happen’ in this analysis for all the links in the chain to bear the weight applied to them without snapping.80 Better to focus only on those productive efficiencies that translate into lower prices for end-users, since these are at least measurable. Producer gains are thus relevant only to the extent they are passed on. Consumer surplus and consumer welfare are therefore the same thing.81
There is an artificial rigidity to both of the above viewpoints. Both are vague as to the time period in which gains and losses to various groups are to be accessed. A more nuanced view would balance short-term losses by consumers against longer-term expenditures by producers on research and development (while, it is hoped, at the same time, not losing sight of the fact that such expenditure neither guarantees successful innovation nor measures innovative efficiency).82 Conversely, one could accept that there might be cases in which total welfare increased even though efficiency savings were not immediately passed on to end-users, although in such cases one should be alive to the possibility that efficiency gains that are not passed on may not endure for any length of time.