THE ANTITRUST MARKET DOES NOT EXIST: PURSUIT OF OBJECTIVITY IN A PURPOSIVE PROCESS

There is no such thing as an ‘antitrust market’. Markets are merely analytical tools, which serve to structure available evidence and enable a comprehensive answer to a particular question. They do not exist as such in the real world but are figments of our intellectual imagination. In that capacity, they can be immensely useful. A pursuit of objectivity in the process of product market definition remains in vain as long as we fail to acknowledge that the utility of antitrust markets lies precisely in their reductive and purposive nature. This article makes two main arguments. The first argument is simple, yet far-reaching: antitrust market definition is useful because it is a method to enable the answer to a question. The implication is that the market is defined by reference to that particular question, rather than as an independent and neutral object. Market definition is ‘purposive’. In the context of competition investigations, this question can concern, but does not have to be limited to, determinations of market power. The second argument is that market definition, even though purposive, does not need to be subjective. Objectivity in market definition can be achieved by aspiring to process objectivity, rather than to objective outcomes.


589
will not be achieved by ignoring the functional nature of the concept, but rather by embracing and explicitly acknowledging its nature and scope. This is a much-needed call, currently underemphasized in academic discussion, which can help competition authorities (including, but not limited to, the European Commission) explain the rationale behind maintaining antitrust market definition. It is also quite timely, since the European Commission has recently embarked upon a reevaluation of its Market Definition Notice, citing efficiency and relevance (including 'transparency' and 'guidance') as evaluation criteria. 4

II. THE NATURE OF MARKETS
'In the process of reflecting the world, we organize it into entities. We conceive of the world by grouping and segmenting it as best we can, in a continuous process that is more or less uniform and stable, the better to interact with it. We group together into a single entity the rocks that we call Mont Blanc, and we think of it as a unified thing. We draw lines over the world, dividing it into sections. We establish boundaries. We approximate the world by breaking it down into pieces.' 5 Rovelli's description of human understanding of the world is an apt starting point to explain what 'markets' are. Although markets do not exist as such in the natural world-as opposed to the rocks which make up Mont Blancthey perform similar intellectual functions as the concept of a mountain. They structure the world in a manner which enables human understanding. They draw a limit which, if not entirely arbitrary, at least corresponds to a particular choice (where other choices might have been possible), and then attribute a specific label to that chosen delimitation. Mountains are to rocks, what markets are to economic life: merely, a word we attach to a group of interrelated objects or occurrences. Human beings select which rock formations belong together, and call the whole 'Mont Blanc'. Similarly, they select particular economic interactions they consider to be related, and call it a market. As discussed below, what is included in that group is wholly dependent on the humans who describe it and the relationship they want to study.
This article is by no means new in describing markets in this manner. It has long been established that the 'market' is a social construct, valuable only insofar as it allows us to describe economic, and thus human, phenomena. 6 Where this article differs is that it emphasizes this constructive nature of the market to clarify both the scope and utility of the concept. This is important, because the concept of the 'market' is used in a myriad of economic and competition law 4 Evaluation Roadmap 1. 5

Journal of Competition Law & Economics
scholarship, yet the meaning and purpose of the concept seems little clarified. 7 As such, this part of the article sets out three key characteristics of markets in general (they demarcate economic interactions; they are useful fictions; and they are particular to the enquiry), before expanding on this in the context of competition law enquiries in the next part of this article. These are three characteristics which, although they may seem evident, are underemphasized in antitrust discussions, despite being crucial to the nature and utility of the concept for the purpose of competition law.

A. Markets demarcate economic interactions
The use of the word 'market' to describe a nucleus of economic interaction, rather than a mere physical location for sellers and buyers to meet, dates back a long time. It finds its roots in economic theory 8 which studies the scarcity of resources and the human response to that scarcity. There are 'two fundamental conditions of human existence': humans tend to have unlimited wants, and there are only limited resources to satisfy those wants. 9 Economic theory studies the ways in which resources are, or could be, allocated among all individuals who need them to satisfy their wants. It describes how individuals select among competing wants, to determine which satisfaction they will prioritize, and what influences these choices. Economics is, as famously proposed by Robbins, the 'science of scarcity and choice'. 10 The forces of supply and demand-the rivalry to acquire and provide the scarce means to satisfy wants-are not physical objects. You cannot measure them as you would measure the width of a tree or the weight of a sack of flour. To enable a meaningful discussion about the interaction of these forces with each other, a framework is needed, which is delimited in time and space. This framework is the 'market'. It is an artificial construct, which draws Antitrust Market Does Not Exist 591 boundaries around different economic decisions of production, acquisition, and consumption. It is used to add structure to thinking about economic interactions, indicate which economic interactions are relevant to the analysis at hand, and enable a meaningful analysis of these interactions.

B. Markets are very useful fictions
Markets do not exist. You cannot leave your house and 'enter' a market. You might walk into a marketplace, where different vendors exhibit their wares in an assembly of stalls. Yet you would not be entering 'the market' in an economic sense. Markets are fictions, much more even than mountains are. While mountains are the result of a human process of ordering and cognition, they do refer to a collection of natural objects (a landform consisting of rocks and earth). Markets, however, do not reflect any 'natural' objects as such. Economic interactions-exchange, supply, and demand-exist only as representations of (aggregated) human activity. Indeed, 'demand' in itself does not correspond to a natural object. You may instinctively understand demandas you as an individual desire and purchase products for the satisfaction of your wants-but your 'demand' is not a physical object that can be grasped, and furthermore does not refer merely to your individual activity but to the aggregation of the desires and purchases of a very big group of people. 11 These phenomena exist only as imagined products of human activity. 12 Since markets are merely our name for a grouping of these phenomena, their existence is entirely fictive: they are cognitive abstractions of different human decisions and interactions. The market is a very useful fiction. It is exactly its lack of 'natural' reality that makes it such a useful tool for analyses of various types. Markets are the 'intellectual machinery', which allow researchers and authorities 'to work things out analytically'. 13 As such, their fictive nature makes them particularly well-suited to this task. Because markets represent a chosen way of looking at relationships between (groups of) people and their decisions, they are very flexible: markets can be defined according to what it is one wants to find out about these relationships. A market is an 'abstract mechanism', 14 drawn according to the nature of the enquiry, used to describe the meeting of a particular section of supply and demand for the satisfaction of a particular 592 Journal of Competition Law & Economics want. 15 They are structures within which actors compete with each other to obtain and provide the means they desire to satisfy that want. 16 The concept may be filled in differently, depending on the activities that are being studied. Markets are drawn according to the nature of the enquiry and will therefore align with the economic phenomenon one wishes to observe. 17

C. Markets are particular to the enquiry
Since there is no true market, the meaning and scope of the concept will vary depending on the context in which it is used. 18 Although the 'market' concept generally refers to the arena where demand and supply meet, the focus of the market depends on the subject of the study. The 'market' includes all decisions and relationships of production, acquisition, and consumption, which are relevant to the question one wishes to answer. The market concept is functional.
Although the basic understanding of a 'market' is largely similar across different fields of study-as the identification of 'arenas' of production, sale, purchase, and/or consumption of goods or services facilitated through a form of exchange 19 -the exact delineation of these activities will depend on the particular field and even the particular question asked. If one wishes to study the pricing process, the market may be defined around only those products achieving similar price levels. If one defines a market as the resources which are optimally allocated, 'optimal' allocation needs to be predefined. If one wishes to understand the social dimensions of exchange, the market may be defined around those persons whose economic behaviour and decision-making are directly related to each other. 20 The economic objects (products, natural, and legal persons), relationships, and decisions included in those varying understandings of the market concept may partially overlap, yet will not be 15

593
wholly the same. The more practical the context in which the concept is used, the more refined and narrow the concept will be. When studying the interaction of prices with demand and supply in theory, there is no need to have a very specific definition which can be used for practical applications. Thus, Cournot defined 'market' as 'the whole of any region in which buyers and sellers are in such free intercourse with one another that the prices of the same goods tend to equality easily and quickly', a notion upon which Marshall built in his Principles of Economics, 21 and has been used in a variety of scholarship since. 22 Yet, in 1952, Machlup recognized the need to have an analytical tool to frame questions of research in a manageable, focused manner. Thus, he contended that the concept of an 'industry' (called 'market' by Mason, Bain, and Stigler) served merely to 'limit the scope of problems of interdependence' of sellers. It was a tool-and nothing but a tool-to focus the analysis and rule out 'negligible' interdependence. 23 An important consequence of the functional nature of the market concept is that there cannot be one 'true' market. First, any market will vary depending on the field of study and the particular objective of a specific enquiry. Second, since any market is drawn, at a particular time to answer a particular question, including economic relationships and decisions relevant to that question, the answer is likely to differ every time the question is asked. A market place may include some sellers who satisfy a specific want, but will not include all sellers and product to whom an individual could turn to satisfy their particular want. The reality is not just that there is no physical place which corresponds to a 'market' in any economic sense; it is that there will never be any collection of sellers and products which could be identified as being the 'one true' economic market. Wants, tastes, desires, and needs are as fickle as the weather.
This has particular implications for calls for 'objectivity'. Since the result of market delineation will differ each time it is perform, the objectivity stems not from similar outcomes, but from a structured and transparent process. What ought to be made explicit, in the pursuit of objectivity, is the purpose of the market delineation, and perspective and selection criteria adopted.

A. Economic markets and antitrust markets
The precise boundaries of markets, which are after all merely the result of an intellectual exercise, will change depending on the context in which they are 'drawn'. The principal distinction for our purposes is the difference between 'economic markets' (the market concept as used in economic scholarship) and 'antitrust markets' (the market concept used in competition policy). Economic markets (also called 'trading markets') are usually interpreted as groupings of relatively homogenous demand, where the products are subject to arbitrage which is sufficient to maintain similar prices. 24 Antitrust markets are defined around the competitive constraints on particular economic entities, which are relevant to determining the factual possibility of particular conduct, and the legal desirability of that conduct, in the context of a specific legal enquiry. 25 The notion of economic markets was initially largely theoretical. When studying the interaction of prices with demand and supply in theory, there was no need to have a definition which could be used for practical applications. Cournot defined 'market' as 'the whole of any region in which buyers and sellers are in such free intercourse with one another that the prices of the same goods tend to equality easily and quickly', a notion upon which Marshall built in his Principles of Economics. 26 No real attempts were made to transform this rhetorical mechanism into a practical tool, for competition policy or otherwise. 27 Some economists accepted that goods ought to be included in the same market if they were substitutes, but this was mainly for statistical purposes. 28 A more practical understanding of markets was provided in the late 30s by industrial organization (IO) economists. 29 IO markets are defined by starting from the position of a single seller and identifying all the considerations, which influence his decision-making, including the desires and actions of buyers and other sellers. Mason, often said to have founded the modern field of IO, 30 stated that market of this seller 'includes all buyers and sellers, of whatever product, whose action he considers to influence his volume of sales.' 31 Thus, the market is defined by reference to all customers for the product of the single seller identified, and all the sellers he considers when making his decisions on price and output. One way to choose the other sellers to include is to identify products with high cross-elasticity with the single seller's product. 32 High cross-elasticity means that changing the price of the first product will have a significant impact on the quantity demanded of the second product. 33 The reasons for this are that the decisions on output and price made by the sellers of these products will have a significant impact on each other. 34 The IO concept of markets focuses not only on the way supply responds to demand in general but also on how the decisions of other suppliers in response to that demand limit the commercial choices available to the seller who is the focus of the analysis. The market is a device through which to concentrate on specific constraints on the behaviour of sellers. Since not all behaviour of, and not all considerations by, sellers can be studied all at once, every study has a narrow scope. Every study will, therefore, have a market defined around a narrow set of constraints. Machlup recognized the need to have an analytical tool to frame questions of research in a manageable, focused manner. Thus, he contended that the concept of an 'industry' (called 'market' by Mason, Bain, and Stigler) served merely to 'limit the scope of problems of interdependence' of sellers. It was a tool-and nothing but a tool-to focus the analysis and rule out 'negligible' interdependence. 35 Accordingly, IO markets formed the ideal foundation for antitrust markets to guide competition enforcement. Indeed, some IO economists saw their research as the basis on which policy and enforcement could rely, and explicitly considered it their aim to devise criteria to guide lawyers and judges. 36 The market concept will need to be refined to fit a certain purpose. An 'antitrust' market, that is, a market defined for competition policy purposes, will have a different focal point than, say, a market for corporate strategy. 37 The 'antitrust' market will ground the general concept within an enquiry into competition. Accordingly, it is necessary not only to determine what the core

Journal of Competition Law & Economics
features of a general 'economic' market concept are, but which economic principles are relevant to antitrust markets in a legal context in particular. Turner has acknowledged that there can be a gap between the economic market, as understood primarily by IO economists, and the antitrust market. Even where economists and competition law authorities start from a similar question-what are the substitutes for this product?-the subsequent choice of which products to include may differ, since gaps in substitution chains or time frames adopted may be more or less relevant depending on the legal issues under investigation. 38 An important reason why the antitrust market may differ from an IO market is that the purpose of defining the antitrust market is very particular: it is first and foremost a legal concept, used within an enquiry into the possibility of conduct which creates harm within the scope of the legal provisions.
Because antitrust markets are legal concepts rooted in economics, their delineation requires an understanding of the economic theory underpinning them. As Robertson puts it, 'law and economics are . . . the raw materials to build the filter of market definition'. 39 These economic foundations serve to add rationality and coherence to the process. 40 However, they are used to inform the definition of markets for competition law purposes, not replace them. Since the market is a functional concept, and competition law inquiries are of a legal nature, there can be a discrepancy between economic markets and antitrust markets in light of the differing objectives of the inquiries.

B. The purposes of product market definition
Market definition is a universal tool, adopted in many jurisdictions as a first step in assessments of the effects on competition of concerted practice, unilateral conduct, and mergers. 41 It is of significance in competition law decision-making since, beyond the mere reference to 'markets' to underpin the legitimacy of competition law, 42 it is used by competition authorities as a framework within which to solve the hurdles necessary to reach a decision. Not only are markets drawn to enable indirect measurements of market power, they more broadly provide the boundaries of the evidence needed to resolve questions of fact, conduct, and anti-competitive effects. The former is the most often cited justification for market definition: findings of market power are 38  required by law, and, since these often require the identification of markets, market definition is too. Yet this article contends that, despite the frequency with which market power is cited as the rationale for market definition, it does not need to be the principal justification for market definition. Rather, the most convincing rationale for market definition is that it provides a tool to draw the boundaries within which to asses a particular question, most prominently concerning the conduct, theory of harm, and anti-competitive effects alleged. Thus, the boundaries of the market will be drawn in function of the question which is asked. This can be called a 'purposive' approach to market definition. Even more strongly, it can be argued that market power is not even a separate purpose of market definition, but merely, one iteration of market definition's broader framing function. Although the market power-rationale and the purposive approach may appear, from their treatment in the scholarship, as two distinct views on the utility of market definition, the market power-rationale could actually be conceptualized as just 'one' iteration of the purposive approach. Market definition is useful, because it enables the identification of the forces at play which are relevant to a question, structuring them in such a way as to elucidate that particular question. This approach can include market power assessments, since a market can be defined for the purpose of identifying those forces which, at the time of the conduct, limited the undertaking's ability to profitably increase price or adopt other strategies indicative of market power. Yet the purposive approach stretches beyond the focus on market power, because it allows for the direction of the market definition towards any question the analysis aims to answer, so that it might include factors which may affect the conduct of an undertaking on a market, even if it does not obviously affect its market power at that time. In the following sections, the market power rationale and purposive approach are described separately, before turning to an analysis of the common baseline in all iterations of a purposive approach.

The market power rationale
The most commonly cited rationale for the definition of relevant markets is the finding of market power. 43 Market definition is useful to assess the competitive structure in which undertakings operate and thus establish the boundaries within which to calculate market shares. Market shares serve as indirect measures of market power. 44

Journal of Competition Law & Economics
practice and jurisprudence as filters to separate cases which involve significant market power from cases in which there is no notable market power. 45 Finding that undertaking(s) have market power is crucial, because such findings are legal requirements in EU competition law. EU competition law requires market power, to a different extent, under Article 101 TFEU (the Treaty on the Functioning of the European Union), Article 102 TFEU, and the EUMR (EU Merger Regulation). Under the EUMR, a merger will be blocked if it will significantly impede effective competition, particularly by creating or strengthening a dominant position. 46 Article 102 findings are predicated on the existence of a dominant position, which is a position of significant market power. Even under Article 101, which does not explicitly require dominance, findings of market power can be significant, since the appreciable effect on competition of an agreement can be determined quantitatively, by looking at the turnover and the market shares of the participants. 47 The General Court (GC) has held that, although market definition may be relevant under both Article 101 and 102, it is only necessary for rulings under Article 102, since under Article 102, power on a given market is the essence of the prohibition. 48 Article 101 is theoretically possible without market definition, except where its application depends on the establishment of market shares. The use of market share thresholds turns market definition into a requisite, since market shares cannot be calculated without reference to a market, as confirmed by the Court in Ziegler. 49 In summary, market power plays a role in the three types of cases, and consequently market definition does too, although to differing extent and for different purposes.
Despite the widespread adoption in decisional practice and jurisprudence of market definition to indirectly enable findings of market power, doing so is not an imperative established by the text of the law. In most cases, with exception of certain written presumptions based on market shares, 50  statutes themselves do not explicitly require 'market definition', instead referring more broadly to 'market power'. Where the law does not use market shares, it could be argued that it is not market definition, but market power which is legally unavoidable. It is conceivable, then, that where market power could be established without defining a market, this process could be avoided. 51 Since there does not yet seem to be a satisfactory method to establish market power directly, 52 rather than via market shares, authorities cannot avoid the process of market definition. Calculating market shares is only possible if one knows the boundaries of the product and geographic market in which the undertaking operates. Therefore, market definition remains,  2. The purpose of the enquiry Despite the emphasis on market power in the scholarship, the assessment of market power is not the only, or even primary, purpose of market definition. In fact, it may only be one iteration of market definition's broader framing function, under a purposive approach. As is explained at length above, markets are analytical tools to find the answer to the question around which an enquiry is framed. 56 In the context of antitrust markets, this question is not merely limited to market power. Market definition is, more broadly, undertaken to obtain a frame of reference 57 within which to evaluate the facts and theories put forward in a particular case. An antitrust market provides useful boundaries for the investigation, within which to identify the key players and factors relevant to the analysis. 58 It will put at the centre of the analysis the factors which influence supply and consumption decisions, principally the relevant competitive forces at play, structuring them in such a way as to elucidate the question an investigation tries to answer. 59 Through the definition of an antitrust market, you organize the available evidence and paint a picture of the different actors in and features of an industry, making it easier to analyze the impact certain decisions and events can have on competition. 60 This picture can then be used to make an informed decision on the alleged breach of competition law, and its consequences on that particular section of the economy, with an eye to the actual forces at play. 61 In that sense, the antitrust market stretches beyond the narrow aim of establishing market power, to provide decision-makers with the environment in which the practices are alleged to have taken place, and with the boundaries within which they can limit their analysis. This view of market definition is a flexible one: since the antitrust market is an analytical tool, it will necessarily be deployed in the manner best suited to the analysis which it is meant to assist. Critically, this means that the relevant antitrust market will depend on the particulars of the case at hand, varying 56  with the analytical purpose for which the market is being defined. Such a view of the antitrust market can be called a 'purposive' approach, 62 since it sees the market not as a static reality but as a process which alters depending on the purpose for which it is used. This 'purpose' is that of the specific enquiry, rather than some broad 'goal' of the laws themselves. 63 The antitrust market is an analytical tool, and thus, its use will be determined by the question the enquiry seeks to answer. Although the articulation of a purposive approach to market definition is most evident in Australian scholarship and jurisprudence, 64 such an approach is not alien to scholarship and enforcement practice in Europe and the USA. 65 In 1977, Breyer, when reflecting on the notion of the word 'market' in competition law for an Australian audience, portrayed the 'market' as a tool to give effect to the particular purpose of the competition law provision. Thus, he argued, 'markets' are not objective realities, but flexible means to assess the conduct at hand by reference to a specific aim. 66 In 1980, Turner explained that the antitrust market fulfils a 'dual role': first, it provides a 'rational economic basis for assessing the consequences of a particular kind of conduct', and only secondly can it be used as an indirect means of assessing market power-a role for which Turner felt market definition may not always be well-suited. 67 To this day, American scholars have asserted a role for market definition beyond market power assessments. Werden responds to the depictions of market definition as 'useless' and 'arbitrary', by reflecting on the utility of the antitrust market for the identification and delineation of the competitive process allegedly harmed. 68 Salop argues that antitrust markets

Journal of Competition Law & Economics
can only properly be defined 'in the context of the alleged anticompetitive conduct and effect'. 69 Glasner and Sullivan refer to several objectives of market definition other than the calculation of market shares, including imposing analytical discipline, organizing the available evidence, screening out implausible theories, and providing a focused framework to guide the analysis. 70 In Europe too, scholars have noted a broader rationale for the antitrust market. Podszun writes that 'market definition is all about . . . identifying the economic circumstances relevant for the behaviour in question'. 71 Robertson advances 'market characterization' as a function of market definition, which may be complementary to the market power rationale. This function, she writes, 'provides the necessary market context in order to understand and apply the competition theory of harm and an analysis of anti-competitive effects.' 72 Scholars can find support for these views in authority guidance, enforcement practice, and jurisprudence. In its Notice on Market Definition, the European Commission describes market definition as a tool which 'serves to establish the framework within which competition policy is applied.' 73 Even when it refers to market power, it only does so as a secondary objective: '[i]t is from this perspective that the market definition makes it possible inter alia to calculate market shares that would convey meaningful information regarding market power'. 74 In 1967, in Brasserie de Haecht, 75 the European Court of Justice evoked the idea that effects of anticompetitive practices could only be assessed by reference to a defined market. More recently, both Advocate-General Bot and Advocate-General Kokott expressed, in Erste Bank and Ziegler, respectively, that the definition of the market had to be performed by reference to the problem to be resolved, the nature of the issue examined, and the likelihood that harmful effects would occur. 76 When the US FTC argued in Whole Foods Market that it did not need to define a market, since antitrust markets are defined only to assess market power, the Court of Appeals referred to Brown Shoe to clarify that market power assessment is not the only purpose of market definition, since the antitrust market provides 'the appropriate setting for judging the probable anticompetitive effect' in a case. 77 Lastly, in its recent define the market to assess the effects of allegedly anticompetitive conduct, 78 as it had done five decades earlier in Walker Process Equipment. 79 What all these statements have in common is that they consider market definition to have utility beyond the mere assessment of market power. They see market definition as dependent on the purpose pursued in the enquiry, rather than being an independent and self-reliant concept. An antitrust market, in each particular case, is to be defined by reference to the question the case seeks to answer. This means that the legal or factual question needs to be articulated before the market can be defined. 80 C. Articulating the purposive approach 1. Potential variations Since a purposive approach merely means that the antitrust market is defined by reference to the purpose of the enquiry, there can be a variety of purposive approaches. This article provides a brief overview of potential variations. These variations can include, but are not limited to, market definition for the purpose of assessing the feasibility of the alleged conduct, evaluating the theory of harm, or establishing market power. The first variations rely on the notion of an antitrust market as the 'forum for economic decisions'. 81 The market is the analytical vessel containing the facts, actors, and forces which have shaped the (anti)competitive situation which has given rise to the case at hand. Selecting which facts, actors, and forces this vessel contains will depend on the perspective adopted. This perspective could be the alleged conduct, the theory of harm, or even the interests one intends to protect. 82 The further, more traditional, market power-rationale can itself be considered a purposive approach. The market power assessment is not truly distinct from the purposive approach but is merely one variation on it: one specific question which market definition can assist in answering.
The first variation focuses on the alleged conduct. If antitrust markets are defined with a view of assessing the feasibility of the alleged conduct, the conduct one believes to be problematic needs to be articulated before the market can be defined. 83 The market would be defined, under this approach, to reveal the (existence or lack of) constraints which would impede the adoption of such conduct by the undertaking (s)  A second approach is the definition of the market to assess the plausibility of the alleged harm. For example, in a case concerning exclusionary unilateral conduct, where the defendant adopted a strategy which interfered with rivals' access to an essential input, the allegation may be that consumers suffered higher prices as a result of the rivals' inability to constrain the defendant with its own offer. In that case, the market could be defined with reference to the preferences of consumers in the absence of the specific conduct, to enable an analysis of the price fluctuations under that counterfactual. All things being equal, would, in the absence of the strategy to impede access, and some consumers have considered the rival product a substitute to the defendant's offer.
Defining markets around the feasibility of the alleged conduct will likely seem more acceptable than a purposive approach centred on the theory of harm. If antitrust markets are defined with a view of assessing the alleged theory of harm, the starting point of the market delineation may vary. An example of this is when an authority takes the group it wishes to protect as the starting point of the exercise. This could be called the 'protected interest' approach. This might occur when the objective of competition enforcement and policy is construed as being concerned exclusively and directly with the harm to the interests of a particular group. In that case, the starting point of the exercise may be the group whose interests are allegedly harmed. 85 Posner stated, for example, that '[f]irst, a group of purchasers entitled to the protection of law must be identified (for example, customers of corned beef in New York City)'. 86 This 'protected interest' approach would, we would argue, only make sense in the context of a competition policy which clearly and unequivocally concentrates on the harm caused to one particular group. In reality, it is more likely that a system will seek to protect the interests of multiple groups. 87 In practice, it is not that easy to separate the 'protected interests' from the 'undertaking and its product' as a starting point, as one will influence the other. It is by reference to the customers (often one of the protected groups) of a product, after all, that a product will be defined. Indeed, Posner's protected group-NYC corned beef customers-are defined by reference to a productcorned beef. It is necessary, therefore, to identify the interested group-the customers-accurately, to identify the product accurately, even if one cares about the interests of other groups as well.
Defining the market with a view to testing the alleged conduct or theory of harm may seem controversial, but it can be argued that this merely recognizes that the concern of competition law is with the deviation from the standard of perfect competition, principally through the use of market power. Collusive conduct, exclusionary conduct, and also exploitative conduct in the EU are prohibited because of the impairment to the functioning, or outcomes, of the competitive market. The logic of market definition is already infused with the concepts of collusion, exclusion, and retrospective and prospective harm. 88 Indeed, the tests currently deployed for market definition are intrinsically linked with the theories of harm we adopt in competition law scholarship and practice. The use of the hypothetical monopolist test, or alternatively the hypothetical cartel test, 89 is rooted in the concern with increased prices as a result of the exercise of existing market power, 90 the same concern which forms the central point in traditional portrayals of antitrust injury flowing from collusion or exploitation. It is not a given that these are the tests which ought to be deployed in each case, 91 but their enthusiastic acceptance when first introduced is due in most part to the link of these tests with 'the real economic question' at the heart of merger cases. 92 Finally, the market power rationale itself can be considered a purposive approach: after all, the factors included in the market are only those relevant for answering a particular question-can the undertaking act to some extent independently, unrestrained by competition? Under a traditional price-centric approach, the market is defined to gauge the ability of the undertaking to increase prices significantly, in the absence of the disciplining force of substitution. Such an assessment is not truly distinct from the purposive approach but is merely one variation on it: one specific question which market definition can assist in answering.
A significant part of modern scholarship focuses on the market powerrationale of market definition, disregarding or overlooking the broader utility of market definition. Where scholars do evoke rationales for market definition akin to the purposive approach described in this article, they often appear to 88  imply that such a rationale is an alternative or complement to market powerjustifications. This is the case for most of the scholars cited in the section above. Robertson cites 'market characterization' as 'another function of the relevant market'. 93 Turner speaks of a 'dual role' for market definition, with the provision of a 'rational economic basis' for the decision as the primary role, and market power assessment as a secondary role. Turner does note that these two roles are, in fact, closely related and that, furthermore, market definition may not be well-suited for the establishment of market power. 94 Similarly, Podszun describes the European Commission's approach to market definition in the Notice as follows: 'the Commission quickly turns market definition from "establishing the framework" to "calculate market shares," not noting that the former could be enabled by the latter'. 95 It is correct, of course, that the Commission and Courts have not made an explicit declaration that market definition is a purposive tool, with market power assessments merely one example of such a purpose. Yet, even in the Notice, the Commission leaves room for such an interpretation, by providing that one of the possibilities of market definition is 'inter alia' to calculate market shares. 96

Competitive constraints
A purposive approach means that the relevant antitrust market will be defined to answer a particular question-for example, the existence of market power, the feasibility of the alleged conduct or theory of harm-and thus be defined in function of that question. A market power-assessment may, at first glance, appear far removed from the view of market definition as a framework to assess the alleged conduct or theory of harm. However, this is not necessarily the case. One the contrary, all these purposes share a common baseline-which is essential to ensuring coherence in market definition. The definition of the market in light of the alleged conduct or theory of harm should be founded on a 'competitive constraints'-baseline. Adopting such a baseline pulls together the idea of market power, as the broader concern of competition law, and more granular purposive approaches centred on the conduct or theory of harm. This is the baseline which most closely aligns with European (and American) jurisprudence. Indeed, authorities and courts in both the EU and the USA have consistently acknowledged that market definition concerns the identification of primary competitive constraints. 97 After all, we know broadly what competition policy cares about: namely, the existence and impairment of competition. Thus, markets ought to be defined 93  around the existing competitive constraints, but the competitive forces that are to be considered relevant may vary depending on the conduct alleged or the theory of harm.
To illustrate this point, think of the difference between merger review and an assessment of unilateral conduct by companies with market power. For mergers, authorities will care about impediments to competition in the future, particularly if the merger were to lead to the creation of a dominant position, 98 and thus define markets by virtue of the constraints on such creation now, and the possible removal or addition of such constraints after the merger. In unilateral conduct cases, the market will be defined by reference to the (lack of) forces which would have obstructed the alleged harm. More specifically, the concern with mergers may be that prices will rise significantly as a result of this new dominant position. The question is whether there are and will remain sufficient competitive constraints, after the merger has been consumed, to keep the undertaking from being able to raise prices significantly. Another concern could be whether innovation will significantly be reduced. In that case, the authorities will have to define the market in line with the competitive constraints which impact the undertaking's decisions with regard to innovation. Price constraints may be-but are not necessarily-the same as innovation constraints. Likewise, a specific concern under Article 102 TFEU may be that the undertaking has been able to foreclose competitors in an adjacent market, not by offering a superior product, but by tying that product to its 'flagship' product. In that case, two markets ought to be definedand the question will not only be whether there are any constraints on the undertaking's ability to sell its flagship product at any price it wants, but more specifically whether it is constrained from selling it with another product. The market for the flagship product would include all those constraining factors. Similarly, to define the market for the tied product, the question would not just be which forces constrained the sale of that product with regard to price and output, but also with regard to making such a sale conditional on the purchase of another product. 99 The competitive constraints' approach to market definition is not only in line with the purpose of inquiries but also bears the best resemblance to the concept of markets as used in IO scholarship. Fisher stated as early as 1987 that 'the "market" must include those firms and services that act to constrain the activities of the firm or firms that are the object of attention.' 100 The purpose of the exercise influences what the starting point and focus of the exercise ought to be. As explained above, IO markets were drawn by reference to a single seller, and its product(s), to identify which forces would constrain the 98 EUMR, Art. seller from adopting certain conduct (for example, raising prices). A similar approach is taken for antitrust markets, where markets are drawn by reference to a particular entity and its product(s), which are relevant to the behaviour of concern. Under Article 102 TFEU, the investigation may be concerned with allegedly anticompetitive conduct which has affected price, and thus the market will be defined by reference to a particular undertaking and the product it sells, identifying the constraints on the ability to substantially adjust the price of that product. Under Article 101, the focus would be on the ability of multiple entities together to adopt certain conduct regarding a particular product. In a merger review, if the investigation centres on the forces which would restrain a merged entity from, say, raising the prices of joint supply, the starting point is likely to be two undertakings, in consortium, selling a particular product (range). The question could be whether the merger is likely to lead to higher prices in the future, so that the market will be defined by reference to future constraints on a merger product offer. The concern may be different, however, if the prices already appear high. In that case, the concern could be whether the currently separate undertakings face competitive constraints premerger, and how the merger would impact those constraints: would the merger render permanent a (high price) situation which may otherwise be transient? 101 The competitive constraints' approach is not incompatible with a market power approach. After all, market power is said to exist when an undertaking can profitably raise price-something it will not be able to do if it is competitively constrained at that time. The value of the competitive constraints approach lies in its ability to paint a fuller picture of the setting in which the conduct occurs, through the lens of the effects the authority cares about. It is purposive, rather than singularly focused on an imagined 'objective' measure (that is, market power). It sheds light, not only on the market power (pricing ability) of the undertaking at that point in time but also on whether it is constrained from adopting certain (harmful) practices by present and potential forces. When the focus of market definition is not just on market power, market definition becomes a more useful tool to identify the key factors relevant to an understanding of the conduct at hand, in a manner which can be used by authorities and courts. As stated by the Commission, 'the determination of the relevant market is useful in assessing whether the undertaking concerned is in a position to prevent effective competition from being maintained This enables the establishment of market power, where required, but strives to go further by providing a lens through which to assess the facts'.

IV. OBJECTIVITY IN A PURPOSIVE PROCESS
A purposive approach to antitrust market definition implies the adoption of a particular perspective when delineating the market, distilling complex, and dynamic relationships into a static picture useful to the enquiry. As will be seen in this part of the article, such a market may be particular (defined by reference to a previously established perspective), but it does not need to be subjective (that is, defined in the function of the personal opinions of the decisionmakers). To reduce the risk of subjectivity and arbitrariness, authorities and regulators ought to strive for process objectivity in the definition of antitrust markets.
Achieving objectivity in the definition of antitrust markets matters greatly, since a lack of trust in the conclusions drawn in particular cases, which start with and often rest on the relevant market, would lead to a lack of trust in the law as a whole. The replicability and coherence of the process, and assurance of equality in enforcement, are necessary preconditions for the rule of law. 102 Calls for objective market definition abound, by scholars as well as the European Commission. 103 What is often missing in discussions on how to achieve such objectivity is a recognition of the true nature and utility of the market as a purposive and particular frame of reference.

A. The particular market
The world is always changing, and our description of it in terms of quantities (length, height, weight, . . . ) is merely the description of relative change. We describe the world by describing how one variable changes in relation to another. Any measurement distils the relationship between variables into a particular point. 104 The dynamic is rendered (artificially) static, to provide human beings with a frame of reference. That frame of reference is not only static, when the world is actually in constant movement, it is also particular. Human beings do not perceive everything about the world. Our vision of reality is blurred: we remain blind to many variables and register only those parts with which we interact and which our brain considers useful for our understanding in that particular moment and place. 105 This is a general truth, which becomes even more pertinent in the context of an investigative analysis with a particular end-goal. If a researcher means to find out the impact of a particular change on a particular system, they keep 'all other things equal', merely investigating the relationship between the change in one variable on the variable of reference. The researcher does not construct a 'study of everything'. The answer to the question depends on the question itself. We assess things by adopting a particular perspective. 106 In that sense, no answer is ever wholly complete and objective: we select what is included in the assessment based on prior choice. Thus, when making sense of the world for a particular purpose, two things are evident: we make the dynamic static and we cannot reach perfect objectivity.
These two conclusions hold true especially for the definition of antitrust markets. An antitrust market is by its nature a blurred picture. It consists of a selection of variables, which have been chosen because they are considered most relevant to answer the specific question at hand. We cannot 'take a view from nowhere', 107 as the 'market' exists not only by virtue of human interaction but more pertinently by virtue of the investigator's perspective. Markets are defined by reference to a starting point: a focal undertaking or a focal want (focal product). A different starting point will likely lead to a different relevant market. Moreover, because of the purposes of product market definition set out above, choosing which economic entities (and even which parts of such entities) to include, which wants to consider, and which customers to take into account will depend on the conduct or theory of harm considered. Antitrust markets cannot be delineated in total isolation from the alleged conduct and theory of harm.
The static nature of antitrust markets may appear confounding to many: the concept is, after all, used to assess decisions and conduct of (natural or legal) persons, phenomena which are by their nature dynamic. This is even more challenging in the context of industries where wants, products, business models, and market participants seem in continuous flux-an issue to be explored in future research. Yet, this staticism is inherent in, and even necessary for, a meaningful analysis with 'objective' parameters. Antitrust product markets do not delimit objects or entities but rather put relationships under a magnifying glass. Antitrust markets bring order where none exists in reality. Product markets are only seen as 'static', because we fail to comprehend that everything is dynamic, that any description of the world by its nature is an attempt to preserve that what cannot be preserved, to make static that what 106 Rovelli describes this quite eloquently: 'Science aspires to objectivity, to a shared point of view about which it is possible to be in agreement. This is admirable, but we need to be wary about what we lose by ignoring the point of view from which we do the observing. In its anxious pursuit of objectivity, science must not forget that our experience of the world comes from within. Every glance that we cast towards the world is made from a particular perspective.' (Rovelli (supra note 5) 132) 107  to an 'imagined' phenomenon. 114 Any such reference to 'subjectivity' really is a reference to the 'particularity' of an antitrust market, previously explained. If, on the other hand, 'subjective' is used in the prevalent sense, to refer to a result which is dependent upon and variable with the feelings or opinions of the persons engaged in the analysis, 115 then an antitrust market is not subjective. It does not depend on personal opinions, but on the conduct or theory at play. Antitrust markets are particular, not subjective. No process, which results in the drawing of boundaries where none naturally exist, is ever devoid of particularity. Evidently, such particularity increases the flexibility of antitrust market definition, potentially increasing the discretion authorities may feel they possess. This discretion, and the risk of arbitrary decisions it entails, would be antithetical both to a rational application of the law and to the ideal of legal certainty and foreseeability, which are essential to the rule of law. It is imperative, therefore, to ensure that the perspective adopted in the definition of the market depends on the previously established purpose of the enquiry, rather than on the personal aspirations of the decisionmaker. It is important to pursue objectivity in antitrust market definition, not through similar outcomes for similar undertakings, but through a structured and transparent process.
Objectivity can be achieved in two distinct manners: either through product objectivity or through the reliance on process objectivity. Product objectivity is attained when the outcomes of an analysis or theory correspond to accurate representations of the natural world-a reflection of phenomena which, even absent the human perceiver, would fit the description given. 116 Colour, though possibly perceived differently by different individuals, corresponds to a particular wavelength which does not alter under the individual gaze. The wavelength is therefore an objective product. 117 Process objectivity, on the other hand, may not lead to products which unalterably correspond to external reality. Indeed, the outcomes of an analysis may be inherently variable with the perspective adopted. Rather than through the realism of the outcomes, objectivity is achieved during the analysis itself, by removing the individual biases of the investigator from the analysis. Process objectivity does not mean 114 The latter corresponding to definition 4b given by the Oxford English Dictionary: 'existing in the mind only, without anything real to correspond to' ("subjective, adj. and n." OED Online, Oxford University Press, June 2020, www.oed.com/view/Entry/192702. Accessed 7 July 2020 lifting the analysis out of a particular perspective, but freeing it from an individual's interests. 118 Though perspective is intrinsic in the process, this 'perspective' is not understood as the individual perspective of the person performing the analysis or formulating the theory, but as the aggregate human perspective which shapes the phenomenon being studied, or the values or presuppositions commonly known and accepted in a field. Removing the individual's feelings and opinions from the analysis is how objectivity can be achieved in an analysis which is perspective-particular. 119 Process objectivity retains a role for values-those of a society or a school of thought-but removes the individual's desire for a particular outcome. Thus, process objectivity would be achieved when outcomes change with a change in overall perspective, not a change of individual investigator. 120 In the context of antitrust market definition, product objectivity cannot be attained. First, because, since there is no 'market' absent human behaviour and human perspective, there is no 'market' in the natural world. Second, and more pertinently for our aims, because, as 'antitrust markets' are defined in order to answer a specific question on the feasibility of alleged conduct and/or theory of harm, they are dependent on a competition law-'purposive' perspective. Thus, removing 'human' and 'purposive' perspectives means erasing the antitrust market altogether. This does not mean, however, that we cannot strive for process objectivity. Though a purposive perspective is fundamental to the delineation of the antitrust market in a case, the individual perspective of the investigator is not. The antitrust market for the same undertakings and even same products may differ from case to case, as the alleged conduct and/or theory of harm varies, 121 but should not depend on the individuals working on the case.
A further compelling argument in favour of process objectivity, as an assurance against subjectivity, is the nature of an antitrust market as a legal 118  analytical tool. Antitrust markets are more than mere factual descriptions relevant to the analysis: they are normative benchmarks. As a legal concept, 122 the antitrust market translates multifaceted economic ideas into predominantly binary choices, to enable legal decision-making. 123 Such choices are normative choices. 124 Under the rule of law, legal decision-making is expected to adhere to principles of certainty and impartiality, 125 notions which align closely with the understanding of 'process' objectivity. Objectivity in legal decisionmaking cannot be achieved by demanding that case outcomes correspond to a 'natural' truth-such truth being at the very least unattainable, if not inexistent-free from societal values. Thus, legal scholarship and doctrine has developed principles of objective decision-making, based on transparency and predictability of the applicable rules and normative criteria, and assurances concerning the neutrality of decision-makers. 126 A decision is objective, where it is free from personal interests, it is based on general and publicly available principles, the values that underlie it are commonly known, and it is supported by well-justified and publicized reasoning. 127 In sum, the rule of law considers the process of decision-making, not its outcomes per se: if the process of decision-making corresponds to the principles of certainty and impartiality, it is deemed to be objective. 128

Achieving process objectivity
Process objectivity implies the absence of individual predisposition, rather than the absence of a specific overall perspective. Thus, a purposive approach of any kind is not inherently at odds with process objectivity. Nonetheless, there is a risk, common to any decision-making process involving a degree of discretion, that a multitude of variations will be adopted without any coherence, so as to suit the desires of individual investigators in specific cases, turning a 'purposive approach' into a 'subjective approach'. It is possible to minimize the risk that a purposive approach be co-opted to serve individual preferences, by putting a few measures in place. First, adopting a common baseline, such as the competitive constraints approach described above, is a first step to ensure 'process objectivity'. Any purposive approach adopted ought to tie into the broader concern of competition law with impairments of competition. This would not only enhance the legitimacy of the process but also provide a touchstone for the predictability of the process. This means, in practice, that we should strive towards a full identification of the different forces which can act as sources of competitive constraint, based on economic theory and experience. Any market definition, performed on a case-by-case basis, would refer to these competitive constraints and select those most appropriate to the alleged conduct and theory of harm. Such a list could perhaps, at European level, be incorporated in a more comprehensive Market Definition Notice, to be expanded beyond the currently rather limited description of demand substitution, supply substitution, and potential competition. Whether in the Notice or in another form, it is crucial that as our understanding of competitive constraints progresses, in light for example of new business models and commercial strategies, any such list should be updated.
Second, process objectivity requires that the approaches (which could be) adopted be made explicit, in order to ensure a degree of foreseeability for subjects of the law. This means that authorities or courts tasked with antitrust market definition need to be explicit about the manner in which the process is conducted. Transparency in the delineation of antitrust markets is extremely important. The antitrust market can be determinative for the final decision in case. 129 A balance should therefore be struck between the utility of a 'particular' analytical tool and legal certainty for the persons involved. Transparency is achieved in two stages: both by making the approach and process explicit ex ante, for example, through guidance papers, as well as in each specific case. What ought to be made explicit is: first, the particular purposive perspective authorities can and do adopt and, second, the common steps of antitrust market definition and how the purposive approach affects them.
Authorities should frankly admit that antitrust market definition depends on the purpose of the enquiry. They need to set out, in clear terms, that the conduct alleged and/or theory of harm is explicitly considered when defining an antitrust market. This ought to be done before cases are brought, through guidance papers, and should furthermore be a standard part of the market definition in each decision. When defining a relevant market in a particular case, the authority should highlight the purpose for which the market is defined, by reference to the competitive constraints of relevance to the investigation. This means that a clear distinction ought to be made between the facts available to the authority 130 and the purposive perspective adopted in the assessment of those facts. The authority should determine the purpose of market definition (identifying the competitive constraints relevant the alleged conduct and theory of harm) and then give due consideration to all the evidence before it which may be relevant (and reliable) 131 to that perspective. The authority cannot omit facts which would indicate the existence of competitive constraints on the basis that they would undermine the theory of harm the authority wishes to put forward. It needs to provide reasoning as to why certain facts are (not) relevant to the identification of the competitive constraints on the alleged conduct and harm. 132 In addition, the specific steps authorities undertake and variables they consider in antitrust market definition should be clearly publicized. Despite the variation in antitrust markets depending on the context of the enquiry, there are indeed specific steps which are uniform across market definition: the identification of the focal product as a starting point, the identification of a candidate market, the identification of competitive constraints, and the ultimate conclusion on a relevant market. Where these steps may vary depending on the scope of the purposive approach, this ought to be made explicit as well. These should be publicly set out, not just in guidance available before any cases are brought, but also in every specific decision: the process of the definition of the market in every case should be made public, not merely the resulting market. 133 Lastly, there is a need for thorough judicial review of relevant antitrust markets adopted by authorities in specific cases. Although administrative authorities, within their own respective legal systems, may enjoy a certain degree of discretion in complex economic and technical matters, 134 accurate portrayal of facts as well as the legality of decisions are, or ought to be, subject to judicial review. 135 First, market definitions ought to be reviewed for the selection and appraisal of facts by the authority within the purposive perspective adopted. In the EU, the extent of judicial review is currently ambiguous in situations of complex economic evaluations, including market definitions. 136 It could be argued therefore that the definition of a market, based on the selection of facts within a purposive perspective, is such a complex economic assessment subject to limited review. Nonetheless, EU jurisprudence has emphasized the need for the GC to ensure that the decision contains all information 'that must be taken into consideration in appraising a complex situation'. 137 Where the Commission has not considered all relevant and reliable evidence, there would be a 'manifest error of assessment'. 138 Thus, even if an argument could be made in favour of limited judicial review of the purposive perspective adopted, the GC ought to be able to hold the Commission to take account where it has failed to consider factors which indicate the existence of competitive constraints on the alleged conduct within that perspective. Second, a legality review ought to be undertaken. Sousa Ferro has argued that market definition, particularly the method applied and key principles (such as whether 'gratuitous' markets can be defined) adopted, entails the application of 'normative criteria' and should therefore be subject to review. 139 That market definition, which entails normative principles, is a proposition with which this article would agree, for a much more compelling reason: if market definition is purposive, it is intrinsically linked with decisions about the aim of individual investigations and thus, ultimately, the goals pursued by the law. Though policy decisions may to some extent be left to the discretion of an administrative authority, matters of law ultimately belong to the Courts. 140 These different suggestions all need to be considered, as they are undeniably linked to one another. A thorough judicial review, for example, can only be