The Tribunal's reasons
47 The Tribunal set out the competing arguments in relation to the construction of criterion (b) at [815]-[817] of its reasons:
[815] A significant issue is the meaning of "uneconomical for anyone" to develop another (similar) facility. The competing views are that "uneconomical" means that: (1) it would not be profitable for anyone to develop the facility (the "privately profitable" test); (2) the total net costs (including social costs) exceed the total net benefits (including social benefits) of developing another facility (the "net social benefit" test); or (3) a single facility can meet market demand at less total cost than two or more facilities (a "natural monopoly test").
[816] To understand the debate, it is necessary to appreciate the different results that might be reached depending on whether or not a privately profitable test is adopted. It was accepted by all the economic experts that the existence of a natural monopoly does not necessarily preclude the profitable development of a second facility. For example, suppose that an incumbent and potential access seeker occupy infra-marginal positions in a related market; ie, their marginal cost is below the prevailing market price. In that circumstance, it may be profitable for a second facility to be built, notwithstanding that it would be more efficient to share an existing facility.
[817] The question is whether Part IIIA is intended to apply in circumstances where it is profitable - albeit less profitable, and potentially less efficient from society's perspective - for a second line to be built. The incumbents say Part IIIA is not intended to apply in those circumstances, because it is concerned with removing "bottlenecks" and criterion (b) should be seen as a bottleneck test. In contrast, FMG and the NCC argue that while bottleneck considerations may be relevant to criterion (a), criterion (b) is concerned with efficiency.
48 The Tribunal then explained at [818]-[835] its reasons for rejecting what it referred to as the "privately profitable" test (which Rio Tinto referred to in this Court as the test of "economic feasibility"). The passage is lengthy but it is worth setting out in full because it provides a comprehensive statement of the competing considerations drawn from the text of the Act, its context and the evidence of the economists. The Tribunal said:
[818] In resolving this debate, the place to begin is with the objects of Part IIIA. There are a number of features of the objects clause which should be noted. First, it refers to "effective competition" rather than competition per se. This is to be contrasted with s 2 of the Trade Practices Act, which relevantly provides that the object of the Act generally is to enhance the welfare of Australians through the promotion of competition. A very useful shorthand description of effective competition is proffered by Professor Hausman. He said: "By 'effectively competitive' economists mean that no individual firm (or group of firms) is exercising significant market power nor is the price above the competitive price." A privately profitable test does not sit easily with the object of achieving effective competition. If viewed as a "bottleneck" test, as the incumbents would have it, then it simply tests whether a person could compete in a related market without access. It does not ask whether that person could compete effectively. It is not hard to conceive of circumstances in which a market is less than effectively competitive because third parties, relying on marginally profitable alternative facilities, cannot truly compete with an incumbent using (a much more profitable) facility with natural monopoly characteristics.
[819] Another feature of s 44AA is that it is concerned with two distinct but related concepts - efficiency and effective competition. The references to efficiencies in "operating", "using" and "investment" in infrastructure connote concepts of productive, allocative and dynamic efficiencies. On any view, the scope of criterion (b) must take into account both effective competition and the efficiencies contemplated by s 44AA(1).
[820] One issue raised by the privately profitable test is whether it ignores efficiency considerations, in particular, the allocative efficiency associated with the use of a natural monopoly facility. The proponents of the privately profitable test contend that if it is privately profitable to develop an alternative facility, there is a strong incentive for the access seeker and facility owner to voluntarily implement a socially efficient sharing arrangement. Professor Willig, for example, explains why it is that the credible threat of a new facility will result in "strong individual incentives" for the incumbent facility owner to enter into an efficient facility-sharing agreement. In substance, his approach suggests that once an incumbent realises that refusing to share a facility will not prevent an access seeker from competing in a related market, the incumbent would rationally share the facility if it is efficient to do so. If the incumbent cannot prevent a competitor entering a related market, the incumbent may as well profit from that entry, if possible. Indeed, Professor Willig argues, if the incumbent still refuses to share the facility in the face of a credible threat, then this strongly suggests that there must be inefficiencies and costs associated with sharing that outweigh any efficiencies. His point is that where it is privately profitable, an entrant will come into the market because it will be competitive (ie there is no market failure) and there is no need for regulation. If an alternative facility is not privately profitable, then regulation might be warranted.
[821] A variation of this argument is that even if the market does not always achieve an efficient outcome when there is a credible threat of a new facility, it still gets it right most of the time and does a better job than regulation. For example, Professor Ordover draws a distinction between a technical natural monopoly test and evidence of market behaviour indicating an "independent [ie third party] commitment to enter/construct/duplicate" the facility. If there is such a commitment, Professor Ordover argues that the technical assessment whether it is or is not economic (or "uneconomical") to build another facility is "trumped" by the revealed behaviour of market participants.
[822] Similarly, Professor Kalt, who deals with the US essential facilities doctrine, notes that under the second criterion expounded in the MCI case (the "impractical to duplicate" criterion), a would-be entrant could not succeed if it were privately profitable to duplicate the facility. In other words, says Professor Kalt, criterion 2 of MCI uses the market to perform the social cost test on the efficiency of new facilities. Professor Kalt acknowledges that market forces do not always produce satisfactory outcomes. But he says that "it is a false standard to place such market forces up against some vision of a court/regulatory process that is costless and without error of its own". He points out that the operative assumption of the US approach is that firms are rational profit-making concerns and this underlies the stringent US policy toward mandating access. His view is that, from a policy perspective, the US approach has the distinct advantage of using competing parties' respective self-interests to compel them to assess accurately the costs and benefits of their alternatives and for each to act on those alternatives in accord with their own individual self-interest.
[823] In essence, the economists supporting the privately profitable test assert that where an alternative facility can credibly be built, private negotiations will necessarily result in efficient outcomes or, at least, will achieve efficient outcomes more readily than regulation. There are several problems with this assertion. First, it assumes that firms always, or usually, behave in an economically rational manner but from empirical observation we know they do not - especially when it comes to dealing with potential competitors. Second, there are often reasons for an incumbent owner who is behaving rationally to deny access to a potential competitor even when sharing would be socially optimal. Forcing the competitor to use a less profitable alternative facility may harm that competitor. The incumbent may seek to exploit the fact that it will take some time to build the alternative facility. It may be that it is only profitable to build an alternative facility with limited capacity (which is lower than the spare capacity on the existing facility which would otherwise be available). The incumbent may be mindful of not giving a fledgling competitor a "leg-up" to facilitate its growth into a larger player. Third, given the potential for market failure, it is far from clear that market forces achieve a better result than regulation as a general rule. In any event, it is doubtful that the Tribunal is entitled to assume that the decision-makers regulating Part IIIA will make errors.
[824] There is another reason why a privately profitable test may not lead to the efficient use and operation of a facility. Suppose that an existing facility is a natural monopoly, ie it can satisfy society's total demand at a lower cost than two or more facilities. Suppose also that it is privately profitable to build an alternative facility which can only satisfy some, but not all, potential demand. On a privately profitable test, a declaration could not be made, even though many potential users would not be able to use the second facility, but could use (and use more efficiently) the incumbent's facility. It is difficult to see how such an outcome is consistent with the efficient use and operation of infrastructure - and for that matter, the achievement of effective competition - contemplated by s 44AA.
[825] Another factor which tells against a privately profitable test is the emphasis on natural monopolies in the background materials to Part IIIA: see Chapter 9, where this material is discussed in detail. For present purposes, it is sufficient to note that the two consistent themes which emerge from this material are that: (1) facilities requiring access exhibit natural monopoly characteristics; and (2) the notion of being "uneconomical" has been linked to the definition of natural monopoly. In Chapter 11 of the Hilmer Report, which concerns access to "essential facilities", there is repeated reference to a facility being uneconomical to duplicate because it exhibits natural monopoly characteristics. Indeed, the Hilmer Committee's concern was with what it described as the "essential facilities problem". They defined an essential facility as one which satisfied two key conditions. First, the facility must exhibit natural monopoly characteristics and "hence cannot be duplicated economically". Second, access to the facility is required if a business is to be able to compete effectively in a related market. That said, the final recommendations of the Report did not refer to natural monopolies, instead recommending that a necessary condition for access was that access be essential to permit effective competition in a related market. This reproduced only the second element of the Committee's "essential facility" definition.
[826] For reasons which are unclear, but probably relate to drafting style, the legislation and Competition Principles Agreement that followed the Hilmer Report adopted a more elaborate series of criteria for declaring access than those which were originally recommended. Criterion (a) seems to correspond, albeit couched in different terms, to the "second limb" of the essential facilities definition adopted in the Hilmer Report. Criterion (b) seems to correspond to the "first limb" of that definition. It is significant that in "reintroducing" this first limb, COAG has continued to associate the first limb with natural monopolies. In September 1994, a draft legislative package was released by COAG for public comment. One of the documents in the package was the explanatory memorandum that described an essential facility as one which "exhibits a high degree of natural monopoly characteristics; ie, a competitor could not duplicate it economically". In the second reading speeches for the Competition Policy Reform Bill in the House of Representatives (Commonwealth, Debates, House of Representatives (1995) Vol HR202, p 2799) and in the Senate (Commonwealth, Debates, Senate (1995) Vol S170, p 2438), it was said that the notion underlying the proposed Part IIIA regime is that access to certain facilities with natural monopoly characteristics is needed to encourage competition in related markets.
[827] While the background material makes repeated reference to infrastructure that displays natural monopoly characteristics, there is no reference to a privately profitable test in terms or in similar language. Nonetheless, the incumbents argue that the Hilmer Committee was principally concerned with bottlenecks, rather than natural monopolies. In this regard, they place particular emphasis on the passage from the Hilmer Report which recommended that a necessary condition for a declaration be that access is essential to permit effective competition in a related market. The Report then observed (at p 251): "Clearly, access to the facility should be essential, rather than merely convenient."
[828] What must be borne in mind is that this comment was made in the context of access being essential for effective competition. It was not about whether access is essential to be able to compete per se. A "bottleneck" approach, as advocated by the incumbents, only tests the latter.
[829] Another point against the privately profitable test is that it would lead to a significant degree of overlap between criterion (a) and criterion (b). All parties accept that whether or not it is privately profitable to build an alternative facility would be a relevant consideration for criterion (a). This is not to suggest that there is complete overlap between criterion (a) and a privately profitable test under criterion (b). There may be circumstances where, for example, it is not privately profitable to build an alternative facility but, for separate reasons (eg the existence of an alternative facility which provides part of the service), criterion (a) is not satisfied. Conversely, one can conceive of circumstances in which it would be privately profitable to build an alternative facility but criterion (a) would nonetheless be satisfied, such as when the use of a less profitable alternative facility would not allow for effective competition. The point is that the existence of a bottleneck (or whether it is privately profitable to build an alternative facility) is not in itself determinative of whether access would promote effective competition and, for that matter, whether access is socially efficient. It is, therefore, unclear why criterion (b) should separately test for a bottleneck on a stand-alone basis. An alternative approach is that criterion (b) is concerned with efficiency (ie the efficient use of existing infrastructure) and criterion (a) considers effective competition. This approach has the merit of avoiding overlap between criteria (a) and (b) and, more importantly, directly addresses the objects in s 44AA(1).
[830] Were the interpretation of criterion (b) governed solely by the forgoing considerations, the Tribunal would have no doubt in concluding that the privately profitable test should be rejected. There are, however, other considerations which muddy the waters. The first is criterion (e), which is whether access to the service is already the subject of an "effective access regime". When considering this criterion, s 44H(5) provides that the minister must have regard to the Competition Principles Agreement. Clause 6(1)(a) of that agreement refers to whether it is "economically feasible to duplicate the facility." Putting these points together, the incumbents argue that the reference to "economically feasible" indicates a privately profitable test. They then argue that criterion (b) must be interpreted consistently with clause 6(1)(a), or otherwise different results may emerge, depending on whether one is assessing an application under a State access regime or under Part IIIA.
[831] The Tribunal acknowledges that this argument has force. Perhaps the most natural meaning of the phrase "economically feasible" connotes private profitability. However, it is not too strained to read "economically feasible" as economically efficient, in the sense that something that is inefficient may be economically unfeasible when looked at from society's perspective.
[832] Another feature supporting the privately profitable test is the reference in criterion (b) to it being uneconomical "for anyone" to develop another facility. The incumbents argue that this reference does not sit well with a natural monopoly or net benefit test. Both of those tests, it was said, ask what is best for society, rendering the reference to "for anyone" otiose.
[833] There is also force in this argument. But, equally, the application of the phrase to the privately profitable test creates difficulties. It was suggested by the incumbents that "anyone" means "any particular individual who can be identified", including, for example, mining companies whose iron ore is sufficiently valuable to subsidise the cost of building the alternative rail facility. An alternative view is that "anyone" means anyone at all, asking whether any hypothetical person could build a rail line and make a profit from providing below rail services.
[834] We think that under either view, practical problems arise. On the incumbent's view, how is one to identify the relevant individual who might profitably build the line? For example, in this case, is one required to consider whether a Chinese steel mill might fund a line in order to facilitate obtaining long-term supply contracts? If the alternative view is adopted, how is one to assess the potential demand for the services and the price for below rail services that "anyone" would be willing to pay?
[835] In the end, the Tribunal does not consider that criterion (b) should be interpreted as a privately profitable test. That test is inconsistent with the enacting history and does not adequately meet the objectives of Part IIIA.
49 The Tribunal, having rejected private profitability as the test erected by the reference in s 44H(4)(b) to "uneconomical for anyone", went on to consider whether s 44H(4)(b) erects a test of natural monopoly or net social benefit. At [836] - [839], the Tribunal said:
[836] Having rejected the privately profitable approach, it is necessary to consider whether a natural monopoly approach or net social benefit approach is to be adopted. The Tribunal has in the past favoured the net social benefit approach. The Tribunal first considered the meaning of criterion (b) in Sydney Airport (No 1). There the Tribunal said that "uneconomical" did not inquire whether duplication was privately profitable. It said the criterion had to be construed in a broader social cost-benefit sense, in which the total costs and benefits of constructing another facility are to be taken into account. The Tribunal went on to explain (at [205]) that:
If "uneconomical" is interpreted in a private sense then the practical effect would often be to frustrate the underlying intent of the Act. This is because economies of scope may allow an incumbent, seeking to deny access to a potential entrant, to develop another facility while raising an insuperable barrier to entry to new players (a defining feature of a bottleneck). The use of the calculus of social cost benefit, however, ameliorates this problem by ensuring the total costs and benefits of developing another facility are brought to account.
[837] The construction of criterion (b) was again considered in Duke Eastern. It was pointed out that the criterion appeared to describe a facility which exhibits natural monopoly characteristics. That led the Tribunal to say (at [64]): "[I]f a single [facility] can meet market demand at less cost (after taking into account productive allocative and dynamic effects) than two or more [facilities], it would be 'uneconomic', in terms of criterion (b), to develop another [facility] to provide the same services." The Tribunal concluded (at [137]) that the "test is whether for a likely range of reasonably foreseeable demand for the services provided by means of the [facility], it would be more efficient, in terms of costs and benefits to the community as a whole, for one [facility] to provide those services rather than more than one."
[838] With respect, we consider that a natural monopoly approach is preferable to a net social benefit approach adopted in previous Tribunal decisions for several reasons. First, the background material to criterion (b) consistently links the term "uneconomic" to the notion of a facility exhibiting natural monopoly characteristics. Natural monopoly rests upon a production cost function which does not take into account social benefits or net social benefits. Second, natural monopoly characteristics are concerned with the costs of production based on the available technology. Third, a net social benefit test gives criterion (b) a role which overlaps substantially, and perhaps usurps, the role of criterion (f). Both would involve a weighing up of many of the same social costs and social benefits. Importantly, in weighing up those costs and benefits, the criteria might arrive at different results. It must be borne in mind that many social costs and benefits are necessarily difficult, and sometimes impossible, to quantify. Accordingly, it may be difficult to conclude, at least in quantifiable terms, that there is or is not a "net social benefit". A requirement to be positively satisfied of such a matter - which would be a requirement if criterion (b) were a net social benefit test - would create a threshold which may, in practical terms alone, be difficult to satisfy. This is to be contrasted with criterion (f), which is framed in the negative.
[839] Moreover, criterion (f) looks at the issue in a different setting. For criterion (f) to be satisfied (although it is expressed as a negative), it is not sufficient for the net costs of access to exceed net benefits, ie even if that is the result of the inquiry, the making of a declaration may yet not be contrary to the public interest. Other factors might carry the day. We will explain why this is so when dealing with that criterion. There is, however, no such latitude given to the regulator were all social costs and benefits to be brought into account in criterion (b).
50 The Tribunal described the test for a natural monopoly in a way which regards criterion (b) differently from previous decisions of the Tribunal. Whereas previous decisions of the Tribunal have regarded social costs and benefits in the course of considering criterion (b), the Tribunal in this case took the view that these costs and benefits should be taken into account in considering criterion (f). The Tribunal said (Reasons at [840] - [855]):
[840] The question is what costs are to be brought to account when deciding whether a facility has natural monopoly characteristics. Here several subissues arise. For one thing, we know that the costs cannot be those incurred by the "facility", for a facility incurs none. The relevant costs are those of the firm. In saying this, we recognise the concept of a firm having a number of plants, one of which may exhibit cost characteristics akin to that of a natural monopoly. This is a concept which some economists (eg Sharkey) call "plant subadditivity". Components of firms, like plants or facilities, cannot be natural monopolies as strictly defined because they are not a firm but only a component of a firm. Nonetheless, criterion (b) requires the notion of a natural monopoly to be applied to facilities.
[841] Whether a firm's facility has natural monopoly characteristics involves determining whether the firm's cost function in relation to that facility is subadditive at all levels of output. This is a purely technical inquiry which looks at a firm's production costs. That is, the cost function to which regard must be had is the cost of the inputs (eg labour, operating costs, capital costs, co-ordination costs) incurred in producing the relevant good or service.
[842] The expert economists put a different proposition. Most supported the position that "costs" are not confined to costs of inputs but should include all social costs. Dr Fitzgerald put it this way: "'Costs' should clearly, as a matter of economics, be defined comprehensively - i.e. not be confined to comparative capital and operating costs, but encompass all costs that differ as between the alternatives, including e.g. diseconomy costs of multi-user c.f. single-user operation" (emphasis in original). With respect to the social cost test, Professor Willig points out that if some version of the test is to be applied, it would be inconsistent with basic economic logic just to take into account construction costs and operating costs. He says it would be necessary in the comparison to include all the impractical costs "engendered by the inevitability and the need to handle as well as possible the conflicts, disputes and paralyses that accompany the mandated sharing of the single set of facilities". It would also be necessary, he says, to include the significant social costs of the disincentives to invest caused by sharing. These costs, Professor Willig rightly says, are difficult to estimate and predict.
[843] The Tribunal has given careful consideration to the views of the experts but is of the firm view that a natural monopoly test under criterion (b) should not take into account all social costs, for three reasons. First, almost all of the extensive literature on natural monopolies and the extensive works that consider whether particular railways exhibit the characteristics of a natural monopoly suggest that social costs should not be taken into account. The literature shows that economists have tested for natural monopoly by only taking into account costs of production: see by way of example only David Evans and James Heckman, "A Test for Subadditivity of the Cost Function with an Application to the Bell System" (2006) 74(4) The American Economic Review 615, 622; Paul Joskow, "Regulation of Natural Monopolies", in A Mitchell Polinsky & Steven Shavell (eds), Handbook of Law and Economics (2007), 11; John Bitzan, Railroad Cost Conditions - Implications for Policy (10 May 2000), prepared for the Federal Railroad Administration, US Department of Transport, 41-46; Marc Ivaldi and Gerard McCullough, "Subadditivity Tests for Network Separation with an Application to US Railroads" (2008) 7(1) Review of Network Economics 159, 165.
[844] The second reason why a natural monopoly test should not take into account all social costs is that, by the nature of the inquiry, many of those costs would not be taken into account even if known. A natural monopoly test is a static test. It assesses the state of an industry at a given point in time by taking a set level of demand and technology. Indeed, a common criticism of the natural monopoly approach is that it fails to take into account dynamic issues such as, for example, the social benefits of facilities-based competition, where competition is enhanced by each firm having its own facility, encouraging it to innovate (and hopefully to lower its costs) to capture market share. Many of the social costs which the incumbents say will be caused by access are dynamic in nature - for example, delays to expansions or the retardation of technological development. It may be possible to annualise some of these costs, but leaving aside the difficulty involved, there remain many dynamic costs which are incapable of being quantified.
[845] The third reason is that, in this case, the social costs which the incumbents urge should be taken into account are, speaking rather loosely, the cost of the diseconomies and the inefficiencies that are said would result from access, those costs being largely in the form of lost production associated with activities in a downstream market. In the Tribunal's view, this confuses the cost of production of the service with the cost of providing access. The diseconomy and inefficiency costs, if they are incurred, are only incurred because of the incumbent's participation in a downstream market. But that is just an historical accident. In other cases, the incumbent owner may have no involvement in a dependent market. It could hardly be supposed that whether or not a facility displays natural monopoly characteristics depends upon whether or not the owner is engaged in a particular trade.
[846] It should be stressed that just because a natural monopoly test does not take into account social costs does not mean that those costs are irrelevant. The costs are clearly relevant to criterion (f) and, perhaps, as discretionary factors. It is just that they are not relevant to criterion (b).
[847] A related issue is the extent to which a natural monopoly test should take into account above rail costs. Criterion (b) is concerned with the economics of developing another facility to provide the service. There are some costs which are unambiguously costs of providing a below rail service (eg capital and operating costs of the track and coordination costs for scheduling and negotiating the allocation of "rail slots"). There are other costs that are not strictly below rail costs, but are costs which are consequential on whether or not a second line is built. For example, if a second line is built it is likely that newer, more efficient, trains would be used, whereas this may not be possible on an existing line. Those costs are properly characterised as above rail costs. The Tribunal considers that taking into account above rail costs in criterion (b) conflates the provision of below rail and above rail services. These costs may, of course, be relevant to other criteria.
[848] There are three main ways to test for natural monopoly: engineering cost analysis, survivor analysis and econometric analysis. The first involves analysis of firm-level data. Survivor studies involve examining data on the growth of industries to determine those firms, assumed to be efficient, which increase market share over time and those firms, assumed inefficient, which decrease market share over time. Econometric studies involve examining cost and output data on firms in a particular industry over a period of time and applying statistical techniques to estimate the cost function for the industry.
[849] Testing for a natural monopoly is notoriously difficult. Mr Sundakov, an economist, recognised the problem with implementing the test because of the difficulty in obtaining relevant cost information. Nonetheless, he said in the course of the experts' conference that "a traditional natural monopoly test, which is a re-examination of the costs of production with and without duplication, is a test that can be universally applied." As regards the difficulties in application, he explained that: "We have a choice here of having a test [the privately profitable test] that is easy to apply but looks at the wrong thing and doesn't quite give us the right number, versus the [natural monopoly] test that we recognise is difficult to apply but tries to answer the right question". Mr Sundakov suggested that it did not seem to be right to "go with the simplicity of the answer if it doesn't quite try to answer the right question". We agree.
[850] In the present context, the question comes down to this: Can each line provide society's reasonably foreseeable demand for the below rail service at a lower total cost than if provided by two or more lines? The relevant costs are, as we have said, the costs of producing the below rail service.
[851] An important assumption of this enquiry is that an existing line can, if necessary, be expanded to meet the reasonable foreseeable demand for the service. This is consistent with the economic theory of a natural monopoly, which takes into account the ability of the facility (or, more classically, the firm) to expand the relevant output: see eg Carl Kaysen and Donald Turner, Antitrust Policy - An Economic and Legal Analysis (Harvard University Press, 1959); Richard Posner, Natural Monopoly and Its Regulation (30th Anniversary ed, Cato Institute, 1999). In the case of an incumbent's line, the additional costs to be taken into account are of operating the line on a shared basis plus the capital cost of any expansion that is necessary to meet the demand. Those costs are to be contrasted with the sum of the costs of operating the incumbent's line (plus the cost of any expansion) for its own use and the cost of constructing and operating a new line(s) to meet third party demand.
[852] It is necessary to determine a point in time at which to calculate whether it is uneconomic to develop an alternative facility. The obvious answer is that the facility must have that characteristic at the time of declaration. However, an access declaration may be made to continue for up to 20 years. It is appropriate, therefore, to consider what the future holds. The problem with that approach is that as cost structures change with ever-changing demand and, as technology changes, what is a natural monopoly today may not be one tomorrow. Provided the future is predictable with some measure of confidence, that future should be taken into account.
[853] In deciding whether a line exhibits natural monopoly characteristics today, and whether it is likely to do so in the medium-term future, it is, we think, appropriate to begin by looking at what the position would be around 2015. We think this is appropriate for several reasons. First, this is the earliest an access seeker could most likely take up access having regard to the likelihood of appeals, the time consumed in negotiations and arbitrations, and the time it would take to construct the appropriate infrastructure. Second (and partly with the first reason in mind), the Tribunal has detailed capacity modelling of the lines (except for Goldsworthy) at around 2015. Third, if a line exhibits natural monopoly characteristics at that time, it is likely to do so now. Our reason for this proposition is simple enough. If in 2015 (or thereabouts) society's demand can be more efficiently accommodated on one line than on two or more, third party demand before 2015, which is likely to be significantly lower, will be more efficiently accommodated on one line due to the high upfront capital costs of constructing a new line, regardless of the level of demand.
[854] Given that plans for the lines beyond 2014/2015 are still at a very early stage, it is impractical to conduct a detailed assessment of the lines beyond then.
[855] In summary, to determine whether a facility is a natural monopoly, it is necessary, first, to determine the reasonably foreseeable potential demand for the facility (strictly the service provided by the facility), and then compare the capital and operating costs of a shared facility to the sum of the capital and operating costs of an existing facility (or an expanded existing facility) and a new facility.
51 The Tribunal then considered the evidence relating to the reasonably foreseeable potential demand for the facility. It is not necessary to set out this reasoning here. It is sufficient to note for the present that the Tribunal's findings in this regard will be considered in relation to the procedural fairness issue.
52 In applying the natural monopoly test in this case, the Tribunal reached a different view from the experts who appeared before it in holding that a natural monopoly test under criterion (b) should not take into account all social costs. These costs were seen by the Tribunal as factors to be taken into account in relation to criterion (f) as the residual discretion (Reasons [846]).
53 The Tribunal noted that for the purpose of its natural monopoly test reaching an assessment of the costs that arise in relation to the Robe line is complicated by the possibility of the Hamersley line being declared. This is dealt with in the Tribunal's reasons at [1001] to [1002]. At [1005] the Tribunal identified that the minimum capital savings to be had from sharing the Robe line are likely to be somewhere in the vicinity of $455-651 m. This was arrived at by taking the cost of duplicating the Robe line ($875m) less the cost of additional consists (ranging from $224 m to $420 m) used as proxies for expansion costs. In the upshot, the Tribunal concluded on the evidence that the Robe line (at [924]-[929]) and the Hamersley line (at [930]-[937]) are natural monopolies.
54 As to the limited declaration made by the Tribunal in relation to the Robe service, the Tribunal noted that access to the south-west section of the Robe line (which runs from Mesa J through to the junction at Western Creek) has significant benefits and relatively minor costs. The Tribunal acknowledged at [1333], that third party constraints in relation to this area would be minimal. At [1334] the Tribunal discussed the greater complexity of use of the northern section at Western Creek through to Cape Lambert. That finding is set out in the last sentence of [1334]:
RTIO's current expansion plans would not fully utilise the capacity of the northern section until at least 2018.
55 The Tribunal gave further consideration to the application of the private profitability test. The Tribunal first revisited its reasons for rejecting the private profitability test on points of principle. The Tribunal said at [952)]-[959]:
[952] While we have rejected the privately profitable approach to criterion (b), we think it may be of assistance for the Tribunal to express its view on (1) how a privately profitable test should be applied and (2) how it would apply on the facts here.
…
[953] Three questions arise: (1) What is meant by "profitable"?; (2) What is meant by "profitable … to develop another facility to provide the service"?; and (3) What is meant by "anyone" in the expression "uneconomical for anyone"?
[954] As to (1), to an accountant profit means sales revenue minus the costs incurred in producing that revenue (eg wages, rent, raw materials, etc). To an economist profit is the minimum dollar value necessary to attract a firm to an industry or to induce the firm to remain in it. This will require credit to be given to all the opportunity costs incurred, including the cost of capital. An activity will be profitable if the return on a unit of output, less expenses (including depreciation and the opportunity cost of capital), is equal to or greater than the marginal cost of producing it. This is in line with the approach of Dr Fitzgerald, who said that an activity will be privately profitable if the entrant could cover "the full private costs of the substitute facility (ie both operational and capital costs), including a normal rate of return on capital, so as to compete in the dependent market." We prefer the approach of Dr Fitzgerald.
[955] As to (2), the question raised is whether the profit must be generated directly by the provision of the service on the new facility, or whether it is sufficient that a downstream activity generates a profit for which the service is an input. The incumbents favour the latter view. They argue that the private profitability test would be satisfied if there is a mining company for which building a railway would be profitable, whether or not the cost of constructing and operating the railway is subsidised by profits from a downstream activity (eg the sale of iron ore). The incumbents' position is to be contrasted with a requirement that a person could profitably develop a new railway as a stand-alone business.
[956] As to (3) (what is meant by "anyone"), there are two views. One, favoured by the incumbents, is that "anyone" means "any particular individual who can be identified", including, for example, mining companies whose iron ore is sufficiently valuable to subsidise the cost of building the alternative rail facility. In other words, criterion (b) will not be satisfied if only one person (eg FMG with its special characteristics of a mining company with a large resource) may find it profitable to construct a railway line to transport its iron ore to a port. An alternative view is that "anyone" means anyone at all, asking whether any hypothetical person could build a rail line and make a profit directly from providing below rail services. In this sense, the approach to the meaning of "anyone" is linked to whether the relevant profit-making activity must be the provision of the service, or some downstream activity for which the service is an input.
[957] There are conflicting arguments regarding which approach is correct. The incumbents' approach is favoured, they say, because it tests for whether a facility is truly a bottleneck. While we have previously rejected the bottleneck approach, it is clear that, if criterion (b) is about bottlenecks, then the incumbents' approach tests for this. The alternative approach does not - it tests the viability of establishing a stand-alone below rail business. It may be unprofitable to establish a stand-alone business, but if it were worthwhile for a miner to build a new railway in order to generate profits from downstream activities, the existing railway would not be a bottleneck.
[958] On the other hand, the plain words of the statute support the alternative view. The statute only refers to it being uneconomical for anyone to develop a facility to provide the service. In other words, the return that would render the construction of an alternative facility profitable is the return from the provision of the service produced by the use of the facility. On this approach, it is not relevant that a reasonable return can be obtained from the provision of some other service or the supply of goods for which the criterion (b) service is an input.
[959] In the end, the Tribunal does not accept the incumbents' approach. The alternative approach accords with the plain meaning of the words in criterion (b) and sets up an objective standard. Whether or not criterion (b) is satisfied is not to be viewed from the perspective of any particular firm. Rather, criterion (b) looks at the position of a hypothetical firm and asks whether that firm could obtain a reasonable rate of return on capital if it duplicates the facility.
56 One may pause at this point to note that the text of s 44H(4)(b) does not include a requirement that "anyone" who might economically develop another facility to provide the service should be able and willing to do so as a "stand alone business". It may be that as a matter of fact it is economical for a vertically integrated participant in the market place to develop another facility to provide the service because of the profitability of some downstream activity of that person. There is no indication in the text or context of s 44H(4)(b) that it is concerned to broaden the gateway to access based on an imagined state of affairs rather than the facts of the market place in which access is sought. As will be seen, it is not necessary to come to a final view on this aspect of the Tribunal's reasoning in order to decide these appeals.
57 Against the possibility that its interpretation of criterion (b) might be unduly favourable to Fortescue, the Tribunal proceeded to give its reasons for concluding that even if the private profitability test were adopted for criterion (b), Fortescue would fail to satisfy that test, in relation to both the Hamersley and Robe lines. The Tribunal said at [960]-[961] and [964]-[965]:
[960] Little evidence was put forward regarding the profitability of establishing a stand-alone alternative rail haulage business. The profitability of such a business would no doubt depend on a variety of factors, most fundamentally the demand for the service, the price which could be charged and the margins to be expected. One would need also to take into account that if too high a price is charged, customers may go elsewhere or look to build their own line. In the end, there is not enough evidence for us to be satisfied that it would be unprofitable to build an alternative railway (ie as a stand-alone business) to any of the lines for which a declaration is sought.
[961] There is, of course, considerable evidence about the profitability of miners building railways as part of their integrated mining operations. We will summarise the conclusions that can be drawn from that evidence.
…
[964] In relation to the Hamersley line, RTIO argues, and we accept, that it is profitable for the Dixon line to be built. This, however, would be developing a facility which provides only part of the Hamersley service. Whether it is profitable to duplicate the remaining parts of the line - which extend for quite a considerable distance - is debatable. Mr Taylor has assessed the prospect of various projects in the vicinity of Yandicoogina building their own railway. However, Mr Taylor only assessed the profitability of building a railway for those projects to connect with the Chichester line. Such a railway, which would be going in the opposite direction to the Hamersley line, would provide a different service to the Hamersley service. We have assessed potential demand from that area as 32 mtpa. That might be enough to justify building the Rosella to Yandicoogina part of the railway to connect with the Dixon line (given Mr Tapp's suggestion that a 30 mtpa project would be able to attract financing for a 260 km line). Similarly, it is possible, but ultimately quite unclear, whether there would be sufficient demand for other parts of the line to be replicated to connect with the Dixon line. Ultimately, we cannot be satisfied that it is unprofitable to develop an alternative railway to provide an equivalent service to the Hamersley service.
[965] As regards the Robe line, it is likely to be profitable for Aquila to build its proposed line. Given that the Aquila line follows a quite different course to the Robe line, the Aquila line would not be providing the same service as the Robe line. Nonetheless, because it appears to be profitable to build a line in the vicinity of the Robe line, we could not be satisfied that it would be unprofitable to build a new line to provide the Robe service.