This article, available here, reviews the use of cost standards in selected European abuse of dominance cases. It shows that a variety of cost standards were employed until recently, and criticises the ECJ’s case law for ignoring challenges with identifying the appropriate cost standard for each case. To address such challenges, it is important to identify the key questions a price–cost test should answer, and agree on the features of such a test.
The paper is structured as follows:
Section II summarises current knowledge on price-cost tests.
This knowledge is outlined in the European Commission’s Article 102 Enforcement Guidance, which puts forward two widely accepted concepts: Average Avoidable Cost (AAC) and Long Range Average Incremental Costs (LRAIC). These tests provide the benchmarks for predatory behaviour in Europe. There is a legal (but rebuttable) presumption that prices below AAC anticompetitively foreclose competition; that prices above LRAIC do not to raise concerns; and that prices between AAC and LRAIC require consideration of other factors that could lead to foreclosure, such as the existence of an exclusionary strategy.
Average Avoidable Cost (AAC) is the lower bound average cost benchmark, for which the Areeda-Turner average variable cost standard provides a proxy. AAC includes all costs that the dominant firm could avoid in the short-run if it no longer produced a given amount of output – including both variable costs (that vary directly with units of output) and fixed product-specific costs (that could be avoided if output was reduced over the relevant period). This test provides a strong indication of profit sacrifice. Long Range Average Incremental Costs represents the upper bound, for which average total cost is a proxy. It includes not only those costs that are incremental over the short-run, but also those product-specific fixed costs (either sunk or not) that would be incurred over the longer period if output expanded. LRAIC would be lower than the average total cost, as it exclude costs that are common or joint across products.
Section III puts the dominant price-costs tests in their historical context.
From an economic perspective, the Chicago School’s criticism that predation could not be a rational strategy has been confined to when there is perfect information. When this is not the case, there are a number of reasons why companies may engage in predation. First, predation can be successful if the predator is able to build a reputation for being tough, which can then be used to extend the reputational effects of predation to other markets. Second, the predator can succeed in signalling that it has very low costs or that demand is too low to sustain a profitable competitor. Third, the predator can exploit imperfect information in capital markets so that the prey’s decline in performance is misinterpreted by investors as inefficiency or underperformance, leading investors to withdraw their financial support. Fourthly, a dominant firm that has already sunk some costs could deprive a more efficient entrant from enjoying the required economies of scale by predating. Crucially, in none of these models does the predator need to price below its short-run marginal cost (for which AAC is often considered a proxy) to deter entry or induce exit.
A seminal article by Areeda and Turner bridged economic and legal tests of predation. They suggested that, absent efficiency or pro-competitive justifications, a dominant firm pricing below its marginal cost should be presumed to have engaged in predation. Because of difficulties in estimating marginal costs, they pragmatically suggested that one could rely instead on average variable costs as a good proxy for predation. This test has attracted criticism for being too simplistic, e.g. it fails to capture strategic factors and long-run welfare effects which may justify predation below average variable costs or may allow for predation above such a cost benchmark. It was in this context that Baumol introduced AAC and LRAIC. These were intended as flexibles concepts that could take into account both the short- and longer-run timeframes of both single- and multi-product firms. While earlier EU case law used the Areeda-Turner test (most notably in the AKZO judgment), the European Commission and national competition authorities – most notably in the UK – have progressively adopted Baumol’s approach.
Section IV considers the key questions that a price–cost test tries to answer and examines in detail some still unresolved issues that are critical to correctly estimating costs.
While the legal test of predation focuses on pricing below some measure of cost, there are no conceptual reasons not to expand the notion of predation to cover above-cost actions that entail a profit sacrifice when this can only be explained by attempts to foreclose competitors. The relevant question for predation is not whether it is pricing below cost, but whether such conduct is profitable for the incumbent. However, for this to pose a competition issue requires that the profit sacrifice (expressed as absolute loss) could only, or principally, be justified by the exit or marginalisation of competitors, absent which it would irrational to sacrifice profits. This is a critical and often neglected aspect in assessing predation. The problem is that it has no connection to the narrow cost standard question that competition agencies address in their enforcement efforts – i.e. the legal test is unsuited to identify and deter most instances of predation.
There are other challenges with the use of a price-cost test for predation. Often the terms avoidable and incremental are used as synonymous in the context of cost standards. However, it is well known that estimating costs as either incremental or avoidable for the same output does not necessarily lead to the same value – this can depend on case-specific factors, e.g. sunk costs or economies of scale. The relevant time-period for assessing an alleged abuse can also have implications for cost estimates – after all, whether a price is variable or fixed can depend on the adopted timeframe. Moreover, if a competition agency knows the timeframe of an abuse the distinction between fixed and variable (or incremental and avoidable) cost becomes irrelevant, since one can focus on which costs were incremental within the actual time horizon over which predation occurred. Another challenge concerns whether one should use ex post actual costs or the costs that the predator expected to incur at the time it took the decision to reduce its prices. A similar difficulty tends to arise concerning the decision of whether the scope of cost increases should be determined by reference to the entire post-predation output of the infringing firm or just the output directly affected by the price reduction. Identifying such outputs is, in any event, even more difficult than identifying the pre- and post-predation outputs, as predation may last long and several factors may affect the dominant firm’s output level over the period. In practice, agencies typically find that the relevant output increment is the dominant firm’s total post-predation output, absent price discrimination. Other issues that can arise are whether opportunity costs should be taken into account – e.g. by reference to the most profitable use to which the inputs to increase output would have been put to absent predation; and how to apply price cost tests in network industries, where fixed costs are lumpy, may be affected by regulatory obligations and amount to the largest component of a firm’s total cost.
Section V argues that the difficulties with using price-cost tests used for identifying predation also arise in the context of other price-based exclusionary abuses.
Throughout this article, the author uses predation as the paradigmatic example of price-test costs. However, his analysis extends to other types of price abuses. Irrespective of the mechanism behind the abuse—that is building a predatory reputation, information asymmetries, capital market imperfections, or economies of scale—predation always involves the predator expanding its output for a period of time. This leads the author to suggest average incremental costs as the appropriate short- and long-term standard. As long as the dominant firm expands its output to foreclose competitors, an average incremental cost standard seems appropriate. This logic largely applies also to other price-based abuses. For example, margin squeeze involves the dominant firm expanding its downstream output, as its aim is to capture downstream sales from its rivals and divert such sale to its own downstream arm. A margin squeeze may not increase the downstream total market output if it involves no changes in downstream prices (but instead a higher upstream price charged to rivals only); what matters is that it involves a downstream output expansion by the incumbent, gaining sales from its downstream rivals.
There are a number of other price-based abuses – such as bundling, rebates, and exclusivity – where it is still debated whether price–cost tests are the best approach. The author considers that an average incremental cost would also be the most appropriate test in these circumstances. For example, in a bundle the main question is whether the incremental price that customers pay for each of the dominant undertaking’s products in the bundle remains above the LRAIC of the dominant undertaking from including that product in the bundle. And in exclusive rebates, the question is whether an as effective competitor could price the contestable part of demand at a certain price. Since an abusive exclusionary pricing practice will always require an expansion of output by the dominant firm, the author considers that ‘average incremental cost is the appropriate cost standard for all price-based cost abuses’.
Section VI concludes by offering some practical suggestions on how to estimate costs.
The author makes six practical suggestions. First, price–cost tests should focus on incremental costs, irrespective of how long the abuse lasted or was expected to last. Second, whenever possible the time horizon should be the period over which predation took place. This makes the distinction between short- and long-run standards less relevant, other than perhaps in cases where it is unclear how long the abuse lasted. Third, the size of the relevant increment must be closely aligned with the prices of the goods and services affected by the abuse. Fourth, when there is evidence that the dominant firm switches inputs from other services to expand its output, the opportunity cost of such action should be factored into the estimation of its incremental costs. Fifth, when there are linkages on the demand- or supply-side, these may require adjustments to the scope of the activities or services for which we should estimate the costs. Lastly, one needs to be extremely careful in using any LRAIC estimates that may have been put together for a different purpose, such as setting ex ante price controls in a regulatory context.
This paper provides an interesting discussion of the challenges that relying on cost-based benchmarks poses to competition law enforcement against price-based practices. I found the section devoted to describing these difficulties – which contains multiple examples of how authorities in Europe dealt with these challenges in practice – most interesting. I was also taken by the argument that predation can occur when pricing is above cost, which is the topic of the last paper reviewed today.
I do not, however, partake the author’s optimistic approach to price-cost tests, in particular his view that they should be extended to all exclusionary practices. As a matter of law, I think this is wrong – there are price-based abusive practices that do not require enforcers to pursue a cost-based analysis. Examples of this include exclusivity rebates or constructive refusals to supply. Furthermore, there are a number of practices where pricing above incremental cost can have exclusionary effects, such as when such practices are directed at conduct that prevents competitors from gaining the necessary scale to lower costs and become effective competitive constraints – as discussed in the papers reviewed above.