The use (and abuse) of price-cost tests for predation Lessons from the Wanadoo case

1. ACE, Toulouse 29 November 2007 The use (and abuse) of price-cost tests for predationLessons from the Wanadoo case Miguel de la Mano* Member of the Chief Economist’s Office DG COMP, European Commission *The views expressed are those of the author and do not necessarily reflect those of DG COMP or the European Commission

2. Definition Ordover & Willig (1981) define predatory conduct as a strategy “that sacrifices part of the profit that could be earned under competitive circumstances were the rival to remain viable, in order to induce exit and gain consequent additional monopoly profit”. • Much broader than just pricing • Strategy in two stages, which is (partly) how to distinguish it from normal competition • Tricky part: why/how does the predator’s behavior today influence whether the prey wants to be in the market tomorrow?

3. Predation is not rational • Predation is costly: and increases with the market share of the predator, while the victim’s losses are smaller, the smaller its market share. • Since predation can only be temporary, the prey will not exit. • Not even a dominant firm can successfully predate on equally or more efficient rivals. • A predator would ultimately raise prices or behave less aggressively to recoup initial loses. • If the industry is profitable in the long term, lenders should be prepared to back the prey through any period of temporary losses. • Predation cannot lead to permanent exclusion: Even if the prey ceased operations during the predatory phase, either it or a successor would reenter during the recoupment phase, making use of the prey’s original assets.

4. Insight 1: Predation may work if prey is financially constrained • The prey is dependent upon some source of external financing (i.e. it is financially constrained) • The predator seeks to manipulate that relationship between the prey and its investors. • For example, the predator may reduce prices in order to reduce the profitability of its rivals. • Lenders may be unable to determine whether the default stems from (a) predatory pricing, (b) or the debtor’s poor performance or (c) see low profitability as as a signal that prospects in this market are limited. • Lenders may decide to pull the plug

5. Insight 2: Asymmetric information reinforces exclusionary effects of predation • Rivals will enter the market if they believe the dominant firm is a high-cost provider, but will not enter the market or will choose to exit the market if they believe the dominant firm is a low-cost provider. • A predator may drastically reduces prices to mislead the prey to believe that the predator has lower costs and to exit the market. • Observing the predator’s low price, the prey rationally believes that there is a least some probability that the predator has reduced costs. This lowers the prey’s expected returns and causes the prey to exit. • Similar models: test-market predation (secret price cuts) signal jamming (public price cuts) • In all cases: Predation to mislead rivals in believing the market is unprofitable

6. Insight 3: Reputational effects can make predation a cheap and effective strategy (across time and space) • The predator seeks to convey a reputation for “toughness” and a willingness to defend its market at virtually any cost. • The predator reduces prices in one market to induce the prey and potential entrants to believe that it will cut price at a later time or in other markets. • The predator seeks to establish a reputation as a cut-throat competitor, based on some perceived special advantage or characteristic.

7. Key criteria in modern theories of predation • Models require some type of asymmetric information—in the possession of the predator that is not common knowledge. • Models typically assume that the predator enjoys some financial or cost advantage over its prey. • If the prey is in a superior financial position or if it is known to have lower costs than the predator, there is no real prospect for predatory behavior. • Models are of limited relevance when the prey’s presence in the predatory market is driven primarily by strategic or defensive considerations rather than financial considerations.

8. Spirit case (US) • Textbook example of what predatory pricing would look like: • Northwest’s prices in the Detroit-Boston and Detroit-Philadelphia city pairs are high. • Spirit enters, Northwest’s prices fall dramatically. • Spirit exits, Northwest’s prices jump up. • Spirit Airlines pulled capacity out of Detroit quickly when Northwest cut its fares but Spirit could not re-enter because Spirit had difficulties of access to gates at the Detroit airport. • Circuit Court seems to suggest that at least in the market circumstances of this case, Northwest’s conduct may have been predatory even if its fare structure exceeds “an appropriate measure of average variable costs.”

10. Policy considerations • Challenge: to distinguish highly competitive pricing from predatory pricing. A price-cutter may simply be responding to new competition, or to a downturn in market demand. • There is the possibility of Type I errors (labeling a price cut predatory when it is actually competitive- a false positive) and Type II errors (labeling a price cut competitive when it is actually predatory – a false negative). • Policy should balance the risk of error in a manner that maximizes expected consumer welfare. • Over-enforcement that results in a Type I error is likely to entail high social costs because it will give firms pause in lowering prices out of fear that such behavior will be condemned as being predatory.

11. legal ATC Requires intent AVC illegal EU traditional approach:AKZO and Wanadoo Dominance + • Note the AKZO approach is not easy to apply. • assessing costs and prices requires a detailed and complex investigation of the dominant firm’s cost structure and revenues. • price-cost tests are generally inappropriate in cases of non-price predation • the concept of intent is too subjective to be operational. Absence of intent does not prove price competition while evidence of intent does not prove predation.

12. 3-step structured rule of reason(de la Mano and Durand, 2005) • Predation is risky but effective under the right circumstances. • Proof of predation involves showing that the alleged predation: (a) has taken place and (b) is economically rational. • Need to show: • Sacrifice: The defendant has sacrificed profits. • Likely Exclusion: As a result, one or more rivals have or likely will be excluded from competing with the alleged predator in some market. • Likely Recoupment: Once rivals are excluded, the alleged predator can exercise increased market power and thereby recoup the initial sacrifice

13. Defense • Evidence of sacrifice, (likely) exclusion and (likely) recoupment creates a presumption of predation. • This presumption should be rebuttable since seemingly predatory behavior can also enhance efficiency and increase consumer welfare. The defendants can provide evidence that • (i) it has not deviated from a short-run profit maximization strategy • (ii) exclusion of rivals is due to exogenous circumstances unrelated to the defendant’s aggressive behavior and/or • (iii) the initial profit sacrifice leads to market expanding efficiencies, which fully offset their exclusionary effect.

14. Proof of Sacrifice • First best: • Evidence that the predator is not optimising in the short-run. But this requires significant information (demand, conduct, cost etc.. - now and in the future) • Second best: • Assessment of alternative course of actions • Incremental revenues vs. Incremental costs • Price-cost tests

15. Recoupment Dominance • Identity of the prey matters • Firms leave, but assets stay • Capacity constrained dominant firm Predation to acquire dominance Reputation as predator is the entry barrier Recoupment • The prospect of recouping is akin to the prospect of becoming dominant • Being dominant may not be bad proxy • But important differences: • Dominance: ex-ante market power • Recoupment: increased market power ex-post • Asymmetry of Entry and Exit Conditions matters Burden of proof on the parties? This does not resolve the under-enforcement problem

16. Structured rule of reason vs.per se rules (e.g. AKZO) • Easier to administer: profit sacrifice can be proven even without a complex and controversial analysis of average costs and average prices and without entering into a debate as to which is the relevant cost standard. • Sliding scale: the more convincing the evidence as regards likely exclusion and recoupment the less detail is required to establish sacrifice. E.g.: evidence of below cost pricing is contentious but: • likely exclusion results from financial constraints, reputational or signaling effects • Likely recoupment is shown on traditional market structure analysis post-exclusion. • Less false negatives: e.g. if a firm engages in non-price predation a comparison of prices and costs can be seriously misleading, if at all possible. • Less false positives: There is no predation if conduct is not rational, either because it cannot credibly lead to exclusion or entry deterrence, or because some competition constraints will remain making the prospect of recoupment unrealistic.

17. Wanadoo: 72% owned by France Telecom From the end of 1999 to October 2002 marketed its ADSL services known as Wanadoo ADSL and eXtense France Telecom held almost 100% of the market for wholesale ADSL services Other ISPs could and did offer similar ADSL services But needed to rely on France Telecom for building blocks Limited local loop unbundling Fast-growing (not emerging) market From January 2001 to September 2002 Wanadoo’s market share rose from 46% to 72% Five-fold increase of the market size over the same period Wanadoo case

18. Wanadoo Decision • Commission decision of 16 July 2003: predatory pricing by Wanadoo • March 2001 – August 2001: Price < Average Variable Costs • August 2001 – October 2002: • Average Variable Costs < Price < Average Total Costs • Plan to pre-empt the French high speed Internet market • CFI Judgement does not depart from earlier case law and confirms AKZO as a legally valid path to establish predation: • Assumption of abuse in case of pricing below AVC; • Prices above AVC and below ATC to be regarded as abusive if determined as part of a plan for eliminating a competitor;

19. CFI Judgement recognises the complexity of applying price cost tests: • Choice of method to establish “sacrifice” entails complex economic assessment and COMP must be afforded broad margin of discretion • Decision unlawful only if the applicant proves that the method used by the Commission is unlawful/ comprises manifest error (§153) • It is not apparent that the method of discounted cash flow (advocated by Wanadoo) was necessary in the present case

20. … and it implicitly opens the door to an effects-based analysis • “for the purposes of applying [article 82] showing and anti-competitive object and an anti-competitive effect, may, in some cases, be one and the same thing. If it is shown that the object pursued by the conduct of an undertaking in a dominant position is to restrict competition, that conduct will also be liable to have such an effect” (§195) • However, warns that identifying effects is not always possible (although the Court may be referring to a problem of timing not assessment): • “where an undertaking in a dominant position actually implements a practice whose object is to oust a competitor, the fact that the result hoped for is not achieved is not sufficient to prevent that being an abuse of a dominant position within the meaning of Article 82 EC” (§196)

21. In fact it shows that effects can (and must) be assessed… to establish fines • “In accordance with [Guidelines on Fines] in assessing the gravity of the infringement, account must be taken of its nature, its actual impact on the market and the size of the relevant geographic market”. (259) • “WIN has denied that the infringement in question impacted on the market. However, various elements indicate the contrary” (§260) • First of all, WIN’s market share on the high-speed market first increased from 50 to 72% (§261) • Secondly, one competitor, Mangoosta, disappeared from the market (§ 262) • Thirdly, during the period covered, a significant decline in the market shares of the competing cable operators was recorded (§ 263) • Fourthly, WIN’s conduct had a deterrent effect on the ability of competitors to enter the market and to develop (§ 264) • The number of new entries on the market was also marginal (e.g. Dixinet only had 10 subscribers) • Harm to consumers: • Article 82 EC is not only aimed at practices which may cause damage to consumers directly, but also at those which are detrimental to them through their impact on an effective competition structure (§266)

22. CFI confirms that proof of recoupment is not a necessary condition but… • … the justification suggests: • (a) it refers to actual recoupment and • (b) it does not want to discourage intervention before it is too late!) • First quotes Tetra judgment: ‘[I]t would not be appropriate, in the circumstances of the present case, to require in addition proof that Tetra Pak had a realistic chance of recouping its losses. […] [aim of Article 82] rules out waiting until such a strategy leads to the actual elimination of competitors.’ • Then CFI concludes • “The Commission was therefore right to take the view that proof of recoupment of losses was not a precondition to making a finding of predatory pricing.” (§217)

23. In fact… • Commission decision described in detail obstacles to entry and growth of rivals that would facilitate recoupment • Disincentives to switching on the part of existing customers (lock-in effects due to experience nature of ADSL) • Costs of entering and acquiring a critical size in a mass market (technical and promotional expenditure, e.g. to build brand image); • Cost of alternatives to FT’s wholesale access (e.g. building own network) • Easy to exploit ex-post market power due to vertical integration with FT

24. Risk of abuse of bright line tests • Typically two considerations: • Probability of making an error • Consequences of making such error • Probability of making depends on the accuracy of the test, but not only. • Suppose we use a “sacrifice only” test of predation implemented using the rule: there is predatory sacrifice if P<ATC • Assume: • In 90% of cases where a firm predates price will be below ATC (5% risk that the firm engages in non-price predation – false negative) • In 90% of cases where there is no predation P>ATC (5% risk that the firm engages in promotional or penetration pricing – false positive) • Is this a reliable test?

25. Not very reliable… • In light of the modern theories of predatory conduct assume market circumstances for rational predation are present rather infrequently (only 1% of firms will find predation rational) • Take 1000 dominant companies • Typically 10 are truly predating. If they predate the price-cost test comes positive in 9 cases (the remaining firm engages in non-price predation or above cost predation) • The remaining 990 are not predating but the test can be inaccurate for them too. Up to 99 of them will price below ATC • So there are 108 positive results in total but only 9 are accurate • So for any case where P<ATC the chance there is predation is low around 8%, not high as may appear at first sight

26. Thanks for your attention