The Wanadoo case Jérôme Philippe ACE 2007 Conference, Toulouse
The context : launch of broadband ADSL internet services in France (march 2001) Wanadoo: ISP (Internet Service Provider), subsidiary (70,6%) of France Telecom France Telecom also active on wholesale offers: IP/ADSL access IP collect Internet connectivity Wanadoo offers broadband internet access through ADSL: Wanadoo ADSL (without ADSL modem) Pack eXtense (with ADSL modem) Wanadoo competing with other ISPs
The Commission’s decision of 16 July 2003 March July 2001: Price < Average variable cost August 2001-October 2002: Average variable cost < Price < Average full cost « eviction strategy » on the French broadband internet access market Fine: € 10,35 million CFI confirmed Commission’s decision Appeal pending at ECJ
The decision’s test for predatory pricing : AKZO (1991) Test inspired from seminal article by Areeda and Turner (1974) Compares price and cost of the alleged predator (no mention of « victim’s » costs) If Price lower than average variable cost = per se abusive if dominance (because there can exist no other goal for such prices than elimination of competitors) If Price lower than average full cost but higher than average variable cost = abusive if set in the context of a plan of elimination of a competitor Note that costs and not fixed or variable by nature ; can depend on contracts (example: hotline cost)
Defences that are not taken into account or rejected in the decision Initial takeoff of the product & market Decrease in monthly cost for each client during « client’s life period » (4 years) Absence of recoupment and of possibility of recoupment Meeting competition defence
Revenues and costs: static, static plus or dynamic approach?
ADSL retail offer: main costs and revenues Each subscriber, when he/she joins, triggers intertemporal costs and revenues on a long period of time (duration of subscription) Costs —Recurring variable costs: network costs and some client management costs —Recurring fixed costs: some client management costs, admin. costs —Acquisition costs (one shot): advertising, promotions, connection Revenues —Subscription (monthly) revenues —Advertising, pay services Average length of subscription : 48 months During subscription: very significant decrease in network costs
Pure static approach (mentionned but not retained by Commission) Revenues = all revenues of year Y Costs = all costs of year Y Ratio revenues / costs is very low, since many new subscribers and small installed base : high proportion of acquisition costs acquisition costs shall be compensated in 4 years, not one, for the subscriber to be profitable Gives ratio revenues/costs of about : 30 % from January to July 2001, 60% from August to December 2001, 83 % from January 2002 to June 2002, but economically not relevant
« Static plus » approach (used by the Commission) Monthly analysis Revenues = monthly revenue of new subscribers Costs = [monthly recurring cost + (acquisition costs) / 48] of new subscribers Example for a Pack eXtense subscriber on 1st June 2001 Costs = acquisition cost / 48 + monthly costs as they are on 1st June 2001 Revenues = monthly subscription revenues as they are on 1st June 2001 Note : this subscriber will stay until (on average) May 2005 Problem = profile of monthly recurring costs (network costs) June-July 2001: C August 2001 – January 2002: C / 1,5 February 2002 – July 2003 (date of decision): C / 2 Post July 2003: even lower Meanwhile, profile of monthly revenues is flat (contract with subscriber unchanged)
« Static plus » approach (used by the Commission): results Period of March July 2001: Revenues on variable costs: 60,3 % Period of August 2001 – December 2001: Revenues on full costs: 89,5 % Period of January October 2002: Revenues on full costs: 92 %
Dynamic Approach: Net Present Value by subscriber (refused by Commission) “in many businesses certain costs legitimately yield benefits in time periods other than those in which they have been incurred. With such costs, it is appropriate that they are recovered over time.” (OFT, Freeserve decision) Net present value (NPV) / discounted cash flow (DCF) is used to assess companies or business plans Big difference : if monthly recurring costs decrease during the subscription period, this is considered in the subscriber’s NPV We consider that a right way to apply the AKZO test would have been to apply the AKZO test to the NPV of each new subscriber : If a new subscriber gives a positive NPV, then acquiring him/her is not predatory If NPV on variable costs is negative, then predation If NPV on variable costs in positive but NPV on full costs is negative, then depends on whether there is (or not) a strategic plan of elimination Wanadoo : all new subscribers (apart from those of March 2001) are NPV positive
Dynamic Approach, Net Present Value by subscriber (refused by Commission): results Proposition : since Commission’s decision is of July 2003, calculate NPV using real figures (cost and revenues) until July 2003, then freeze the July 2003 values for the period posterior to July 2003 then: all new subscribers (apart from those of March 2001) are NPV positive ratio price/cost (Pack Extense): Revenues on variable costs: - from 103 % (March 2001) to 116 % (Sept 2002) Revenues on full costs: - 99 % in March from 100 % (April 2001) to 113 % (Sept 2002)
Net Present Value – Case Law OFT BSkyB decision (squeeze effect): DCF rejected Futures costs and revenues difficult to foresee Positive NPV can also show that abuse (predation) was a profitable strategy OFT Freeserve decision (squeeze effect) “The Director is aware of the criticisms of the DCF approach and has made certain adjustments to take these into account when conducting his investigation.” French Competition Council decision of 6 November 2002 on Direct Insurance (Decision n° 02-D-66) DCF is used (on an emerging market)
Net Present Values – Why the Commission refused to apply it Rejects any kind of retrospective analysis « Necessary to assess whether a behaviour is abusive at the time when it takes place » Intent or objective behaviour? What is important in not individual net outcome per subscriber but an overall assessement of the activity’s financial situation Says that a high growth rate it naturally predatoty !!!! (para 93-94) Most serious criticism: a positive NPV may be the signal of efficient predation if the period used for calculation includes recoupment This is true: wasting money to get it back later by recoupment is the essence of predation Is there an indistinguibility theorem?
Is there an indistinguibility theorem? A positive NPV can correspond to 2 situations : —Normal story: new service triggers high costs (acquisition + no learning by doing yet + no scale effects yet), then development of the service leads to break even and profitability is obtained after some time —Predation story: absence of initial profitability drives competition out of the market, then price can be increased to supra-competitive levels, which makes for initial losses How to distinguish ? One solution only: assess whether the profit making period is made in a competitive or non-competitive context This boils down to assess whether there has been effective recoupment or whether there exists a possibility of recoupment ie analysis of recoupment / possibility if recoupment cannot be ruled out
In the Wanadoo case… The Commisison refused to formally include an analysis of recoupment (before July 2003) or possibility of recoupment (after July 2003) Howeber the decision mentions that after October 2002 (ie when profitability starts), the market is competitive (para 386) This rules out the possibility that profitability after 2002 results from anticompetitive recoupment… but the NPV of subscribers is still positive This should have lead to absence of predation but… CFI : « the Commission is right to consider that revenues and costs posterior to the infringement should not be taken into account in order to assess the price/cost test during the infringement […] Wanadoo is thus not allowed to include any element of price or cost posterior to October 2002 in its calculation of price-cost test» (para 152)
Recoupment
Economic theory: recoupment is essential in predation necessity of barriers to entry Exclusion of competitors Recoupment with supra-competitive prices European law: Tetra Pak: additional evidence of recoupment is not needed with respect to the circumstance of the case AG Fennelly, Compagnie Maritime Belge: recoupment is part of AKZO test Member States: France: no need to show recoupment, but must show possibility of recoupment, AOL France and AOL Europe (n° 04-D-17, 11 Mai 2004); UGC Ciné-Cité (n° 04-D-10, 1 Avril 2004). UK: ruling of the CAT in Aberdeen Journals (2003) rejects condition of recoupment USA: Matsushita Cargill Brooke Group
Brooke v. Brown&Williamson, US Supreme Court, 21 June 1993 « Without recoupment, even if predatory pricing causes the target painful losses, it produces lower aggregate prices in the market, and consumer welfare is enhanced », Brooke « The plaintiff must demonstrate that there is a likelihood that the scheme alleged would cause a rise in prices above a competitive level sufficient to compensate for the amounts expended on the predation, including the time value of the money invested in it », Brooke « For recoupment to occur, below-cost pricing must be capable, as a threshold matter, of producing the intended effects on the firm’s rivals, whether driving them from the market or causing them to raise their prices to supracompetitive levels in a disciplined oligopoly (See Areeda & Turner) », Brooke
Matsushita v. Zenith Radio, US Supreme Court, 26 March 1986 Cargill v. Monfort of Colorado, US Supreme Court, 9 December 1986 « In order to recoup their losses, [predators] must obtain enough marker power to set higher than competitive prices, and then must sustain those prices long enough to earn in excess profits what they earlier gave up in below-cost prices», Matsushita « Evidence of below-cost pricing is not, alone, sufficient to permit an inference of probable recoupment and injury to competition. Determining whether recoupment of predatory losses is likely requires an estimate of the cost of the alleged predation and a close analysis of both the scheme alleged by the plaintiff and the structure and conditions of the relevant market (See Elzinga & Mills) », Cargill ‘It would be ironic indeed if the standards for predatory pricing liability were so low that antitrust suits themselves became a tool for keeping prices high », Brooke
Recoupment is not part of the Commission’s test for predation « showing the possibility to effectively recoup initial losses is not required by European law » (para. 338) It is sufficient to show that price < costs (and intent in some cases) However, in the Wanadoo case (good is a subscription service), assessing recoupment (before July 2003) and possibility of recoupment (after July 2003) would have been the only way to distinguish normal behaviour from predatory behaviour Predation is different from reaching break-even
Other topics How to assess the existence of a « plan of predation »? A few individual s are not sufficient It should not be an infringement to have (relatively) aggressive words on competitors « Meeting competition » defence Totally rejected by the Commission for a dominant company (but is not needed for a non-dominant company!) Exit of a competitor How to distinguish natural exit from exclusion? « That below-cost pricing may impose painful losses on its target is of no moment to the antitrust laws if competition is not injured », Brown Shoe
CFI Wanadoo ruling of 30 January 2007 (T-340/03) Upholds the Commission’s decision « the Commission is right to consider that revenues and costs posterior to the infringement should not be taken into account in order to assess the price/cost test during the infringement […] Wanadoo is thus not allowed to include any element of price or cost posterior to October 2002 in its calculation of price-cost test» (para 152) « although it is true that meeting competition is not in itself abusive, it could not be ruled out that it becomes abusive when it aims not only at protecting its interests, but also at reinforcing dominance and abusing of it » (para 187) « scale economies and learning by doing cannot, by nature, change the conclusion on predation, since the dominant company can benefit from these effects precisely thanks to predation » (para 217) « the Commission is right in considering that it shall not demonstrate recoupment » (para 228) « it was not necessary to show that Wanadoo had a real possibility to recoup its losses » (para 227) Appeal pending at the ECJ
Towards a structured test? Would fit economics better… Inspired by Bolton, Brodley & Riordan (2000), post-Brooke proposals in the USA: 1. Market study: Does market structure make predation possible? Is there a likely possibility of recoupment? 2. Case study: Is there evidence of a scheme of predation? Are there significant sales with prices effectively lower than variable/full costs? 3. Defence: If yes, are there lawful business justifications for achieving these sales?