Substantive merger control is the bar higher than
“Substantive merger control: is the bar higher than it used to be, and to what extent? ” Insights from the assessment of recent mergers Dr Cristina Caffarra 83 rd Lunch Talk of the GCLC 16 September 2016 1
MNO mergers: 4 -to-3 s now “back to 4”…. Case Year Country Type Outcome T-Mobile/tele. ring 2006 Austria 5 to 4 Cleared with remedies T-Mobile/Orange NL 2007 Netherlands 4 to 3 Cleared unconditionally T-Mobile/Orange UK 2010 UK 5 to 4 Cleared with remedies H 3 G/ Orange Austria 2012 Austria 4 to 3 Cleared with MVNO remedies Three Ireland/O 2 IE 2014 Ireland 4 to 3 Cleared with MVNO remedies Telefónica DE/E-Plus 2014 Germany 4 to 3 Cleared with MVNO remedies Telia. Sonera/Telenor 2015 Denmark 4 to 3 Withdrawn (prohibition) H 3 G/Telefónica UK 2015 UK 4 to 3 Prohibited H 3 G/ Wind Italy 2016 Italy 4 to 3 Cleared but only with a “back to 4” DIY MNO kit… 16 September 2016 2
A few more tools in the box…. Simulation “lite” and “superlite” UPP, IPRs, CMCR – various applications Bertrand price competition with differentiated products, various functional forms for demand system – multiple consumer good cases, branded goods, retail goods Bertrand-Edgeworth competition in price with homogeneous products and capacity constraints – multiple commodity cases… Cournot competition in quantities with homogenous products (and competitive fringe) – commodities, iron ore, energy… Bidding models to estimate impact of reduction in number of bidders – multiple bidding markets Competition in price, sequential moves (and vertical component) – cable/content Competition with sequential choice of product quality and price – telecoms 16 September 2016 Bargaining models – music, content, telcos, multiple others 3
“What can be said at all, can be said clearly…. ” Predictive value? What to expect from a model? Interpretation of results? But no model can provide a reliable “point prediction” of the size of price changes… How do we benchmark? What “thresholds” do we use? “All models are wrong, some are useful” (G. Box) “The purpose of computing is insight, not numbers” (R. Hamming) We can hope for: - Clear assumptions - Indication of relative importance of drivers - Order of magnitude of effects 16 September 2016 4
Issues 1. Is the analysis itself becoming so demanding that – by implication – standards seem tougher? 2. Economic tools vs… market shares? 3. Is the perception the *bar is higher* in fact reflecting there are deeper questions we don’t know (yet) how to tackle? 16 September 2016 5
1. Is the analysis too demanding? 16 September 2016 6
Some unnecessary complications. . … e. g. Liberty/Ziggo – “indirect effects”? Merging two cable operators in the Netherlands – no overlapping footprint, hence no direct loss of competition - but “indirect effect” hypothesis “Indirect” price effects (unilateral and coordinated) because national pricing policy of an unconnected third party – KPN – alleged to “competitively link” the pricing decisions over the two merging networks Just cannot be true in any model in which parties move “simultaneously”. Need to postulate “sequential pricing” (first mover, second mover, etc) Effects depend on specific assumptions about sequencing of pricing decisions of parties and KPN after the merger But how do you decide how to model the order of moves? How is it possible to make any prediction from pre-merger pricing? ? 16 September 2016 7
Some tiresome but inevitable complications…. Dealing with complainants’ models of vertical/conglomerate effects • Competitors who are also customers…well resourced and willing to invest a lot into a complaint • Responsible for comparatively quite a lot of modelling (variants of raising rivals’ costs/partial foreclosure models, bundling models) which are put to parties to deal with but typically lead to intervention in terms of remedies possibly over-enforcing in some of these cases e. g. Vodafone/Liberty JV – “foreclosure through mixed/mobile bundling” Theory: “in response to 3 P/4 P bundling by KPN, parties will also bundle fixed/mobile services – putting upward pressure on prices, locking-in customers because of switching costs, and disadvantaging non-integrated competitors” In the end often bad models with very fragile claims, but time consuming and may not avoid a Phase II or a remedy 16 September 2016 8
Some more “legitimate” complications… Bidding analyses – first price, second price auctions – does it matter? e. g. GE/Alstom First price • Bidders shade bids as a function of number of other bidders Second price • Look at how often merging parties were “winner” and “runner up” “Average effect of a reduction in the number of bidders can be quite similar, the form of the effect can be quite different depending on format of the auction”. Are we overcomplicating? Not really. But what ultimately matters is getting information on closest competitor: “participation”, “who wins”, “who’s the runner up” all contain information on closeness of competition if properly interpreted 16 September 2016 9
2. Economic tools vs… market shares? ? 16 September 2016 10
Oligopoly models pervasive to gauge price effects Intense work and seemingly arcane debate…. • Put forward by advisors or generated internally by CET • Motivation for parties: show that in a stylised setting that captures the main features of the market, merger effects are “limited” (especially if we incorporate synergies), and/or can be undone by reasonable divestments • Difficult call: how quickly will it topple over, and show the opposite? Sensitivities every which way… • Broader concern: do these models “help” with robust predictions, given they are such hyper-simplifications of reality? BUT: IS THE ALTERNATIVE JUST DOING MARKET SHARES? SOME DEALS WOULD HAVE NEVER BEEN CLEARED ON A “DOMINANCE” TEST…. 16 September 2016 11
Realistic expectations on what we can show… e. g. Ball/Rexam Ball Rexam Crown Can-Pack Other EEA sales shares 60 -70%, combined with a 30 -40% increment… More “local” markets may help salvage some more “peripheral” assets (but law of averages…) Can use econometric model to predict “who wins” contract (as a function of distance, incumbency, capacity constraint…) and then see whether candidate remedy plants can be deemed “closest competitors” and impose enough competitive constraint on the parties Problem in many cases is high value of incumbency = how incumbent is the existing plant – if it gets divested, does it retain the incumbency? Difficult to show with little data…. 16 September 2016 12
3. Does the perceived *higher bar* reflect concern we don’t really grapple properly with the deeper questions? 16 September 2016 13
We know how to do short-term price effects…(just) Huge effort on precise size of short-term price effects But do not deal well with critical “big picture” drivers of deals “Looking for keys under the lamppost”? Standard merger simulation framework completely geared towards measuring (static) short-term price effects as precisely as possible (e. g. telco mergers) – because we know how to do that Concern that we fail to incorporate important factors that in reality drive deals: • • Investment constraints and dynamics Counterfactual (e. g. structural shifts) Perception we are spending too much time “playing” with these stylised models, while failing to make progress on “bigger” industry issues that may motivate deals 16 September 2016 14
What are we really leaving out? True “dynamics” (investments) is the key problem Calibration of investment incentives in a model not practically possible, also given variation in the data we have Big issue in mobile mergers: industry claimed strong increasing returns to scale deter investment in new technologies when the market remains fragmented: “Need to consolidate to make adequate returns to invest in [4 G/5 G/fibre rollout…]” Trouble has been finding robust evidence to support this • Does investment depend on retained profits? • Evidence of causal relationship “concentration/investments”? • Is progress towards 4 G/5 G further along with more concentration? Or does competition drive investment? Academic studies neutral/sceptical. Collective industry failure to show “need for consolidation” to the required standard. 16 September 2016 15
What are we really leaving out? (2) “Bigger picture” counterfactual analysis is another problem e. g. financial constraints to investment e. g. structural shifts in comparative advantages see EU chemical industry post 2008 hit by production shifting to the Far East, reflecting loss of competitive advantage in inputs and transformation costs. Local capacity redundant, needs shakeout. Entry/exit literature shows industry restructuring/exit of firms takes place mostly through acquisitions. => Seen as “industrial policy”, left to Introduction to the Form CO, the lobbyists and never really getting traction in the competition analysis But these issues are a big part of the motivation for the deal. Only difficult to integrate them in the tools we have. . Need much better integration / understanding of e. g. corporate finance into the analysis 16 September 2016 16
Conclusions The *bar* is not particularly *higher* generally Economic analysis ever more central and technically demanding – focus on price effects (we “know” how to do that, and much better than not doing it…) but there are some important “black spots” Telco mergers? Special circumstances (UK, DK) and growing perception / evidence that earlier remedies (Austria, Ireland, Germany) were a fudge and not likely to work – hence “ 4 -to-3 -back-to-4”… Remedies are quite involved and increasingly prescriptive… but do they work? Need more retrospective evidence! 16 September 2016 17
- Slides: 17