1. Economics RBB LONDONADRIAN MAJUMDAR Foreclosure, Rollback Rebates & Tomra 1 May 2007 adrian.majumdar@rbbecon.com

2. Economics RBB 1 MAY 2007 2 ADRIAN MAJUMDAR LONDON What is “anticompetitive” foreclosure? A practice that raises barriers to entry and/or expansion, where entry and/or expansion would otherwise have occurred and materially benefited final consumers NB: a practice that “excludes” a rival does not necessarily harm consumers (e.g. when Domco lowers its price, all rivals suffer lost sales but consumers may well benefit) In theoretical models, the story of consumer harm is typically “dynamic”. By keeping the rival smaller than it otherwise would have been, the rival is denied economies of scale. As such, the rival is less effective a constraint on Domco.

3. Economics RBB 1 MAY 2007 3 ADRIAN MAJUMDAR LONDON Foreclosure – Initial Screens Dominant Supplier Final consumers Rival supplier ‘loyal’ buyers Other buyers? 1: dominance? What degree? 2: alternative routes to market to become a viable competitor? Are ‘loyal’ buyers the gateway to the market? Is self supply viable? (No entry barriers in downstream market) Is direct supply viable? How great are scale economies? 3: how does discount scheme affect incentives of buyers? What is the reward? How easy to meet target? Natural growth? Greater promotional effort? Direct substitution of Domco for rival?

4. Economics RBB 1 MAY 2007 4 ADRIAN MAJUMDAR LONDON Rollback: once you hit the target, the lower price relates to ALL units purchased (also called ‘retroactive’ or ‘back to unit one’) Incremental: once you hit the target, discount applies to those units purchased at or above the target Incremental and rollback rebates (1)

5. Economics RBB 1 MAY 2007 5 ADRIAN MAJUMDAR LONDON Incremental and rollback rebates (2) Units purchased target List price Discounted price target Unit price List price Discounted price extra rebate with rollback when hit target Total needs of “representative” buyer AMC “Appropriate measure of cost” (AMC)

6. Economics RBB 1 MAY 2007 6 ADRIAN MAJUMDAR LONDON Implied profit with rollback rebates Units purchased target Unit price AMC Total needs “Break even” “Profit sacrifice zone”: If Domco can sell a lot of units at the list price, it will throw away (sacrifice) profit by rolling back the discount to all units “Profit increment zone”: But, if Domco could not sell many units at the list price, inducing the buyer to reach the target is profitable. “profit sacrifice zone” “profit increment zone”

7. Economics RBB 1 MAY 2007 7 ADRIAN MAJUMDAR LONDON What should an equally efficient rival do? Units purchased target Unit price AMC “stay put” “expand a bit” “expand a lot” NB – we measure the rival’s sales from right to left

8. Economics RBB 1 MAY 2007 8 ADRIAN MAJUMDAR LONDON Which strategy is most profitable? “profit sacrifice zone” “profit increment zone” “stay put” “stay put” is more profitable than “expand a bit” “expand a lot” is more profitable than “stay put” Domco sells at its target Domco’s sales fall below its target Notes: Blue arrow measures Domco’s sales Yellow arrow measures Rival’s sales

9. Economics RBB 1 MAY 2007 9 ADRIAN MAJUMDAR LONDON Implication of a “captive base” Domco has a captive base of sales (e.g. “must- have” good, switching costs faced by buyer) Notes: Blue arrow measures Domco’s sales Yellow arrow measures Rival’s sales “profit sacrifice zone” “profit increment zone” “stay put” is more profitable “expand a lot” is more profitable Does the captive base extend into the “profit sacrifice zone”?

10. Economics RBB 1 MAY 2007 10 ADRIAN MAJUMDAR LONDON The EC Commission views rollback rebates as a form of “fidelity rebate” just because this form of discount can reward a buyer for sourcing a high share of its needs from Domco (e.g. where the target is set close to the buyer’s total needs) The EC Commission says that because any rollback rebate involves a “profit sacrifice zone” (or “suction effect”), it MUST raise barriers to growth (i.e. it must incentivise the rival to be “stay put”) But the EC Commission failed to acknowledge that: After a point, it becomes profitable for the rival to “expand a lot” – so it is important to identify whether other barriers exist (e.g. that give rise to Domco’s captive base) There may be sufficient sales not affected by the target to allow the rival to attain an efficient scale (i.e. is the target a relatively small share of the buyer’s total needs). If the (combination of many) targets in question do not deny the rival an efficient scale then harmful foreclosure is unlikely. Tomra

11. Economics RBB 1 MAY 2007 11 ADRIAN MAJUMDAR LONDON Conclusion / Questions for discussion Is foreclosure feasible on the market? – Gateway buyers affected by Domco’s schemes? – Rival cannot reach efficient scale through alternative routes to market or by product differentiation? Is the rival denied an efficient scale? – Domco sets target close to total needs of gateway buyers? – Domco has a large captive base with these buyers? Are rollback rebates indispensable to achieving certain efficiencies? Fact based story of harm… what standard of proof required? – Does the rollback scheme accentuate other barriers to growth? – Is the rival denied an efficient scale of production? – How does this ultimately harm consumers? – What standard of proof: “capable” vs “likely” vs “very likely”? – What is the risk of precedent that deters pro-competitive behaviour?