Global competition Law centre Lunch talk, 4 February 2005 Oracle/PeopleSoft The arguments for and against Claes Bengtsson Member of Chief Economist Team, DG Competition Disclaimer: the views expressed in this presentation are those of the presenter and are not necessarily those of the Chief Economist, DG Competition, the Commissioner or the European Commission.
Background facts Three large competitors: Oracle, PeopleSoft and SAP Many small competitors with less complete product range Different pillars of software, often sold in bundles (FMS, HR, CRM and SCM) Bidding market, sophisticated buyers Solutions tailored to particular needs of customer Vertical component to competition
This case is different “We don’t plan to integrate the two companies…There is no business integration risk at all. Our intention is we’re not going to actively sell the PeopleSoft products anymore” Larry Ellison, CEO of Oracle
The databases IBMOracleMircroSoft OracleSAP PeopleSoft Databases Software
How to model the bidding process? Individual bidding process, but often a “best and final” offer The main part of the costs are sunk before the bidding contest Each bidding contest is unique and the fit of each of the proposed solutions is uncertain Prices do not reflect costs Buyer does not commit to an objective and transparent valuation procedure
Uncertainty about what the software is worth “…we're going to try to do our best to learn as much as we possibly can, and if they share with you, they'll share with us something, but they're negotiating with us. So they're not that motivated to tell us everything, and to tell us everything exactly right.” “…I have been here, and I've noticed that none of the customers wanted to actually share their actual TCO and internal valuation numbers because they actually said they didn't want the vendors to find out.” Saffra Catz, Oracle
Model outline The buyer receives a sealed bid from each of the three bidders p Oracle, p SAP, p PS The value of each type of software is privately known to the buyer: u Oracle ~ N(μ Oracle,σ Oracle ) Oracle wins iff (u Oracle - p Oracle )> (u SAP - p SAP ) and; (u Oracle - p Oracle )> (u PS - p PS ) Calibrate –μ to fit market shares –σ to fit probability of sale
Symmetric simulation Symmetric caseQuality=1 s.dev.=0,10,20,30,40,50,60,81 Prices Post merger0,180,350,520,640,720,790,931,06 Pre merger0,120,240,350,470,580,670,840,99 Effect50,0% 48,0%35,5%23,8%16,8%10,3%7,6% Probability of sale Post merger100,00% 99,68%96,63%91,80%86,89%0,79%72,38% Pre merger100,00% 99,92%99,19%97,46%0,93%87,83% Effect0,0% -0,3%-3,3%-7,5%-10,8%-15,1%-17,6% Consumer surplus Post merger0,880,760,640,590,58 0,600,63 Pre merger0,970,930,900,870,84 0,860,90 Effect-9,0%-18,7%-28,3%-31,9%-31,5%-30,6%-29,7%-29,4%
Fact that fits the model A bidder with a high quality product is likely to ask a higher price than a low quality bidder. The customer does not always pick the highest quality package. Sometimes the difference in price is high enough for the customer to pick a lower quality solution. On average, the supplier with the highest quality offer will get the highest market share and the highest profit. Differences in prices broadly in line with industry reports
Other factors Bidding data –Did not find any difference in bidding behavior when Oracle bid against SAP or PeopleSoft compared to other bids –Did not find any difference in bidding contest with many bidders compared to few US district court ruled in favor of Oracle –In particular, found against DoJ on market definition Uncertainty about the coverage by the old test Clearance