In light of the conclusions regarding the market definition, it is not possible for the Commission to uphold its preliminary conclusion that this group of customers will be confronted with a de facto absence of choice after the merger.

In order to assess whether or not the notified concentration would lead to non-coordinated effects, the Commission ran a number of regressions on each of four datasets: one data set from PeopleSoft, one Oracle data set compiled for the European Commission and two of the Oracle datasets from the US Court proceedings. The purpose was to investigate to what extent the competitive situation of a particular bid (measured by the number of final round bidders) had an impact on the discounting offered by the seller in question (i.e. PeopleSoft in PeopleSoft’s dataset and Oracle in Oracle’s datasets).

The Commission generally found that there was a very strong relationship between the size of the deal and the discount offered. Deals that tended to produce very high list prices were overall likely to also attract very large discounts. The Commission also found that in a number of regressions in which this relationship between the deal size and the discount was not properly taken into account, there appeared to be an effect on the competitors on the discounting behaviour. The initial finding mentioned in the Statement of Objections was indeed based on such a regression. Once the size of the deal was taken into account, the number of final bidders did no longer lead to such effects.

A finding, as in this case, that the number and identity of competitors in a given bid appear not to have an effect on a firm’s behaviour does not in itself prove that there will not be harmful effect of the merger on customers. There may be a variety of reasons why such an effect is absent from the bidding data. Based on the information available, it does not appear possible to reach a final conclusion about which of the different reasons is the most pertinent. It is not necessary, though, for the Commission to reach a conclusion on this issue since it is clear that the absence of an appreciable effect of competition on Oracle’s behaviour makes the bidding data unsuitable to rely on as determinative proof of an anticompetitive effect of the merger.

In the Statement of Objections the Commission had based its concerns further on the finding that, in addition to the non-coordinated effects, the two remaining players would be in a position to further soften competition by coordinating their competitive behaviour. The theory of coordination was based on a definition of the market which after the merger would consist of only Oracle and SAP and relied in particular on the symmetrical market shares of a combined Oracle/PeopleSoft and SAP and an alignment of incentives between them due to the fact that SAP is by far the largest reseller of Oracle databases. As parameters for adopting a common policy, the Commission identified the allocation of customers in a duopoly, a reduction in price competition and a common understanding to slow down the addition of further functionalities and the improvement of products.

In light of the conclusions regarding the market definition, it is not possible for the Commission to conclude that the merger would lead to a collective dominant position of a combined Oracle/PeopleSoft and SAP on the basis of coordinated effects.

In a market in which — in addition to Oracle, PeopleSoft and SAP — Lawson, Intentia, IFS, QAD and Microsoft are also present as vendors of high-function FMS and HR applications, it appears difficult to argue that these players would reach a common understanding as regards the parameters outlined above. In particular, an allocation of customers will not be possible due to the larger group of possible vendors of such software. The larger number of vendors also reduces the transparency in the market and would make retaliatory actions more difficult. Furthermore, the players’ market shares are not symmetrical. The vendors other than Oracle, PeopleSoft and SAP are significantly smaller in the relevant markets. They do not have similar structural links as SAP and Oracle in the area of databases. Lawson’s resale of Oracle databases appears to be marginal compared to the sale of Oracle’s databases by SAP, while Microsoft in any case uses its own databases for its ERP products.

Also, the Commission cannot conclude that the merger will lead to a coordination of only a combined Oracle/PeopleSoft and SAP. The other vendors mentioned in the section on market definition — Lawson, Intentia, IFS, QAD and Microsoft — appear to be suitable alternatives as the Commission’s data and the dataset submitted by Oracle after the oral hearing (ultimately on 27 September 2004) show that those vendors have won bids for software in the relevant markets. A coordination of the competitive behaviour of a combined Oracle/PeopleSoft and SAP could therefore not be successfully sustained.