Oracle/PeopleSoft: assessing competitive effects
In the early years of an antitrust system, market definition
and structural analysis – i.e. analysis of competitive effects
based on market shares – usually tend to dominate. But as
the system matures, the structural approach tends to give
way to a more economically sophisticated analysis of
competitive effects. The Oracle/PeopleSoft merger
provides an illustration of this trend, and gives a flavour of
the type of analysis we can expect to see increasingly often
as the European Commission applies the new EC Merger
Regulation (ECMR).
Oracle/PeopleSoft, at over 12 months from notification to
decision, was the longest proceeding in the history of EU
merger control.
1
There are many facets of interest: it was a
true “gap” case as it was decided under the old ECMR, but
fitted neither the traditional single dominance nor collective
dominance (coordinated effects) models.2
Moreover, it
involved an interplay between the EU process and evidence
brought out in simultaneous court proceedings in the US,
where a federal court in San Francisco defeated the DOJ’s
attempt to challenge the merger. It was also characterised
by the collection of a substantial amount of new evidence
(both from the US trial and from Oracle) at a very late stage
in the EU process, facilitated by a 6-month “stop the clock”
period.
But in this memo we concentrate on the competitive effects
analysis that led the Commission to its eventual clearance
decision, which well illustrates the decline of traditional
structural analysis in favour of economics-based analysis of
effects.
Background to the transaction
Oracle is the world’s second largest supplier of software,
after Microsoft. Its traditional strength is in database
products, which are a key part of the infrastructure on which
business software applications run. In recent years, Oracle
has also become a major supplier of enterprise application
software (EAS), the software that is used to automate a
wide variety of business processes, including areas such as
accounting, human resources management, manufacturing
processes, supply chain management and customer
relationship management.
The worldwide leader in EAS supply, by some distance, is
the German company SAP, which is among the foremost
suppliers in all the major areas of EAS. PeopleSoft, the
target of Oracle’s hostile bid, was a pioneer in the
development of human resources (HR) software, but has
expanded into all the major areas of EAS. Overall, the
worldwide EAS industry remains highly fragmented, but the
competition agencies on both sides of the Atlantic feared
that an Oracle/PeopleSoft combination would bring together
two of only three serious players in a segment of the
1 Lexecon Ltd advised Oracle throughout the EU merger review.
2 Many commentators argued that the old ECMR suffered from a “gap” in its
scope of application, being unable to deal with cases that involve neither a
clear market leader (single dominance) nor scope for coordinated behaviour
(collective dominance).
industry, especially in relation to HR and financial
management systems (FMS) software.
“High end” market definition
Both in the US and in the EU, the agencies’ concerns
focussed on the “high end” of HR and FMS supply – defined
in terms of product features and capabilities; and in terms of
the size and complexity of the customer organisation.
3
EAS
transactions are concluded by individual negotiation, and
the agencies’ aim was to identify a subset of transactions
that could be regarded as a single relevant market,
corresponding to the “high end”. This was where the
agencies saw the greatest overlap between the merging
parties, and the least constraint from rival vendors.
The Commission’s approach was based on a general view
that SAP, Oracle, and PeopleSoft are particularly
prominently represented in the largest HR and FMS deals,
involving supply to the biggest customers; and that these
three vendors’ products are generally technically superior to
other vendors’ offerings. In accordance with this view, the
Commission defined a separate relevant market for “highfunction
EAS purchased by large enterprises with complex
needs”. This definition has a customer-related dimension
(“large and complex enterprises” – defined for practical
purposes as organisations with over 10,000 employees
and/or with revenues of at least €1 billion); and also a
product-related dimension (“high function” – notionally
related to the capabilities of the product, but identified for
practical purposes by contracts involving at least €1 million
in net new license revenue).
Not surprisingly, the Commission struggled to put the
concept of the “high function” market into practice – in
particular, it had to acknowledge the difficulty of finding
meaningful criteria to use in order to draw the line between
“high function” deals and others. Precisely the same issues
arose in the US trial, where the judge emphatically rejected
DOJ’s attempt at defining a “high end” market:
the court cannot delineate product boundaries in
multi-billion dollar merger suits based upon the
mere notion that there is “something different” about
the merging products and all others, especially
when that “something different” cannot be
expressed in terms to make a judgment of the court
have meaning. More is required.
4
While the Commission’s basic approach of defining a
separate “high function” market did not change materially
throughout its investigation, its conclusion on who competes
in that market did change crucially. The Commission’s
hypothesis at the time of its Statement of Objections (SO) in
March 2004 was that only SAP, Oracle and PeopleSoft
3
The only area where the Commission and DOJ disagreed materially was on
geographic market definition: DOJ argued for a separate North American
market (which was rejected by the US court), while the Commission defined
the market as global.
4
Final Order of Judge Vaughn Walker in USA v. Oracle (Case C 04-0807).
London: 99 Bishopsgate, London EC2M 3XD
Tel +44 (0)20 7664 3700, Fax +44 (0)20 7664 3998
Brussels: Tel +32 (0)2 627 1400, Fax +32 (0)2 627 1444
Amsterdam: Tel +31 (0)20 521 9326, Fax +31 (0)20 521 9333
Paris: Tel +33 (0)1 70 38 52 78, Fax +33 (0)1 34 29 67 40
For more information please contact:
Matthias Pflanz, Vice President (mpflanz@crai.com)
www.crai.com
were credible suppliers in the market. By contrast, the final
decision (October 2004) acknowledges that a number of
other vendors – including Microsoft, which is one of the
leading suppliers in the EAS industry overall, but less
focussed on the largest customers – should be regarded as
credible competitors for at least some customers in the
“high function” market.
Unilateral effects assessment
Both the Commission’s and the DOJ’s substantive
assessment of the effects of the merger focussed on
unilateral (or “non-coordinated”) effects – that is, on effects
that would arise directly as a consequence of lessened
competition between the merging parties, without any
coordinated response from other rivals remaining in the
market. A coordinated effects theory was unsustainable in
the light of industry circumstances,5
while a traditional
single-firm dominance finding was similarly hard to justify in
the light of the very strong position of SAP in the relevant
markets, however defined. From an economic perspective,
a unilateral effects analysis was clearly the most
appropriate way of assessing the transaction, given the
reality of competition in EAS supply.
In a unilateral effects analysis, the key questions are how
directly the merging parties’ products compete, and how
closely they would be constrained post-merger by nonmerging
rivals. This determines the extent of any loss of
competition resulting from the merger. The more closely
the parties compete before the merger, and the less
effectively others are able to constrain them post-merger,
the more likely it is that the merger will reduce competition.
The investigation of these questions is well suited to
empirical analysis. In the case of a market where prices are
individually negotiated, as in the present case, the key
empirical evidence takes the form of bid data, showing who
is actually competing in individual bids, and what effect the
presence of particular competitors has on the terms offered
by the merging parties. Formal market definition is not a
central issue when there are data that allow a direct
assessment of the closeness of competitive constraint
between the merging parties and others in the market,
because market shares become largely irrelevant in this
situation.
Economic analysis of merger effects
At the time of the SO, the Commission had only limited
evidence on unilateral effects available to it. But
nevertheless, it concluded provisionally that there was a
serious prospect of anti-competitive unilateral effects. Its
evidence consisted of data on a small selection of individual
bids, supposedly showing close competition between
Oracle and PeopleSoft, as well as two more formal
economic analyses:
• an econometric analysis provided by PeopleSoft,
claiming that there was a link between the number
of bidders and the discount offered in a set of
PeopleSoft’s bids; and
• a theoretical merger simulation model, calibrated
only on estimated market shares, predicting that a
reduction from three to two bidders would harm
customers by raising prices and reducing product
variety.
5
Coordination is implausible in the EAS industry owing to a combination of
factors such as highly differentiated products and heterogeneous customer
preferences, lack of transparency regarding prices and other terms offered,
scope for disruption of any attempted coordination by “fringe” vendors, and
the importance of technological innovation.
Neither of these provided economically robust or convincing
evidence against the merger, and the Commission did not
ultimately rely on either as evidence in its decision.
6
Instead, the Commission’s decision was able to draw on a
substantial amount of new evidence that had not been
available earlier in the procedure – notably from the US trial,
and from the additional bid data gathered by Oracle in
response to the Commission’s request after the oral
hearing. Detailed econometric analysis of these new data
sources formed an important part of the basis for the
eventual clearance decision, despite the concerns
expressed by the Commission earlier in the process.
We used the extensive bid data provided to the Commission
by Oracle, covering hundreds of bids potentially falling into
the “high function” markets, to perform a detailed
econometric analysis that was submitted to the
Commission. Our analysis addressed two key questions:
a) does the number of rival bidders systematically
affect the discounts offered by Oracle?
b) does the identity of the rival bidders systematically
affect the discounts offered by Oracle – in
particular, does the presence of PeopleSoft lead to
especially high discounts?
Thorough econometric analysis allowed both of these
questions to be answered in the negative. The Commission
accepted these conclusions in its decision, and agreed that
the bid data offered no evidence in support of the
hypothesis that a merger between Oracle and PeopleSoft
would eliminate a particularly important and irreplaceable
competitive constraint. While the use of bidding analyses to
assess merger effects is not new,
7
both the scope of the
econometric analysis in the present case, and the weight
given by the Commission to its conclusions, were
remarkable. Econometric bidding studies are now firmly
established as a mainstream, powerful technique in
unilateral effects cases.
Final observations
The EU review of Oracle/PeopleSoft ended with an
unconditional clearance, despite the earlier issuance of an
SO expressing strong concerns. This places it in a small
group of cases where the Commission substantially revised
its assessment between SO and final decision. The basis
for the clearance decision – namely the collection and
analysis of a significant amount of additional data after the
oral hearing – was also highly unusual, as was the interplay
between evidence aired in a US court and the analysis
carried out by the Commission.
The Commission’s willingness to accept new evidence late
in the process, and ultimately reach an unconditional
clearance after an adverse SO, suggests that the new
system of internal checks and balances within the
Commission is working. The central role played by
sophisticated econometric analysis in the final decision is
testament to the importance of the chief economist’s team
in major merger cases. Both of these are welcome
developments.
December 2004
© Charles River Associates (published originally by Lexecon Ltd,
prior to the acquisition of Lexecon by Charles River Associates)
6
We provided the Commission with a detailed critique of the PeopleSoft
econometrics. The Commission’s decision agreed that the PeopleSoft
results were unreliable.
7
We have used econometric bidding analyses to predict competitive effects
in numerous merger cases – for a description of an early example of the use
of such techniques (in the 1997 Boeing/McDonnell Douglas case), see
http://www.lexecon.co.uk/assets/boeing_mcdonnelldouglas.pdf.