When a merger kills competition

November 8th, 2009

In the next days the European Commission will have to take a position on the merger between Oracle and Sun Microsystems. As it always happens in front of mergers between such large corporations, a deep antitrust evaluation is necessary to verify whether the deal can have anti-competitive effects and ultimately hurt current or future consumers through an increase in prices or a reduction in innovative efforts.

At first glance, a merger between a company that is mainly focused on software (Oracle) with one focused only in part on software and to a large extent on hardware (Sun) would not appear to raise antitrust concerns. However, a closer look at the interests of the two companies in the area of software can provide a different impression, especially if we recall that on a number of occasions Oracle has engaged in acquisitions aimed at substantial business change that led to higher prices (for instance in the case of the BEA Web Logic products) or discontinued production (for instance in the recent case of the Virtual Iron products). Moreover, while Oracle is focused on commercial software, most of the software by Sun is open source, as in the case of Solaris, the most popular operating system derived from enterprise Unix, and MySQL, the most popular database for websites: one may wonder what will be the destiny of these technologies and of their current users, and what will be the impact on the large European open source sector.

If the merger goes through Oracle will become the dominant company as a single supplier of enterprise hardware and software. In this unique position, it may try to limit competition, reduce consumer choice and increase prices. The problem is crucial in the market for database management systems, whose leader is Oracle if we look at revenues, and Sun’s My SQL if we look at units. The reason for this difference relies on the different business models: while Oracle provides a proprietary database and gains from  both licenses and maintenance fees, Sun freely distributes its open source product and gains only from maintenance and support fees. Together, Oracle and Sun account for the large majority of database sales and about 90 % of the Linux/Unix segment (which is almost half of the total). The other two relevant competitors in this market are IBM and Microsoft, followed at distance by Teradata (focused on data warehousing), Sybase and a fringe of open source products unable to provide valuable alternatives (as Ingres and PostgreSQL).

In the global database market, which is worth more than 20 billion dollars, Oracle and Sun compete directly, with the latter traditionally ahead as a web server database and rapidly growing as a business database (for both SMEs and large companies) and as an embedded database (to be used within other products). Thanks to many recent improvements, Sun has made available by far the best open source alternative, which has forced Oracle to invest heavily in R&D to maintain its lead. This was a sort of “escape competition effect” due to the entry pressure exerted by the best fruits of the open source community. Moreover, the free distribution of MySQL has created a binding price constraint on Oracle, which has been unable to perpetuate a high mark up policy and has been forced to reduce its prices to match the low “total cost of ownership” associated with the open source solution (characterized by free distribution plus cheap maintenance). This outcome is reminiscent of the behaviour of a market leader facing endogenous entry of competitors associated with the dominant firm theory, and has guaranteed an efficient evolution of the market until now. One may compare these competitive effects to those exerted by Linux and the competitive pressure of the open source distributors of operating systems on the innovative and pricing strategies of Microsoft for Windows.

For the innovative and competitive pressure to be effective in the database market, Oracle should keep being constrained by a fringe of rivals and by the most efficient one in particular, something that is trying to avoid through a merger with the latter (a sort of “escape competition merger”). No other company could replace (“fork”) the role of MySQL in the short run, because Oracle, as the new copyright owner of its source code, could adopt the dual license, excluding commercial exploitation by other firms and at the same time enjoying any enhancement made by others under the open source GPL. Since the existing database vendors have been unable for years to became a substantial threat, this simple merger would allow Oracle to get rid of its main competitive stimulus (and possibly to find a method to deal with similar threats in the future), and therefore to soften competition. The consequence for the database market would not be that different from what would happen in the market for operating systems if Microsoft was allowed to be in control of Linux.

After the merger, absent any competition between Oracle database and MySQL, innovative efforts for the latter will probably slow down, turning MySQL into a niche product, locking in its customers and re-directing all toward its proprietary database and its applications. Moreover, this would allow Oracle to maintain or increase mark ups for its own products. Of course, followers as IBM and Microsoft may gain from a softer competition, but the largest gain would be for the merging firm (not by chance IBM tried to merger with Sun before Oracle) and the largest loss will be for the customers.

Last but not least, post-merger Oracle would control also the celebrated Java technology, which is a widely used standard for application-hosting used on over 6 billion devices worldwide (way more than Windows), including mobile phones and server platforms. At that point competition for Java-based application servers and for the wider Java middleware market (accounting for about $ 5 billion) would be limited to two major companies, Oracle and IBM, without any strong competitive pressure. Again, this is the typical situation in which a merger softens competition and allows the duopolists (that, by the way, have been historically close cooperators) to increase margins and profits while hurting the final consumers. Finally, one may wonder whether Oracle will use Sun’s technologies for its own purposes only and will pursue anti-competitive strategies toward other firms. In particular Oracle may have incentives to restrict licensing for its competitors, for instance the European SAP (a leading competitor of Oracle in business software) whose business is delivered on Sun’ server systems, or to increase prices for the other firms, for instance, the many European telecommunications technology firms as Nokia, Symbian, Alcatel, Ericsson or Siemens, all licensing Java.

Of course, all of these critical points can be the subject of antitrust concern only if they lead to consequences that hurt current or future consumers, and not if they end up hurting the competitors per se. However, it seems that there are all the conditions for an intervention aimed at least at limiting the welfare reducing consequences of the Oracle-Sun merger for the database market.

Send us your feedback and comments about this article