|
Article Excerpt
1. Introduction
Intel and Microsoft are perhaps the best-known pair of complementary firms in the world. More than 80% of the personal computers (PCs) worldwide ship with an Intel microprocessor running Microsoft's Windows operating system. Since 1980, when IBM chose both Intel and Microsoft as the core components of the first IBM PC, Intel and Microsoft have been joined at the hip. In this paper, we tap recently available data revealed by the U.S. Department of Justice to explore the Intel-Microsoft relationship and model the dynamics of competition between complementary players.
Very little research has been published on the alignment of incentives among tight complements. In this article, we demonstrate that natural conflicts emerge over pricing, the timing of investments, and who captures the greatest value at different phases of product generations. In a system of complements, like the PC, the value of the final product depends on how well the different components work together. This, in turn, depends on the firms' investment in complementary R & D. By studying the incentives for complementors to cooperate on R & D and pricing, we explore whether profit maximizing complementors will cooperate fully to make the final product as valuable as possible.
Our interest in the topic was sparked by claims from Intel and Microsoft executives that their "partner" was not doing enough to advance the PC platform. For example, in an interview conducted by the authors on September 10, 2002, Frank Ehrig, former Microsoft relationship manager for Intel, asserted:
Intel is always trying to innovate on hardware platform and thus, always needs software. When software lags, it creates a bottleneck for Intel. Microsoft, on the other hand, wants to serve the installed base of computers in addition to demand for new computers. Therefore, a natural conflict exists between both companies. In addition, the question always remains--Who will get the bigger piece of the pie? The success of one is seen as ultimately taking money away from the other.
Problems between Intel and Microsoft arose when one force was not pushing as hard as the other would like. For instance, Intel's 32-bit microprocessor (the 386) waited 10 years before Microsoft produced a 32-bit operating system (Windows 95) (Casadesus-Masanell et al. 2003b). Intel's growth was dependent on the demand for its newest processors. As the company strove to stay ahead of its competition in terms of CPU performance, Intel management became dependent on complementary software to stimulate demand for new products. Microsoft, on the other hand, was motivated to not only meet the demand for new computers but also wanted to serve the installed base of computers.
The PC industry is an ideal setting to study complementors and their relationships. Following IBM's decision to set up an open standard for its PC in 1980, the microcomputer industry became gradually more horizontal, which led to specialized players increasingly dominating each component layer. In effect, the horizontal industry structure produced a wide array of complementary firms that needed to work together. Undeniably, the two most important components in a PC are the microprocessor and the operating system, with Intel and Microsoft as the leading players in each of these domains. (1)
We approach the question by setting up a dynamic mixed-duopoly model. Intel and Microsoft are both profit maximizers, but the shape of their objective functions differs because their product technologies are inherently different. As a software company, Microsoft derives profit from selling operating systems (OS) with each new PC and from selling applications to the installed base. Intel's profits, however, come from the sale of microprocessors in new PCs. Frank Ehrig's quote above suggests that the installed base is at the root of conflict between Intel and Microsoft. Therefore, the only substantive asymmetry between Intel and Microsoft that we explicitly model is the effect of the installed base on profits: Intel does not profit from the installed base, while Microsoft does. This allows us to see to what extent conflict between Intel and Microsoft can be explained from differences in objective functions implied by technology (software versus hardware).
We find that Intel prefers to charge high prices early in the life of a generation, while Microsoft is better off charging low prices early to increase the installed base. Microsoft then raises prices gradually as the life of the generation approaches its end. Intel charges high prices early because it has no (direct) interest in the installed base and to take advantage of Microsoft's low prices. Contrary to intuition, we find that Intel and Microsoft's net present value of profits coincide. This is surprising because the objective functions and the equilibrium price paths differ. As a result, initial investments in complementary R & D are equal. This result suggests that there is no conflict regarding investments in complementary R & D. The only conflict that the benchmark model identifies (other than the usual free-rider problem) is that Intel sets prices too high and this forces Microsoft to sell its operating system at lower prices than desired.
We then extend the model from one to two consecutive PC generations. The expanded model allows us to shift focus from equilibrium investment size to equilibrium timing of releases of new microprocessors and operating systems. Because the asymmetry in objective functions does not appear to generate conflict regarding investment size, we investigate whether it creates differential incentives regarding the timing of new product releases. We find that Intel wants to release a new generation microprocessor earlier than Microsoft. Clearly, Intel would be better off if Microsoft's actions followed Intel's timing. Microsoft, however, is better off milking the first generation to the last drop and wait longer than Intel to introduce the new generation OS. The switching to a new generation implies that the first generation installed base will begin to fade; Microsoft has more of an incentive to stay longer with the first generation than Intel. We conclude that there is a two-sided asymmetric conflict between Intel and Microsoft. From the point of view of Microsoft, Intel sets prices too high (especially early in the life of a given PC generation) and, from the point of view of Intel, Microsoft releases its new generation operating systems too late. In the final part of the paper, we speculate about possible solutions to both these conflicts.
While this model builds on a specific case, we believe that the implications are more general. One can think of any two firms that produce complementary products, where one firm cares more about the installed base than the other (for example, Nokia, which makes money from new sales of phones, and service providers, who profit more from usage in the installed base). Similar conflict exists between software vendors and PC companies. Most PC original equipment manufacturers (OEMs) need new exciting software that requires customers to upgrade to new PCs, but application providers generally prefer to sell to the larger installed base. Going back in history, Kodak could not entice camera manufacturers to develop cameras for their new film, which led Kodak to forward integrate into cameras themselves (i.e., the lack of compatible incentives between complements was so great, that they had to internalize the innovation). Finally, pricing conflicts among complements are rife in recent times: Apple iTunes, for example, wants very cheap songs (complements) to sell more iPods, but record labels want to raise the price.
The emergence of independent, complementary firms would seem to be expanding. An increasing number of industries has evolved away from vertical integrated to more horizontal structures. These industries range from computers and communications to consumer electronics, automobiles, and health care. In these new horizontal industries, products that used to be designed and manufactured by a single firm are now produced by different players that must coordinate activities, and of course, some of these players are complements to each other. The result of these horizontal structures is that it is becoming more important for firms to create new skills for managing complementary relationships. This requires an understanding of the fundamental economics governing relations between complementors: What determines the relative bargaining power to appropriate value? How can one firm build a dominant position vis a vis complementors? How can a firm make sure that complementors invest enough in complementary R & D? etc....
This paper is related to the literature on co-opetition pioneered by Brandenburger and Nalebuff (1996). In fact, co-opetition presents Intel and Microsoft as one main motivating example on the tension between cooperation and competition that characterizes many business relationships. According to Brandenburger and Nalebuff (1996, p. 4): "The demand for Intel chips increases when Microsoft creates more powerful software. Microsoft software becomes more valuable when Intel produces faster chips. It's mutual success rather than mutual destruction. [...] Business is cooperation when it comes to creating a pie and competition when it comes to dividing it up." To the best of our knowledge, however, there is no previous dynamic mixed-duopoly model analyzing the relationships between complementors by use of non-cooperative game theory. Most of the literature on cooperation versus competition focuses on relationships between competitors or between buyers and sellers. The analysis of complementors is not as well understood.
This paper is organized as follows. Section 2 introduces a simple dynamic mixed-duopoly model. Section 3 extends the analysis to two product generations and studies the investment game played by Intel and Microsoft at the end of the first generation. Section 4 concludes by discussing some managerial implications that can be drawn from the analysis. The online appendix (provided in the e-companion) (2) presents all the proofs.
2. Benchmark Model
We model the relationship between Intel and Microsoft as a two-stage game. In the first stage, the "value creation" stage, the companies choose how much to invest in complementary R & D. In the second stage, the "value capture" stage, Intel and Microsoft set prices to capture as large a share of value as possible. We assume that the second stage has length T [member of] (0, [infinity]] which, for now, is an exogenous parameter. Thus, Intel and Microsoft set prices [p.sub.I](t) and [p.sub.M](t) every instant t from time to T. After the life of the generation is over at time T, Microsoft may still realize some profit for S [member of] [0, [infinity]] more periods. This additional profit originates...
|
