Running Neck and Neck

Phenom. Quad-core. FASN8. These may sound like a new car model, the type of engine under its hood, and a vanity license plate for it respectively, but in reality they are all microprocessors recently introduced by major chipmakers. More generally, they are moves in one of the most competitive current product races in the technology industry: the ongoing battle to develop the fastest processor. Phenom, according to a recent Financial Times article, is Advanced Micro Device’s (AMD) latest processor family, developed to compete with Intel’s quad-core processors—which are essentially four processors tied together. AMD has also upped the ante with an eight-core processor, FASN8 (pronounced “fascinate”), in a bid for further leadership. Of course Intel is not idling in the face of these moves. The semiconductor stalwart’s new 45nm (one nm, or nanometer, is one millionth of a meter) chip is expected to leapfrog AMD’s offerings, only the latest in a series of lead shifts between the two firms: in 2003 AMD released a 64-bit processor that wrested the top position from Intel; soon afterward the latter recaptured the lead with its Duo processors.

The dynamics of a sector caught up in research and development races such as these have intrigued economists and other business scholars for years. As such, the technology industry has been particularly fertile ground for their studies and theories. Among the phenomena they try to model and predict is how some sub-sectors, such as operating systems, evolve long-term dominance. For example, Microsoft Windows has been the leading system for more than a decade, improving every few years and keeping rivals far behind. Meanwhile, others such as microprocessors bear witness to almost-quarterly lead changes.

Professor Johannes Hörner of the Kellogg School of Management’s Department of Managerial Economics and Decision Sciences considers exactly such research and development races in a recent paper in the Review of Economics Studies. Specifically, Hörner derives and presents the conditions under which different competition scenarios and outcomes are more likely to occur. For example, he found that when firms are either very impatient or overly patient, their sector or sub-sector is more likely to become a monopoly with one dominant firm, as exemplified by Microsoft’s major lead in the operating systems space. In other cases, there is perpetual leapfrogging, as seen between Intel and AMD.

An interesting and counterintuitive finding by Hörner is that firms do not necessarily invest the most when they are truly tied in a specific race. In fact, a firm will most likely invest more in a given area when it lags just slightly behind a rival, motivated by the fear of falling too far behind and never being able to catch up. In this scenario the same is true for the leader, who may recognize that a bit more investment could mean leaving rivals so far behind that they cease to be threats.

According to Hörner, his interest in such competition scenarios stems partly from the firm-reputation issues he studied for his doctoral dissertation; a paper from his dissertation examines the value of reputation in the context of perfect competition. Hörner was generally interested in the strategic management of assets, whether intangible ones such as reputation or tangible holdings such as technological advances or patents. With his current study, Hörner’s ambition is to develop a rich framework that encompasses several different kinds of industry outcomes in the context of R&D races, including the “absorbing state,” with one company having an effective monopoly, the case of perpetual leapfrogging, and others.

“The previous literature was modeling such races as either the first competitor to run 100 meters, or the competitor to run the biggest distance in the first ten seconds of the race,” Hörner says. By taking this approach, these earlier models of development races envisioned them as games in which the payoffs are distributed at the end of the race. “Instead of modeling a game with no flow payoffs and an end,” Hörner adds, “I model it with flow payoffs and no end in sight.” This no-finish-line approach is clearly superior for describing many of these real-world races, including the one between Intel and AMD. Both firms naturally care more about current profits and the sum of future discounted profits than about which of them will be first to introduce a 16-core chip. While the hard-finish-line approach could be more relevant to pharmaceutical firms’ races to patent specific medications, even in that industry much of the competition and intellectual property is aimed at making incremental progress, versus being the first player to introduce a new product.

In line with his conceptualization, Hörner’s model places two racing firms on the same line, constantly observing each other’s position. There is an infinite number of periods of observation, but for simplicity in each period only the firm that is ahead is viewed as making a profit, regardless of the actual distance of its lead. This simple model reflects an interesting behavior pattern previewed earlier: when firms are very close together on the line, they do not have as much motivation to compete as when they are a bit farther apart but still within reach of each other. That is, the lagging firm seeks to catch up and make sure that it does not fall so far behind that the race is effectively over, while the leader wishes to increase its distance from all followers. On the other hand, when the firms are far apart, neither is likely to invest much in development because the leader knows it will continue to derive profits safely and the follower knows that it will likely never be able to catch up.

These insights, seemingly intuitive in the context of Hörner’s explanations and proofs, were absent from the previous literature. As Hörner and others in this intriguing area continue to refine their models of research and development races, one thing is certain: there will no dearth of product development competitions for them to observe, especially in the tech sector.

Further Reading

Nuttall, Chris (2007). “AMD Launches Chips to ‘Leapfrog’ Intel,” Financial Times, May 14.