Executive Summary
A substantial body of theoretical literature indicates that network effects may hinder the entry of higher-quality products into markets in which network effects are important. However, Tellis, Niraj, and Yin (2009) provide compelling evidence that, in general, higher-quality offerings win out in software markets after a short time lag. Because software markets are commonly considered susceptible to network effects, this finding provides important empirical evidence against the hypothesis that network effects impede entry. Because the authors obtain their results across a large number of product categories and because their analysis holds up across various methods, their evidence that high quality trumps network effects is impressive.
However, before concluding that higher-quality products overcome network effects, certain limitations of their work should be noted. In particular, Tellis, Niraj, and Yin do not consider products that were not developed or introduced because network effects were judged by managers to be too difficult to overcome or because of lack of access to ways of overcoming these effects, such as installing new software on original equipment. Although there is no obvious way to address these considerations with the data at hand, they create a potential sample selection bias.
The nature and extent of the network effects present in the markets that Tellis, Niraj, and Yin study might also be questioned. The authors define these effects as “the increase in a consumer’s utility from a product when the number of other users of that product increases.” Consistent with this definition, they find that market shares tend to increase with the cumulative share of a given product, a measure of network size. Although this finding is consistent with utility increasing with network size, other explanations for the finding are also plausible. For example, a positive relationship between demand and network size could result from diffusion due to word-of-mouth communication or to bundling of software into new computers. In summary, there are alternative explanations for the evidence about network effects presented in the article.
A potential source of network effects is that developers may not develop programs to use with a given piece of software without a critical mass of users of that software. Thus, network effects might be expected to be strongest for operating systems. Among operating systems, Windows and its predecessor, DOS, have been dominant for more than 25 years, despite not being demonstrably superior to alternatives, such and Macintosh and Linux. It is possible that availability of programs that run on Windows and compatibility with other Windows computers has played a role in this. This might be regarded as a counterexample to Tellis, Niraj, and Yin’s finding that quality dominates network effects.
There is a need to further examine whether network effects really are important for the different types of software. When network effects are important, there is a need to understand whether these create a barrier to developing new products that may preclude some suppliers from competing in these markets.
Biography
Brian T. Ratchford is Charles and Nancy Davidson Professor of Marketing at the University of Texas at Dallas. He has an MBA and a PhD from the University of Rochester. His research interests are in economics applied to the study of consumer behavior, information economics, marketing productivity, and the Internet as a marketing institution. He was editor of Marketing Science from 1998 to 2002; is currently an associate editor of Journal of Consumer Research; and is currently on the editorial review boards of Journal of Marketing Research, Journal of Marketing, Journal of Retailing, Journal of Interactive Marketing, and Journal of Service Research.
J Marketing Research, Volume 46, Number 2, April 2009
View Table of Contents.