The Estimation of a Cusp Model to Describe the Adoption of Word for Windows*

Abstract

This article revisits earlier work in this journal by Paul Herbig (1991) that proposed a catastrophe model of industrial product adoption under certain conditions. Catastrophe models are useful for modeling situations where organizations can exhibit both smooth and abrupt adoption behavior. It extends Herbig's work by focusing on organizations' adoption of new products when network externalities are an important part of the decision process, and it presents an empirical estimation of the model. Network externalities occur when firms do not want to adopt a new innovation or product unless other firms do. The reason is that they do not want to end up with an innovation that ends up not being a standard of some sort. Mistakes of this nature can be costly as the firm must invest twice and loses time relative to competitors who have not made such a mistake. However, when such externalities exist, for example with regard to technological adoptions, then normal diffusion gives way to sudden discontinuous shifts as all firms seemingly act together an move to a new technology. Since, technology is an area where the authors expect network externalities to exist, that is the focus of this article. The specific application is developed from two sets of panel data on the organizational adoptions of Microsoft's (MS) Word for Windows software by organizations that previously were using either Word for DOS or Word for Macintosh (Mac). The theoretical framework for the analysis is based on work in the economics literature on network externalities. However, the organization and new product development catastrophe model comes primarily from Herbig (1991). The article focuses on an area of organizational adoption where relatively little empirical research has been done, namely organizational adoption “for use.” Longitudinal data provided by Techtel Corporation is used to develop the estimations. Results of the empirical analysis are consistent with the theoretical framework suggested in Herbig's article and in those found in economics and catastrophe theory literatures. This lends clear support to the idea that organizations will adopt a bandwagon‐type behavior when network externalities are present. It further suggests that in such markets, the standard S‐shaped diffusion curve is not an appropriate model for examining organizational behavior. From a managerial perspective, it means that buyers and sellers may face nonstandard diffusion curves. Instead of S‐shaped curves, the actual curves have a break or rift where sales end, and there is a sudden shift to a new product that is relatively high very early on. Clearly, for new product development (NPD), it suggest that organizations' “for‐use” purchases may be similar to regular consumers and may change rapidly from one product to another almost instantly, as in the case of the switch from vinyl records to compact discs (CDs). From an old product seller's viewpoint, the market is here today and gone tomorrow, while for the new seller it is a sudden deluge of sales requests. To put it in more everyday terms, sudden changes in adoption behavior are a September 11‐type experience for the market. It is the day the world changes.