We study firm-level adoption of packaged software products across 3,891 sites in the United Kingdom over four years (2000)(2001)(2002)(2003). We treat the entire bundle of software in a firm as its software product and divide the bundle into related, but distinct subsystems. Diffusion is studied across those subsystems. We introduce three factors that may affect the adoption decision: product complexity, architectural linkage and competency destroying/enhancing scale. We find that firms

more »

... g their competency are more likely to adopt new technologies, while there is a negative relationship between architectural linkage and product adoption. We also find that changes in the complexity of core subsystems hinder switching. * This work is part of the project "Explaining Productivity and Growth in Europe, America and Asia" funded by the Anglo-German Foundation, whose support is gratefully acknowledged. We would also like to thank participants from the "The knowledge-based society workshop" (sponsored by the Research Council of Norway 2006Norway -2009) ) for their comments on an earlier version of this paper. 1 We use the term 'technology' as a catch-all for all goods and services. 2 Geroski [Ge00], Stoneman [St02], Nelson et al. [NPS04] and Rogers [Ro95] review the different models of technology diffusion. Karshenas and Stoneman [KS93], Zettelmeyer and Stoneman [ZS93] and Grajek and Kretschmer [GK06] provide comparative analyses of the different models. 3 The range of products studied is enormous and we will not attempt a review of the existing studies.