Patterns of discovery

From a given directed weighted network of knowledge links between technology fields, the paper develops a multisector dynamic model of incremental innovation and R&D activity in these fields. The model is focused on the equilibrium share distribution of these variables, which is proved to be locally stable, with reference to a simple low dimensional case. Simulation methods suggest that local, and also global, stability extend to any model dimension. It is also shown how different network structures map to different asymptotic share distributions. Using the NBER patents and patent citation data files, the analytical framework is then used to analyse some general features of the pattern of knowledge creation and transfer in the period 1975-1999. From a descriptive viewpoint, the changes in the share distribution of innovation activity predicted by the model match reasonably well the actual changes in the period.