The temporal stability of spatial diffusion channels

Abstract

This paper attempts to answer the question of how stable are spatial channels of innovation diffusion over time. The notion of stability of diffusion channels derives from the concept of spatial autocorrelation and from the assumption that neighboring spatial units exhibit relative similarity in economic development preconditions necessary for innovation adoption. Stability, however, declines over time as the diffusion process approaches saturation. Innovations with similar properties may tend to produce similar diffusion channels, facilitating forecasting of future diffusion of innovations based on knowledge of diffusion of similar ones. A case study is provided through an analysis of the diffusion of automobiles in Ohio using three methods: visual analysis of maps of cross-sections for identification of channels, transitional matrices for analysing the nature of change in adoption situations, and surface analysis to analyse correlation between adoption surfaces in the short and long ranges. The conclusion of these analyses is that over most of the study period the channels are relatively stable in the short-range, with a general tendency of decline in stability over the long-range. A comparison to a study of diffusion of another innovation in the same area supports the assumption that channels may be more universal than innovation-specific.