The Spatial Diffusion of Technology
Executive Summary — Technology disparities are critical for explaining cross-country differences in per capita income. Despite being non-rival in nature and involving no direct transport costs, technology diffuses slowly both across and within countries. Even when a technology has arrived in a country, it takes years and even decades before it has diffused to the point of having a significant impact on productivity. Why does technology diffuse slowly? How do we explain cross-country differences in its speed of diffusion? In this paper, the authors study the diffusion over time and space of 20 major technologies in 161 countries over the last 140 years. The spatial effects they identify for technologies vanish over time. For most technologies, this implies that the effect of geography is initially strong, decays over time, and eventually disappears. This is the first paper to document these patterns in adoption rates for a large number of technologies and countries. Estimates provided of structural parameters can be used to inform spatial theories of growth. Key concepts include:
- This paper uses a new data set of direct measures of technology to study technology diffusion across time and space.
- Technology diffuses slower to locations that are farther away from adoption leaders.
- Interactions are more frequent for more recent technologies.
- Understanding technology diffusion over space is crucial to understand the speed of technology diffusion.
- Evidence on the significance of the spatial and temporal links in technology adoption could prove helpful to stimulate future research in these areas.
We empirically study technology diffusion across countries and over time. We find significant evidence that technology diffuses slower to locations that are farther away from adoption leaders. This effect is stronger across rich countries and also when measuring distance along the south-north dimension. A simple theory of human interactions can account for these empirical findings. The theory suggests that the effect of distance should vanish over time, a hypothesis that we confirm in the data and that distinguishes technology from other flows like goods or investments. We then structurally estimate the model. The parameter governing the frequency of interactions is larger for newer and network-based technologies, and for the median technology, the frequency of interactions decays by 73% every 1000 kms. Overall, we document the significant role that geography plays in determining technology diffusion across countries.