Fast and furious: the productivity effects of the geography of experienced internet speeds

The discussion about the productivity gains from digital technologies is almost as old as digital technologies themselves. From early futuristic approaches to the various forms of the productivity paradox, there are still open questions regarding if and how the internet can lead to positive economic effects as well as their spatial heterogeneity. The usual caveat to unpack this relationship is data about internet usage that is detailed enough to be linked both to economic outputs and places. We develop a multilevel modelling framework and combine firm-level microdata with novel internet speed microdata illustrating how connectivity has been experienced by end-users. Although it is not possible to directly observe the online activities of individuals, these data allow us to approximate business usage. Additionally, we can distinguish between upload and download speeds, which is important as these channels can support different internet functions (e.g. video streaming vs. synchronous communications). We then observe firms, their productivity, and other firm characteristics and estimate the firm productivity effects of broadband speeds after accounting for spatial effects and platial characteristics, which are also associated with the underpinning digital divides. We find that increases in a place's internet speeds can lead to higher firm productivity whilst the opposite holds for instability of internet speeds. These results vary by sector and firm characteristics. Our results have significant policy implications, highlighting the economic impact of high-speed internet infrastructure planning decisions and presenting policymakers with a clear efficiency vs. equity trade-off.