Working Paper: NBER ID: w9979
Authors: Chris Forman; Avi Goldfarb; Shane Greenstein
Abstract: The authors test opposing theories on how urban locations influenced the diffusion of Internet technology. They find evidence that, controlling for industry, participation in the Internet is more likely in rural areas than in urban areas. Nevertheless, talk of the dissolution of cities is premature. Frontier Internet technologies appear more often at establishments in urban areas, even with industry controls. Major urban areas also contain many establishments from information technology-intensive industries, whose presence could reinforce the concentration of frontier Internet technologies in these areas. However, information technology-intensive industries are numerous and widespread. Hence, so is the use of frontier technology.
Keywords: internet adoption; urban density; industry composition
JEL Codes: L63; L86; R0
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| urban density theory does not hold for participation applications (R23) | probability of adopting basic internet applications (L96) |
| urban size or density (R12) | probability of adopting basic internet applications (L96) |
| rural areas may adopt internet participation more readily (R50) | probability of adopting basic internet applications (L96) |
| urban density theory holds for enhancement applications (R11) | likelihood of adopting complex applications (C52) |
| population density (J11) | likelihood of adopting complex applications (C52) |
| IT-intensive industries in urban areas (L86) | variations in adoption rates (J12) |
| urban density theory and industry composition theory interact positively for enhancement applications (R32) | likelihood of adopting complex applications (C52) |
| urban density theory does not hold for participation applications (R23) | widespread use across various locations (R14) |