Abstract
In this article, I explore the interaction effects of social networks and local spillovers in the diffusion of computers using the panel data from 47 Japanese prefectures for the years 1988–2000. Controlling for unobserved prefecture-specific fixed effects and an endogeneity bias of the lagged dependent variable (the computer possession rate), I find that people are more likely to own computers in areas where the possession rate of computers is higher and where social networks are more strongly connected. Further results suggest a robust complementarity between existing computer owners and the social network. Overall, the empirical study provides evidence that the people learn from neighbours who own computers via a social network enhancing spillovers and reducing transaction costs. That is, the social network reduces the cost of being acquainted with experienced computer owners and so being able to learn from them, thereby attracting new adopters.
Notes
1 There are some empirical studies. See, e.g. Irwin and Klenow (Citation1994) and Goolsbee and Klenow (Citation2002).
2 Social network reduces various kinds of transaction costs such as the job search costs. See, Holzer (Citation1988).
3 Empirical study reveals that density also increases productivity (Ciccone and Hall, Citation1996).
4 We use nonlogarithmic HOUS and INCOM.