Abstract
This paper develops an identification strategy to generate unbiased estimates of Internet usage spillovers using a unique data set of US households. This strategy addresses the effects of variables that affect consumers' decisions but are unobservable to the researcher that lead to biased estimates. The estimator developed here examines changes in household behavior over time and uses county level instrumental variables for county level spillover measures.Multiple potential sources of learning are identified including those from the household's locality, from educational Internet subsidies, and from universities. There is general support for all sources, but the locality and subsidy results are both more robust and larger. These findings directly address the policy relevance of these spillover sources and they have implications for policies to encourage Internet use as well as for identification strategies for the effects of the Internet on behavior.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have