Spillover Effects from Wiring Schools with Broadband: Implications for Universal Service Policy

Abstract:     
Debates continue about whether broadband Internet access should be made a part of universal service policy. While both the US and the EU have not explicitly adopted such a policy, other countries, particularly developing nations hoping to leapfrog the developed world, have done so. Further, some states in the US and countries in the EU, have adopted strategies to ensure widespread broadband availability and affordability even in the absence of formal mandates. Typically, such strategies pursue specific projects that promote the local adoption of telecommunications services and its spillover to society at large. One such form is wiring schools with broadband. For several reasons we suspect that connecting schools via broadband contributes to increased overall ICTs usage. By equipping its central offices to provide access to schools, a provider paves the way to connect households and firms in the neighborhood. Then, as youngsters enjoy the broadband experience at school, they are likely to request that their parents install broadband at home, thus exposing older generations to new ICTs services.

These technical and social factors can be expected to complement each other to increase the potential positive externality of providing Internet access in schools. To establish the existence and to measure the magnitude of such externalities we collected data on the Internet usage in schools and by society at large in a given country between January 2006 and December 2008. This country had connected all schools to the Internet through DLS (1 Mbit/s or more) by January 2006. Usage in schools is measured by the amount of traffic through the DSL router at the school premises. Usage by population at large is measured by the amount of traffic through a carrier’s Central Offices (CO). The carrier sampled is the one that provides the DSL connection to all schools in the country who, coincidentally, has a very significant share of the broadband market in the country. We complement these data with yearly information on population density and a number of education related variables such as basic education rate, drop out rate and illiteracy rate. We proceed to regress the current levels of traffic in the carrier’s COs on past school’s traffic and past CO traffic to find that the former is highly statistically significant.

All else being equal, our results show that doubling Internet traffic in schools increases the CO traffic by roughly 5% in the subsequent year. These results are robust across a number of different models, namely using aggregated school traffic per CO, average traffic per school to the closest CO and cumulative school traffic in two previous years. We control for endogeneity between CO traffic and school traffic by using an instrumental variable that measures the quality of the broadband connection at schools, computed using the number of computers connected to the Internet per school and the distance between schools and COs. This instrument is a good predictor of school’s traffic, unrelated to CO traffic and proved very efficient. Our result supports our spillover hypothesis whereby students carry the habit of using Internet at school to other Internet access points, most notably home, where they can act as liaison to older family members and friends transferring the specific Internet related training they get in school to people that otherwise would find it hard, and perhaps useless, to obtain. This result has also deep implications for Universal Service Policy (USP) because it shows how, in the absence of such a formal policy, specific USP-like projects that fully engage local communities with telecommunication operators through the local Government can contribute to a more widespread effective adoption of broadband by society at large.

Number of Pages in PDF File: 18

Accepted Paper Series