Diagnosing the Disconnected: Where and Why is Broadband Access Unavailable in the U.S.?

Abstract

This paper constructs a framework for modeling the determinants of broad band availability in the United States, and applies it to a zip code-level database of economic, demographic, geophysical, and policy variables constructed by the author. Statistically significant state effects exist, suggesting that, after controlling for a long list of economic and demographic factors, state-level policies appear to have played a role in accelerating or retarding broadband deployment. I find that geophysical variables are important determinants of differences in broadband availability across zip codes, presumably because of their effects on broadband deployment costs. My results here suggest that omitting these variables from supply side analyses is likely to result in significant estimation biases in empirical econometric work. Geophysical variables are also excellent candidates for use as instruments in future work on broadband demand, where dealing with unmeasured variation in quality of service chosen by consumers is clearly an important problem for researchers, and availability of instrumental variables to deal with this issue has been problematic. Interesting results on the impact of a variety of socioeconomic variables on broadband deployment are available. Income and wealth variables unsurprisingly seem to be among the determinants of lack of broadband availability, but population density’s role is much more problematic. When a full set of explanatory variables is used, the results suggest that absolute market size, not some measure of density relative to physical area, is the key determinant. I note that the 'digital divide' reducing African American use of broadband, cet. par., seems to have vanished. A statistically significant impact of the eRate (Universal Service Fund) program linking schools and libraries was evident in 2005, with a positive impact on reducing areas without broadband service.