Dynamics of Quality Improvement in US Broadband Networks: An Exploratory Study
University of Texas at Austin - Lyndon B. Johnson School of Public Affairs
August 29, 2012
This very preliminary paper is concerned with measurement of broadband quality, making use of a unique, rich new data source that allows us to directly measure the evolution of many dimensions of broadband quality over the period 2011-2012. Its primary focus is on measuring broadband speed (bandwidth), widely perceived as the most important dimension of broadband quality. My principal conclusion is that average download speed delivered by U.S. ISPs, within speed tiers, increased at an average annualized rate of 5 to 7 percent over the spring 2011-spring 2012 period.
This quality increase captures only the effects of improvements in delivered speed within speed tiers, and excludes upgrades to speed tiers for customers on existing service plans that may have been given gratis to existing customers, as well as purchase of new and more expensive plans by customers opting to switch to higher speed plans. For the purposes of quantifying quality improvement relevant to calculation of a quality-adjusted broadband price index, then, this would be a lower bound on the extent to which faster download speeds were a characteristic tracking improved quality in existing broadband plans.
Taking this quality improvement into account seems likely to result in rates of quality-adjusted price decline significantly greater than previous estimates of changes in quality-adjusted U.S. broadband prices. Using as our proxy for quality-adjusted price decline the nominal service price divided by download speed (a measure popularized by the OECD and others), a flat nominal price divided by a five to 7 percent faster speed results in a 5 to 7 percent annual rate of quality-adjusted price decline. While unimpressive when juxtaposed with the historical record in semiconductors or computers, this is still double or triple the rate suggested by earlier studies.
In addition, the analysis has shown that variation in speed from hour to hour within a single day is roughly 5-6 times greater than variation from day of the week to day of the week, at a given hour. Several state holidays have a small but highly statistically significant negative impact on net speed. Weekends, and not weekdays, seem to be the new “prime time” for internet speed slowdowns. This may ratify the claim that the internet has evolved from being primarily a platform for scientists, engineers, and techies, to an intra-business communication and networking platform, to an infrastructure that today, generates traffic whose ebbs and flows are now dominated by consumers.
Number of Pages in PDF File: 25
Keywords: broadband, internet, quality measurement, price measurement, broadband policy
JEL Classification: C43, L63, L86, O30, O31, O38, O47
Date posted: March 30, 2012 ; Last revised: September 23, 2012