Measures of Spectrum Holdings and Spectrum Concentration among Cellular Carriers

Abstract

Regulators in most countries limit the amount of spectrum that a carrier is allowed to obtain as a way of reducing the risk that rival cellular carriers will be unable to obtain the spectrum they need to compete effectively. For similar reasons, regulators may consider how a proposed merger would affect concentration of spectrum holdings. These policies require appropriate measures of spectrum holdings for a carrier, and of spectrum concentration across carriers. Spectrum holdings of a carrier are typically measured simply by considering the total bandwidth. However, the total bandwidth of a carrier may not be the most useful measure, as this measure does not consider the frequencies of the spectrum held, although frequency greatly affects the cost of building out and operating infrastructure in that spectrum. A promising alternative is to use a frequency-dependent weighting function when quantifying spectrum holdings. Spectrum concentration across carriers is typically measured with the Herfindahl-Hirschman Index (HHI), but analysis in this paper shows that other measures could be more appropriate, and why the high costs of spectrum may yield economies of scale which benefit large carriers. This paper empirically investigates how use of these alternative measures affects the relationship between concentration of market share and the concentration of spectrum holdings using a variety of different measures to quantify spectrum holdings and concentration of spectrum holdings. Some of these measures of spectrum holdings take frequency into account, and others do not. Some of these measures are more appropriate if there is a linear relationship between the amount of spectrum a carrier has and number of customers it serves, and others are not. Greater correlations between spectrum concentration and market share concentration were observed with measures that put more weight on lower frequencies, although the differences are too small to support strong conclusions. The issue deserves further empirical research with more extensive data.

Keywords: spectrum screen, spectrum cap, frequency-dependent weighting function, market concentration