How Do Limitations in Spectrum Fungibility Impact Spectrum Trading?

Abstract

Markets in naked spectrum have been discussed since Coase rst suggested market-based spectrum allocation in 1959. While the initial attention was paid to primary markets (i.e., spectrum auctions) researchers and policymakers soon realized the limitations of primary markets in the absence of secondary markets. In time, regulators relaxed policies so that broker-based secondary sales emerged. Broker-based markets lack the price transparency and liquidity of exchange-based markets, so researchers have considered these as well.

Regardless of type, most analyses assume that the spectrum bands being traded are fungible. However, outside of limited circumstances, this assumption is difficult to defend in practice. Data from auctions shows that different prices obtain for different bands, arguably due to different physical properties. Fungibility is even less defensible as one considers paired vs. unpaired spectrum and differences in regulatory rules associated with different bands.

We extend to consider the impact of fungibility limitations on the liquidity of the market for naked spectrum (i.e., spectrum license trading). To do this, we modify the Agent-based Computational Economics (ACE) based SPECTRAD model developed in to allow us to study how deviations in the fungibility assumptions aspects the liquidity of trading markets. In particular, we consider how the coverage differences associated with different frequency bands aspects the substitutability of the bands, which, in turn impacts the liquidity of the underlying markets. Some examples of this include the AWS-1 band and the 1755-1850MHz band that is currently being considered for sharing, or, alternately, the 2300MHz band). We will also consider how differences in quality (i.e., signal-to-noise ratio) aspects the liquidity of spectrum markets. An example of this kind of effect is the 700MHz A block compared with the B block.

While it seems certain that decreases in fungibility will result in decreases in liquidity, it is unclear what the magnitude of those are for particular spectrum trading scenarios. For instance, we will examine at which frequency offset liquidity decreases significantly. Finally, the overall goal of this paper is to (1) understand the role of fungibility limitations on spectrum markets and (2) provide policymakers aiming to craft rules for spectrum markets with a quantitative insight on what promotes liquid markets. We expect that an practical consequence of this work could be the development of more focused spectrum markets rather than a single, general purpose market as has been modeled previously.