Hierarchical modeling and prediction of spectrum auction revenue by a posteriori clusters

Abstract

The electromagnetic spectrum is a crucial asset for wireless technology innovation and for the industry's economy. We examine spectrum auction-related databases, construct a hierarchical data matrix, and propose an optimal auction revenue model. The model incorporates statistically derived deterministic and stochastic variables to develop a spectrum valuation methodology for its primary market. Our study reveals that the underlying regulations, bidding behavior, and spectrum demands due to wireless technology advancement are effectively elucidated by the selected explanatory variables and are highly correlated with auction revenues. This paper is the first to introduce a hierarchical modeling technique that incorporates a multilevel data matrix culminated from 3995 licenses of 15 Federal Communications Commission spectrum auctions to enable economic valuation for future spectrum auctions. Furthermore, the results of this study can be applied to evaluate the significance of sunk costs, winner's curses, and associated cost synergies, which are economic implications of spectrum auctions. Our research contributions are twofold. First, the hierarchical auction model can maximize spectrum valuation methodologies, thereby assisting spectrum regulators and the wireless industry. Second, we compare the reproducibility of hierarchical and ordinary least squares modeling techniques to support adequate validation and their extensive utilization in academic disciplines.