Auctioning the Digital Dividend: A model for spectrum auctions

Abstract

We model a spectrum auction where firms purchase units to participate in a constrained, multi-product, downstream market. We use dynamic programming techniques to numerically solve for the optimal bidding strategy in a clock auction. Firms value constraining competitor market power, so incumbents often bid aggressively to shut out entrants. We find that high cost firms may hold up the market, so the auction may be inefficient and generate zero revenue. An auction may be optimal for a regulator maximising total surplus. A regulator maximising auction revenue sets reserve prices high enough to restrict spectra sold, effectively behaving as a monopolist.