Novel-Infrastructure Bidders’ Impact on Cost-Based Subsidy Auctions: Evidence from RDOF Phase I

FCC Auction 904 (“RDOF Phase I”), a reverse-subsidy auction for broadband ISPs, determined its reserve subsidy amounts using the Connect America Cost Model (CAM). The CAM estimates the incremental cost needed to provision broadband service at a location, based in part on fixed-line infrastructure models such as fiber-optic or cable networks. The auction rules also included clear preferences for higher quality services, but incorporated other aspects of technology neutrality. In the end, three of the top four winning bidders, who collectively received over one-third of awarded subsidies, were not fixed-wireline incumbent ISPs. One of these bidders, SpaceX, committed to provide broadband internet service using a constellation of low-earth-orbit (LEO) satellites, and was awarded about 10% of total subsidies in the auction. Since the cost structure of deploying broadband using LEO satellite constellations differs greatly from the cost structure underlying the CAM model (and therefore RDOF Phase I’s reserve subsidies), we use the RDOF Phase I results to understand the role of novel-infrastructure bidders in subsidy auctions.

We find that LEO satellite ISPs bid far more for low-subsidy areas than other bidders, consistent with the idea that infrastructure cost advantages made these areas desirable when they were less so for wireline ISPs. Given the auction rules’ preference for Gigabit bidders, we further find that when LEO bids against non-Gigabit wireline ISPs, LEO almost always wins due to its infrastructure cost advantages. Finally, we provide a conservative initial estimate of the premium paid by the FCC to preference high-quality Gigabit bidders of $2.5 billion, or 27% of the auction’s total subsidy award.