Using data on all FCC auctions of spectrum related to cellular services from 1997 to 2015 we attempt to identify intrinsic spectrum values from winning auction bids. Our data set includes 17 auctions and close to 7,500 observations. We add two components to previous literature on this topic. First, we control for license and block specific auction rules that Connolly, Salisbury, Trivedi and Zaman (2017) catalogue. Second, we introduce two technological measures to separate out technological progress that effectively reduces spectrum scarcity from technological progress that increases demand for mobile applications. Previous papers have included simple time trends to reflect technological changes. Time trends are unable to distinguish between markets within the United States and conflate the effects of these two types of technological progress. Our results confirm previous theoretical and empirical findings for basic measures of demand such as population, population density, income levels, frequency levels, bandwidth, paired bands, and national licenses. Astonishingly, 49 percent of all cellular licenses since 1997 have been won by small bidders: 44 percent were won using small bidder credits, 14 percent were won in set-aside/closed licenses, and 9.5 percent were won in closed licenses using bidding credits. Our results further quantify the negative impact on headline winning bids when won using bidding credits, in closed licenses, and with the imposition of the open access requirement in the C block of Auction 73. Increased spectral efficiency appears to be reducing spectrum scarcity as evidenced by its lowering of winning bids, while market level communications infrastructure has a significant positive impact on the demand for and price of spectrum. Additionally, auction results confirm that the relative value of higher frequency spectrum is increasing over time as new technologies develop.
