After advising on spectrum auctions on five continents, what we have learned is the unpredicatbility of what happens during the actual auctions. The “what” has included security breaches, a bidder bidding on the next auction, the dropping out of an established local operator (the first time ever in the case we were advising on), and the casting aside of the spectrum valuation figures that a reknowned investment bank had projected.
Along with the case where two parts of the government were openly arguing over how the auction should unfold, these were not examples of how game theory is usually applied to spectrum auctions. In fact, they illustrated how auctions can have games within games, plus side games and meta games.
In the “next auction” case, the winning bidder was effectively bidding in two auctions at once, submitted an extremely high winning bid in the first auction—six times as high as the next highest bid—in order to have an impact on the second auction in a second country. The general conclusion in the first country was that the bidder (a mobile operator) was either irrational or had seriously overestimated the value of the market and the associated spectrum. This perception of irrationality led competitors to forego bidding in the next auction, where the stakes were 20 times as high. As a result, by thinking of the first auction as one of a series, the irrational operator saved more than a billion dollars in the second auction.
The case of the two government agencies fighting over the outcome of an auction happened in India. The official auction management entity was charged with raising the highest fees possible for the government while the other, the regulator, sought to end the auction as quickly as possible on the theory that this would keep service prices from rising. The regulator’s announcement of a proposal to levy significant annual spectrum fees after the auction was over resulted in financial analysts down-valuing some of the operators, their stock prices dropping, and the auction ending within a few days, after having run for weeks.
In still another case, the auction rules called for the three pre-qualified bidders to be sequestered during the auction with no access to newspapers, TV, radio or the Internet. The staff serving them meals were to draw lots before each meal so that no communications in or out of the compounds where the bidding teams were confined could be pre-arranged. Patroling helicopters were to ensure that no carrier pidgeons were being deployed. Finally, only one of the bidders had its owner on the bidding team. This positioned that bidder quite differently from the others, who had to bid within non-adjustable bidding limits.
The other main lesson that we draw from our spectrum auction cases is the impact of the financial cycle on winning auction prices. When the cycle was up, as was the case in early 2000, winning prices set all-time records on a per MHz/POP basis. No peak since has resulted in higher prices nor any trough in lower ones than those that followed in 2001 and 2002.
That said, no spectrum authority or regulator has, to our knowledge, developed a framework or set of rules for factoring in financial cycles when deciding on the timing of auctions. In general, the auctions come when the consultations and preparation process for a spectrum auction ends, irrespective of whether the authority is trying to maximize government revenues or to stabilize mobile service prices. Barring a call from upstairs, that’s when the auction takes place.
Our spectrum auction practice extends back to the early 1990s. Since then we have worked on auctions in Brazil, Canada, Israel, Italy, Jordan, Malawi, Taiwan, Tanzania Thailand and other countries. We have trained C-Level bidding teams using bidding software and mock auctions, structured reverse auctions, prepared valuations, and advised operators on how and how much to bid. We have also prepared a series of case studies, including the ones summarized above, on how auctions have unfolded as well as assessed the factors underlying winning auction prices across OECD markets during a 15-year period.