Student researchers at SIEPR have the chance to explore the institutions and issues that shape our world while working to improve economic policy.
By Krysten Crawford
More than two decades ago, Stanford economist Paul Milgrom played a key role in the design of the first wireless spectrum auction. Since then, the framework he helped create has been used in more than 80 auctions in the United States, generated billions of dollars in government licensing fees — and been replicated around the world.
So it made sense for the Federal Communications Commission to tap Milgrom in 2011 when the agency needed a new way to free up more broadband for mobile devices. It took him and a small band of fellow economists and computer scientists 18 months to design the auction, which finally opened last month after years of regulatory procedures, software development and presentations to potential bidders.
When the auction ends later this year, the country’s wireless landscape will never look the same.
The FCC is banking on this auction to solve a vexing problem: the public demand for mobile services has exploded since the iPhone debuted in 2007, and wireless service providers are struggling to satisfy their customers. And the old formula for spurring investment in wireless services by taking unused spectrum and auctioning it off to the highest bidders wasn’t going to work. The government didn’t have enough high-quality spectrum available to meet consumer needs.
So, for the first time ever, the FCC has set out to buy back spectrum rights already under license. Once done, the agency plans to auction the recovered spectrum to as many as 125 broadband providers – companies like Comcast, AT&T, Verizon and T-Mobile and a host of smaller regional players looking to secure spectrum set aside for them as part of the auction design.
The U.S. Incentive Auction, which kicked off March 29, is an undertaking of unprecedented magnitude and complexity. It is rife with political, economic and technological challenges, not the least of which have to do with the central players: a mobile broadband industry on the rise and the country’s 2,200 TV broadcasters in an industry on the decline.
The TV broadcasters hold the spectrum licenses that at the heart of the Incentive Auction and that many in the wireless world call “beachfront” property.
“This is by far the most complicated resource reallocation ever attempted, anywhere in the world,” says Milgrom, a senior fellow at the Stanford Institute for Economic Policy Research.
Milgrom’s No. 1 priority was to create an easy process, especially for TV broadcasters.
“We wanted to figure out how to put complexity under the hood,” he says. “You step on the gas and you don’t worry how the engine works.”
At first glance, the auction seems straightforward: it’s essentially two bidding contests in one. The first is a reverse auction, which allows TV broadcasters to sell their spectrum back to the government. The second, a forward auction, enables broadband providers to buy the rights to the newly available spectrum, which resides in the highly valued 600 MHz band.
The forward auction is expected to gross as much as $40 billion. As long as those proceeds are greater than prices offered to TV broadcasters in the reverse auction, the process works. The money left over goes back into the federal budget. And broadband consumers will ultimately benefit from faster, more reliable service.
“The FCC auction is designed to provide a market-based mechanism to move spectrum to more efficient and consumer-friendly uses,” says Greg Rosston, SIEPR’s deputy director and a SIEPR senior fellow who also helped design and implement the first spectrum auctions and is advising T-Mobile in the current auction.
“Converting spectrum use from lightly demanded over-the-air television to mobile broadband is likely to provide billions of dollars in consumer welfare gains,” he says.
But the challenges to building the well-oiled machine Milgrom envisioned were enormous. Thomas Wheeler, the FCC chairman, used a different metaphor when he described last year the process as having “more moving parts than a Swiss watch.”
For help, Milgrom pulled together an interdisciplinary “dream team” of top experts in economics and computer science: Jonathan Levin, also a SIEPR senior fellow and faculty member in economics; Ilya Segal, a professor of economics at Stanford; and Kevin Leyton-Brown, a computer scientist at the University of British Columbia who earned his PhD from Stanford.
Consider this challenge the design team faced: The spectrum that the FCC buys back from TV broadcasters has to form a contiguous block of signals spanning the country. Without that uniformity, wireless companies can’t offer nationwide coverage. This means that, to ensure a seamless block of signals, the FCC will have to condense broadcasters that do not sell in the auction into a smaller, more tightly packed spectrum band. The process is called “repacking” and can happen only after the reverse auction, when the FCC knows how much spectrum it has to sell to broadband providers, and before the forward auction.
Repacking has been called the linchpin in the FCC’s overall plan, but it isn’t easy to do. In effect, the FCC assigns a broadcaster that hasn’t given up its spectrum rights a new channel on the TV dial. But switching channel assignments is tricky from an engineering standpoint. It can cause interference with another station’s signals, which means viewers will see static on their screens. By law, the FCC has to assign TV stations channels that are interference-free and cover the same potential audience.
As it turns out, there are roughly 2.7 million ways in which a channel reassignment could cause interference. The FCC needs a way to flag them almost instantly as it weighs which TV station licenses to buy back and at what price.
The best commercial software wasn’t up to the task. The best algorithms on the market could do an interference analysis in five minutes and give the FCC a definitive answer on whether a channel combination would work only 50 percent of the time.
Commercial software was a non-starter. The auction would take decades to complete.
So Leyton-Brown, the University of British Columbia professor, built an algorithm that can resolve any interference issue in less than one minute 98 percent of the time. Often, the software takes less than a second to come up with a definitive recommendation for the FCC.
The interference question is just one of the team’s achievements. But it helps tell the story of what they set out to accomplish, and the need to take an interdisciplinary approach to solving problems
“We wanted to make this process easy for bidders,” says Milgrom, “and we achieved that.”
Krysten Crawford is a freelance writer.