"Even on a cost avoidance basis, the AWS3 bids could reflect economic values," a reliable source emailed me. More spectrum makes it cheaper to add wireless capacity, especially now that LTE-A aggregation is widely available. I ran some numbers and to my surprise even $45B is plausible - if not exactly cheap!

AT&T certainly didn't intend to pay $15-20B for about 40% of the 50 MHz available. John Stankey put $10B on the table for the auctions. That seemed like enough to scare off smaller bidders.  Someone wasn't scared, probably Charlie Ergen, and pushed the bidding up. 

Telcos need to keep up with traffic increasing at 30-70% per year. They can increase capacity with more cell sites, more antennas, and network efficiencies instead of spectrum. They calculate how much less they could spend on capex if they had additional spectrum available. They calculate cost avoidance numbers from that and develop a maximum bid range. Then they hire nobel-calibre game theorists and auction economists to ensure they spend much less.

The savings in capex just might justify the high prices. AT&T has just cut capital spending by $4B, the largest decline in the last decade anywhere in the world. Just a quarter of that, $1B, would support over $30B in spending on spectrum, at 2.8% effective interest rate.

(CFO John Stephens tells the street they are paying 2.8% these days.) Verizon faces similar costs. Assuming each is bidding $20B, that requires saving about $600M/year in capex, only 3% of their capital budget.

Low interest rates encourage high bidding. "Debt financing is cheap and plentiful." NYT "2014 was a time for deal makers to pop the corks of their Champagne bottles. The year was one of the best for mergers and acquisitions." The same logic applies to spectrum buys.

The interesting question may not be why bids are so high in this auction but why prior auctions were 50-75% lower. Prices in private sales were in line with auction prices. Verizon paid about $4B for 20 MHz spectrum. A similar amount of spectrum in this auction is going for over $15B.

There were over 10 significant wireless networks in the U.S. a decade ago, enough to assume lots of competitive bidding. There are only four now, too few to assume bidding will go to the actual value (efficiency.) It's not as absurd as most rural broadband auctions, unlikely to have more than one or two likely bidders. No one else has the necessary backhaul and local facilities. In Britain, British Telecom won every single rural auction because there were so few bidders. They were accused of charging twice what things would cost with more bidders.

Part of the answer is some telcos are willing to pay more today is that network quality has become a more important factor in marketing with smartphones. T's reputation was hit hard by a poor network in 2008-2012 as the iPhone took off. The pattern of increased traffic demand has been established. AT&T's Stankey works from an assumption traffic demand will increase ~40% going forward, but that's very uncertain. He wants to be ready if the growth turns out higher. AT&T is now much closer to Verizon in quality but many consumers don't believe they are catching up. (Verizon has cut capex by about 10%, narrowing the gap between them.) 

Each carrier faces different strategic choices; Verizon and France Telecom are believed to have superior networks and hence can charge higher prices. Making sure they have plenty of spectrum in reserve is important to them. T-Mobile U.S. emphasizes lower prices. TMUS customers are much less likely to complain if occasionally things slow down a little. 

     $20B is mind-boggling unless you're running a phone company doing $120B/year and you work with $billions all the time. 

A few notes:

With $10's of billions in play, these companies hire Nobel Prize caliber game theorists and economists to advise on strategy. Neither I nor any other analyst I know can work at that level. It's our job to come closer to the truth, but figuring out strategies here is hard. At 5 a.m. Christmas Day I couldn't sleep and spent some time reading the major auction theorists. I couldn't understand some of the papers.

A remarkable collection of theorists came to a CITI Columbia seminar June 20th. Peter Cramton, Larry Ausubel, Paul Milgrom and Greg Rosston were among the two dozen speakers. 

It's not impossible the high bids are coming because Verizon and AT&T are going hard against each other. Few believe that's likely, especially with another 50-100 MHz coming soon from the broadcast incentive auction. 

The 100% annual growth in the U.S. was an artifact of the introduction of smartphones. Where most people now have smartphones, traffic growth is down dramatically. The many policy people still claiming 100% wireless growth are either fooling themselves or presenting "politician's truths."  http://bit.ly/growthslowing.

In the calculation above of the $600M/year cost of carrying $20B in spectrum I have no amortization. That implies that the value of spectrum in the future will be at least as high as today. That's of course unproven. 

The "exclusionary value" of controlling the spectrum is greater than > $10B.  If anyone else wins spectrum, that company will need to win millions of customers to pay off the cost. Many will necessarily come from the Bells, whose combined annual sales are > $160B. Losing 2% of that revenue is > $3B/year in gross. The lost net would be close to $1B for each company because the marginal customers have lower cost to serve. 20 MHz is 4-6% of the working spectrum in the U.S.,so 2% or higher is realistic. 

On the other hand, spending $20B on spectrum has serious negative effects. It's spooking Wall Street and dropping the share price considerably. The higher debt ratio could affect the debt ratings, an expensive problem. The company's fixed interest payments will be higher, a severe problem in hard times.  

Undiscussed above is the importance of spectrum to AT&T's plans to shut down landlines in 50-75% of their territory and go wireless only. Both AT&T and Verizon are going slowly on rural fixed wireless for now. In a few years, with MIMO and 100 MHz carrier aggregation, they expect to have a much more competitive offering. By 2020 or so, both could offer 20-40 megabit speed, 100-150 gigabyte caps at a consumer price. Most Bell rural spectrum is unused today but extra spectrum could make a more robust offer in the future. Fixed wireless can use 4 and 8 antennas for MIMO, meaning in some terrain 20 MHz can deliver hundreds of megabits (shared.) MIMO doesn't deliver this performance everywhere, however. Ericsson Australia CEO HÃ¥kan Eriksson told me there was little MIMO gain in the flatlands of the Outback. MU-MIMO, still in the labs, may solve that problem.