Despite it coming as a “surprise” to many reporters (and Wall St analysts) that DISH ended up with more total winning bids (before DE discounts) than Verizon in the AWS-3 spectrum auction, and that DISH got a 25% DE discount on its bids, the outcome is exactly what I predicted from the bidding patterns back in November. I was particularly amused to look back at Jonathan Chaplin’s comment from his December 7 report which poured scorn on my thesis, stating:
Some have suggested that DISH is distorting prices by bidding against themselves (DISH has three bidding entities that can’t communicate with each other during the auction). While possible, this is highly, highly unlikely.
[As a reader suggests, perhaps I should take this opportunity to note Chaplin's follow-up proposal on January 11 that DISH should do a LightSquared and set up a wholesale capacity business generating $10B per year. While possible, this is highly, highly unlikely.]
Its useful to examine exactly why DISH was so successful in driving up the price of the AWS-3 paired spectrum to a price far beyond anyone’s expectations. One of the key objectives for a bidder in the early rounds of an auction is to discover the amount of spectrum that its rivals are looking to acquire (only later does it become possible to discover how much they are prepared to pay for that spectrum). The price usually rises fastest in the key cities and then as the mix of demand becomes clear, bidders can switch over to second tier licenses knowing roughly how much spectrum they will end up being able to win.
We know that AT&T was looking to buy a paired 10MHz block, and it seems likely that Verizon would have been seeking roughly the same. Meanwhile T-Mobile wanted to selectively pick up one or two paired 5MHz blocks. If DISH hadn’t been bidding then everyone could have got what they wanted at close to the reserve price. However, adding DISH to the mix meant that the four key players were trying to buy more than the 2x25MHz of paired spectrum that was available.
More importantly, DISH was bidding through three separate entities and instructed them to bid on all the licenses simultaneously in key cities, to ensure that AT&T, Verizon and T-Mobile simply didn’t know how much spectrum each other and DISH were looking to buy.
The chart below shows the bidding patterns for the G, H, I and J blocks in New York (the G block is a smaller 2x5MHz CMA license, while the H and I blocks are 2x5MHz BEA licenses and the J block is a 2x10MHz BEA license).
We can see that all three DISH entities bid on every one of the New York paired license blocks they weren’t already holding all the way through Round 15, by which time the total combined gross price had reached $2.81B ($2.28/MHzPOP). In fact, it wasn’t until Round 18 (when the price reached $3.81B or $3.12/MHzPOP) that DISH’s bidding on these licenses began to slow (and SNR even overbid its own winning bid in Round 17).
[Incidentally, DISH's 3 entities combined were the biggest bidder for much of the auction, notably as late as Round 63, where they held $14.7B of gross PWBs or 35% of the $41.6B total, compared to $12.6B for AT&T, $10.5B for Verizon and $2.1B for T-Mobile. When the reserve price was met in Round 13, DISH held a total of $5.4B of PWBs, 44% of the $12.3B auction total at that point in time, compared to only $2.7B for AT&T, $2.1B for Verizon and $1.3B for T-Mobile.]
DISH clearly wrote the instructions to its DEs very well, because in the end there were very few cases where the final winning bid from SNR was topping an existing bid from NorthStar or vice versa (the largest license I’ve seen where this happened is the B1 unpaired license in Tampa BEA034 which sold for $21.4M before the DE discount). And it does seem that DISH complied with the letter of the rules: even though the FCC still needs to rule on whether the DE discount should be granted, it seems unlikely the FCC would want the auction to descend into chaos (which could theoretically result in a re-run).
However, its clear that the rules for future auctions will need to be rewritten significantly – I would expect severe restrictions on DE discounts and common ownership of different bidding entities at the very least. Indeed, it will now be very difficult to come up with a workable structure to advantage smaller operators like Sprint and T-Mobile in the incentive auction next year.
Where does the outcome leave us? Ergen did not buy a readily deployable collection of spectrum, instead seeking a blocking position in key cities (including New York and Chicago) in an attempt to force other operators to make a deal with him. Interestingly, most of DISH’s paired AWS-3 spectrum is in the G block, which is adjacent to and perhaps more quickly usable with the AWS-1 spectrum band, rather than being aggregated directly with the adjacent AWS-4 downlinks in the longer term like the J block. DISH also acquired most of the unpaired uplink blocks, which appears to be a hedge against the potential (and now perhaps likely) loss of LightSquared.
However, with AT&T winning enough AWS-3 to meet its spectrum needs (and make it highly indebted) for the next few years (not to mention AT&T’s ownership of DirecTV which makes a tie-up with DISH very difficult), it seems clear that Ergen is setting his sights squarely on a deal to sell DISH (or perhaps more likely lease its spectrum, given the difficulty of reaching agreement on a sale price) to Verizon.
So now, as I pointed out in November, the key question is whether Sprint will take this opportunity to satisfy Verizon’s spectrum needs through a sale of 2.5GHz spectrum? Given everyone in the industry is fed up with Charlie, that certainly seems like a plausible next step.