Spectrum Auction Theory v. Competition Theory

As I’ve previously reported, Congress is weighing spectrum legislation as part of the Payroll Tax Holiday and Everything Else extension. One critical change pushed by House Republicans (with the enthusiastic support of AT&T, surprise surprise . . .) involves whether the FCC should be able to keep companies that have a lot of spectrum (AT&T and Verizon) from bidding on some licenses in the future. This is called “eligibility restrictions” (i.e., are you eligible to bid in the auction or not). The FCC has authority to impose eligibility restrictions now, but generally doesn’t. As the spectrum gap between AT&T and Verizon and everyone else in the wireless world gets bigger, however, there is some talk of possibly bringing them back.

Needless to say, AT&T and its supporters think this is both unfair and bad policy. Others, such as the folks at T-Mobile (now no longer being absorbed into the Bell Borg) have responded to the fairness argument.  For myself, I am always deeply suspicious whenever incumbents start arguing about “fairness,” as it usually means “please consider this particular detail in a total vacuum without ever thinking about all the unfair advantages I have, and use my framing because I appeal to basic values and use sports metaphors like ‘level playing field.'”  But lets set that aside and do the cold-hearted  policy wonk think. As Paul Krugman occasionally likes to say “economics is not a morality tale.” And in any event, even if we decided this on “fairness,” we’d still want to know the right answers and outcomes, right?

Two usual policy arguments are advanced for no eligibility restrictions. The first is that “auctions put spectrum in the hands of those who will use it most efficiently.” The second is that auctions with open participation increase total revenue. Lets pretend for a moment the first statement is true. Based on this, I shall prove below that not only does open participation decrease revenue, but it creates a serious conflict with competition policy. If we maximize auction efficiency, it is inevitable that the largest players will win the majority of the licenses and that this problem will grow worse over time.

So meet below the line for a video blog explaining this and some serious policy trade off discussions . . . .

As for whether some eligibility restrictions would increase or decrease the total revenue to the Treasury, I’ll refer you to this old piece from my friend Greg Rose. To summarize quickly, if people believe AT&T and Verizon will generally win whatever licenses they want, auction revenue will be lower. No one bets in a poker game unless they think they can win. Similarly, folks don’t bid very much against AT&T and Verizon because they are generally regarded as unbeatable for any licenses they want.

This, of course, begs the rational question: will AT&T and Verizon always win? After all, everyone has a shot, and someone like Apple with bags and bags of money could decide to jump into the auction. Why do I say that under auction theory, AT&T and Verizon will generally win the licenses they want at auction no matter what?

Here is a video explaining it.

To summarize for those who didn’t feel like watching. Auction theory predicts that spectrum will go to the entity that (a) values it most provided (b) that entity can pay for it. (Auction supporters sometimes say that auctions put spectrum to its highest, best use — which is crap. I’m collecting dumb-ass things like this for my future blog post “The Crazy Crap Coase Never Actually Said.”) Pretend we have a situation two large networks (AT&T and Verizon), some smaller networks (Sprint, T-Mo, U.S. Cellular) and potential newcommers everyone yaps about but who never actually show any interest in bidding (Apple). There’s a spectrum license worth, in some abstract theoretical sense, $5 billion. For Apple to take advantage of it, it will need to build a network, get customers, and otherwise incur lots of costs. So Apple will have to spend $3 billion to extract the value of the license. This makes the rational maximum bid for Apple $5 bn – $3 bn= $2 bn.

Now sure, Apple is supper rich, so in theory it can bid $20 squinjillion dollars for the license. But rich companies don’t generally spend more money on something than they expect to be able to make — that’s why they are rich. So Apple is extremely unlikely to come in and bid more than $2 bn for the license.

Next up is Sprint or T-Mo. They have a network and customers already. But it is not as big as those of AT&T and VZ. They will therefore not achieve the same economies of scale as AT&T and VZ. So for Sprint and T-Mo, it costs $2 bn to extract the value, making the rational bid $5 bn- $2bn= $3bn.

Now we come to AT&T and VZ. They have economies of scale, so they can realize the full value of the spectrum with only $1 bn in additional costs. So the rational bid for AT&T and Verizon for the same block of spectrum is $5 bn – $1 bn= $4bn.

But we now add a further advantage to AT&T and Verizon. Because they have more spectrum, they enjoy a competitive advantage. That is also valuable. If Sprint or T-Mo or a new entrant like Apple wins the spectrum, they will face stronger competition. So there is an additional value to the spectrum just to keep it out of the hands of rivals. This is called “foreclosure value.” If we assume another $1 bn in foreclosure value to AT&T and Verizon, their rational bid becomes $5 bn, as compared to a rational bid of $3 bn for Sprint or $2 bn for Apple.

Even if Sprint or T-Mo were prepared to bid more for the spectrum than was rational based on projected return for the asset, they face a second problem. In addition to being willing to pay, you have to have the capacity to pay. Sprint and T-Mo could get financing for a $3 bn bid, because they can make a profit and repay the loan. But if they go to investors and say “please lend us $6 bn so we can outbid AT&T and VZ, even though we can only hope to get $3 bn in value from the license,” I can tell you what investors will say. They will say “No.” Not all will, of course. Some will say “Hell NO!” But either way, Sprint and T-Mo are not going to get financing to bid more for licenses than they can expect to get back in profits.

By contrast, AT&T and VZ will have no problem getting financing, because they can actually exact that mush value from the license.

I am not saying a smaller competitor can never win any licenses. But the ability of a T-Mo to occasionally win some licenses at an inflated cost does not change the overall trend and where we ultimately end up. It is rather like the ability of the Yankees to attract talent v. the ability of the Pittsburgh Pirates. The Yankees are going to be able win the best players at auction, since there is no salary cap. Being a winning team, they will stay rich. The Pittsburgh Pirates will never be able to outbid the Yankees and other rich teams, so they will stay in the bottom of the league. This does not mean the Yankees win every game or the Pittsburgh Pirates lose every game — but that is generally the way to bet.

Spectrum Efficiency v. Competition

So if the duopoly is the most efficient user of spectrum (at least as measured by the ability to extract monetary value from the spectrum), that’s good — right? The system works as predicted and the auction puts the spectrum in the hands of the company that can generate the most value from it. indeed, taking this to its natural conclusion, the most efficient spectrum use is from a monopoly. A “natural monopoly” if you will.

Unfortunately, this conflicts with our desire to have competition. Setting aside the question as to whether the current wireless market is “competitive” — whatever that would mean — we have generally embraced the idea that we would like to see more than two competitors in the wireless market to ensure that we get the benefits of competition. Our auction theory tells us that, without the FCC’s ability to force AT&T and Verizon to occasionally sit out a spectrum auction or two, then the remaining players will become increasingly unable to compete because they will be unable to get enough spectrum at auction. Auction theory also tells us this tendency will get worse over time, and that no rational new entrant will try to enter the market.

Which brings me to one of my favorite aphorisms: policy is about trade offs. We can go with pure auction theory. But if we check that box, we know that AT&T and Verizon will continue to dominate. Or we can decide we like competition theory, in which case we need to make sure that competitors get enough spectrum to actually compete. But it is ridiculous to claim that our policy is to promote competition, to rely on competition to protect consumers and promote innovation, and then refuse to take any action when both theory and real world experience tell us the outcome will be to enhance consolidation and undermine competition. It’s like saying I want to stay dry in the rain, but refusing to use an umbrella because that would somehow be “unfair” to the rain or would be “picking winners” between umbrellas and raindrops.

We shall see whether Congress decides to protect competition by preserving the FCC’s ability to use eligibility restrictions when necessary, or if we will end up with a duopoly in the name of “fairness” and to avoid “picking winners and losers.” I know some people have trouble with making explicit trade offs like this. But life is unfair. It is rather naive to expect economics to somehow be “fair” when the reality it describes is not.

Stay tuned . . . .




Be Sociable, Share!
This entry was posted in Spectrum, Tales of the Sausage Factory. Bookmark the permalink. Both comments and trackbacks are currently closed.


  1. Your video’s theme music is audible; your voice isn’t. Without that, it is not clear why unrestricted auctions must produce less money; “may” is no problem. Are bidding costs significant? If not, under your example, Sprint really ought to bid 3 billion if ATT is in the auction, to cut down ATT’s profits (yes, ATT will cheerfully bid 3,000,000,001; but at least the duopoly is being taxed).

  2. Harold says:

    The video doesn’t cover that. The link to Greg Rose’s 2006 study for Center for American Progress does. In the short term, the auction dynamic raises revenue, as you predicted from my explanation. this is why the 700 MHz almost doubled the projected revenue. The institution of anonymous bidding foreclosed certain kinds of strategic behavior, forcing bidders to compete.

    The problem of declining revenue occurs over time. As it becomes clear the game is unwinnable, participation declines and revenues decline. Klemperer did some work in the early 00s on this to demonstrate why the UK model worked only for the UK when Europe did its spectrum auctions. The answer was that UK was the first spectrum auction in the EU. They could take advantage of the dynamic that occurs the first time the game is played before participants learn from experience.

    If we are only planning to have one auction ever, and we only care about maximizing revenue, then there is an argument for multiple, open, ascending auctions with anonymous bidding and small license areas as the revenue maximizing auction design.

  3. […] (I’ve talked about how this creates barriers to entry to new entrants and smaller competitors at length here).  But DOJ notes that because the number of licenses is extremely small, it becomes possible for a […]

One Trackback

  1. […] (I’ve talked about how this creates barriers to entry to new entrants and smaller competitors at length here).  But DOJ notes that because the number of licenses is extremely small, it becomes possible for a […]

  • Connect With Us

    Follow Wetmachine on Twitter!


If you do not have an account: Register