I rarely gush enthusiastically over a Notice of Proposed Rulemaking (NPRM) from the Federal Communications Commission (FCC), but I will make an exception for the recently released Incentive Auction NPRM and associated Appendix on auction design. As Republican Commissioner Ajit Pai observed in his separate statement, it has become almost cliché to observe that this is “the most complicated set of spectrum auctions ever held by any country.” What the NPRM explains, if you are willing to plough through it, is why it is so insanely complicated.
Unfortunately, the complication has given rise to a number of misunderstandings about what is actually going on here. In this case, a failure to understand why this is so complicated, rather than simply knowing that it’s complicated, can result in bad policy. The most critical misconception I have encountered to date is that the incentive auction involves wireless companies bidding for broadcast licenses, with the FCC acting as a sort of spectrum Christie’s. That is, after all, how this got sold and broadcasters and wireless companies seem to be the main players.
Below, I explain why this is not merely wrong, but why visualizing the auction in this way leads to policy choices that almost guarantee failure. It also bears directly on one of Commissioner Pai’s questions: why has the FCC proposed ending the auction as soon as the “victory conditions” set by Congress are met, rather than keeping the auction open as long as there appears to be the possibility of more willing bidders. . . .
Lets start with the framework. Congress set down a bunch of rules in the Spectrum Act of 2012 that limit how the FCC can run the Incentive Auction. The limits bite hardest on how much leeway the FCC has to move around the surviving broadcasters (“repacking”). Next hardest is that you only get to run one Incentive Auction. So if something goes wrong, like broadcasters pricing themselves out of the market or a band plan that attracts few bidders, there is no do-over. The usual way people establish the market price, by offering something and seeing how the market responds, is not an option here. So the FCC needs to make a lot of guesses about how both broadcasters and wireless carriers will react to various possible offerings.
The easiest way to run an incentive auction of the kind required by the Specrtum Act, therefore, would be for Wireless Bureau Chief Ruth Milkman and her auction-expert sidekick Gary Epstein to borrow Doc Brown’s DeLorean, run the reverse auction (where the broadcasters sell their spectrum rights), take the results back in time six months, develop a band plan based on those results, then come back and run the forward auction. Barring time travel, it gets more complicated. For the Incentive Auction to succeed:
Broadcasters must sell their licenses (and other spectrum rights), at below market price, to the FCC.
The FCC must then simultaneously sell them, preferably at above market price, to wireless broadcasters.
That’s not just my assessment, that was also the assessment of the Congressional Budget Office (CBO). When asked to score incentive auctions for the first time last year, CBO noted:
“It is difficult to predict how much spectrum would be auctioned by 2021 because of the time and cost involved in moving existing users. For example, the amounts auctioned as a result of incentive auctions would depend on the willingness of two satellite licensees and dozens of television broadcasters to sell their existing spectrum rights at a price that is below the market value of their licenses.” (emphasis added)
Why do we need to do it this way? So the government can make a profit. If we didn’t care about the government making money from the Incentive Auction, we could have the FCC set up some kind of Kickstarter-like mechanism where in order to have an auction you would need to have some minimum number of broadcasters express interest. If we only wanted the government to make a modest amount of money, we could let the FCC take a commission of 5-10%. But we not only want the government to make money, we want it to make gobs and gobs of money. To make that happen, broadcasters must surrender their licenses for much less than what wireless carriers will pay for the same MHz/Pop. (Megahertz per population density, or MHz/Pop, is the standard for measuring the value of a license. You divide how much someone paid per MHz by the population of the license covered.)
To make it more fun, everyone involved – the broadcasters, the wireless carriers, and the FCC – all have the same data on past auctions and the same set of consultants. They even have the same data to tell them, in a general way, what the expected cost of giving up spectrum rights is to a broadcaster and what the expected return would be for a wireless carrier. Ideally, the rational broadcaster would sell its spectrum rights at cost + $1 profit. Unless, of course, the broadcaster thinks it can get a lot more than that – in which case the rational broadcaster will try to maximize profit by charging what it thinks the carrier will pay – $1. Even in the best case where the broadcaster bets right, that leaves no money for the federal government.
But it is even more likely that, in the absence of anything else, the broadcaster would set the price too high. Why? Because even with the same general data, there is a range of estimates as to the value of the spectrum to wireless carriers. Broadcasters have every incentive, absent some countervailing pressure, to believe the most optimistic estimate. Wireless carriers, by contrast, have every reason to believe the most pessimistic estimate.
This is how the market for homes works under normal circumstances (if anyone can remember back to when we had neither a bubble driving up values or a recession driving values down). Sellers looked at what comparable houses in the neighborhood went for, then put a home on the market for a price in the same general price range. Usually, though, the home owner will err on the high side in the firm belief that his or her home is superior to other similar houses. Potential buyers come and look at the house, and decide that it would do if they make all sorts of changes to things like the wallpaper and the countertops. Mentally deducting all these things from the offered price, the potentially buyer offers something on the low end of the price range.
In the housing market, the parties negotiate and, ideally, arrive at a solution. But we don’t get to do that here because broadcasters are not negotiating with the wireless carriers. We need them to hand off spectrum rights to the FCC so the FCC can do all the fix ‘em work and resell the repurposed spectrum to the wireless carriers. Broadcasters have no reason to sell, other than their desire to make money. So why not offer at the highest price? If the auction fails, you just keep on being a very profitable broadcaster.
How Can The FCC Get Broadcasters To Offer Lower Prices?
To make this work, the FCC has to create some kind of pressure to encourage broadcasters to bid lower than what the FCC hopes to get from wireless carriers. What the FCC has proposed is a mechanism for getting broadcasters to bid against each other by limiting the amount of spectrum it will accept back. Since the broadcasters want to make a profit, the fact that they are effectively bidding against their fellow broadcasters (hopefully) pushes them toward the lower end of what the broadcasters will consider acceptable prices for the relinquished spectrum rights.
To provide an example, lets say five stations in the market might be interested in relinquishing their spectrum rights in various ways. Lets pretend this includes exiting the market (which yields 6 MHz) and other possible moves (such as relocation to VHF or channel sharing) for a total of 20 MHz recovered out of a potential thirty. If the FCC offers to take all 20 MHz, then nothing pressures the broadcaster to lower its bid unless the broadcasters have seriously underestimated wireless carrier demand. The most likely result is that the broadcasters will demand about as much money as the wireless carriers will pay, leaving little if any profit for the federal government.
But suppose the FCC says it will only take 10 MHz in the market rather than the full 20 MHz that the broadcasters in the market will offer. The 5 broadcasters offering a combined 20 MHz now must bid against each other to have their spectrum rights accepted. This forces them to demand a lower price than the wireless carriers will bid, because they won’t even reach the wireless carriers unless the FCC accepts their offer. Ideally, the broadcasters will bid each other down to their lowest acceptable price, then the FCC will turn around and resell the reconfigured capacity to wireless carriers at the highest price obtainable from the market.
The downside of this approach is that it imposes a limit on the spectrum the FCC can recover. In fact, the only way to maximize revenue for the government through this method of getting broadcasters to bid against each other is to guarantee that the government will accept less spectrum than it could have obtained at a higher price. It also raises a second question: at what point do you close the auction? If the broadcasters are still bidding while the wireless carriers are bidding, the higher price offered by the wireless carrier impacts the willingness of the broadcaster to accept a lower price.
The NPRM tries to deal with this question by creating pressure on wireless carriers to bid high while simultaneously creating pressure on broadcasters to bid low. Under the proposal, the auction ends when all the “victory conditions” set by Congress are met: enough money to pay for the auction and reallocation. Ideally, the fear that the auction will end drives broadcasters to bid lower and wireless carriers to bid higher. But it runs the risk of cutting off the bidding when wireless carriers are still willing to bid more.
All of these complicated questions are raised by Commissioner Pai’s first question in his concurring statement. Pai is quite right that Congress wanted to maximize auction revenue, so asking how to maximize revenue – while normally a prohibited question in FCC auction design under the relevant statute – is a perfectly appropriate question here. (This does not make me happy, but it is one of the many ways this auction is different from all other spectrum auctions.) Unfortunately, Pai’s analogy to a “reserve price” (which seems to treat the government as a bidder in both auctions) and appeals to “traditional auction theory” worry me that he (and others) are misconceiving the shape of the auction. It is not enough to construct an auction in which wireless carriers will bid high. It is also necessary to construct an auction in which broadcasters will offer their spectrum rights at the lowest possible bid.
I am not convinced that traditional auction logic will help us here. The FCC faces the difficult task of providing both parties with the same information and pushing them in opposite directions. The most obvious way of doing that is to create in both parties a fear that they will be locked out of the auction unless they make their “best” bid early. The alternative is to hold the reverse auction and the forward auction sequentially. While this may provide greater certainty, it also significantly delays the time it will take to get spectrum to the carriers.
Unfortunately, traditional auction theory does not have much to say on this subject or on how to evaluate these concerns. Typical two sided auctions like this generally involve the platform operator taking a commission rather than buying the licenses and “flipping” them. in addition, the intrinsic value of any given broadcast license that the FCC would “buy” at auction is very different from the potential value when re-banding and “selling” new 600 MHz licenses. (This is one reason why the FCC has proposed the modular “5 MHz licenses” to try to limit the uncertainty, but I will save that for a different post.) In evaluating the comments as they come in, stakeholders and Commissioner both need to constantly remind themselves how radically different this two-sided auction will be from a traditional auction.
Stay tuned . . . .