Assessing the 700 MHz Order Part IV: Lingering Doubts and Details

The Wireless Bureau has released its Public Notice (“PN”) for the 700 MHz auction. In addition to setting the date for the start of the auction as January 16, 2008, the PN also addresses a bunch of questions left over by the Order. The biggest of these are: (a) Setting rules for package bidding; and (b) setting reserve prices on a “per block” (rather than “per license” basis) for the auction.

What does that mean? Well, the Commission in the Order decided to protect itself politically from accusations that it had set the rules too aggressively and therefore killed the auction. The Commission therefore used its authority to create “reserve prices,” or minimum prices that bidders must meet before the Commission will award the licenses. The Commission has used reserve prices before, but generally on a license by license basis not on a “block by block” basis. Nor has the Commission ever guaranteed a reauction if a block of licenses fails to meet a set reserve price.

“Package bidding,” as discussed in previous posts, means allowing people to bid on a set “package” of licenses rather than requiring a bidder to bid on each license individually. This encourages people to bid because it means I won’t get stuck with licenses I don’t want if I fail to win one or two critical licenses that make it worthwhile (this is called the “exposure” problem). So if I only want the C Block licenses if I can get national coverage, I will still participate in the auction because I know if I lose any C Block licenses, I won’t get stuck paying bilions for the licenses I did win but now no longer want.

The use of this combination of factors, along with the failure of the Commission to adopt an “either/or” rule that would require a bidder to go after either the D Block license or C Block licenses, makes me uneasy. I can see scenarios where a bidder gets the D Block cheap, then chooses to enhance coverage by bidding aggressively for one or two C Block licenses. That’s not necessarily bad, except it may prevent the creation of a second national player because it deprives the second national player of licenses it needs to complete its package (I’m not postulating deliberate blocking, you understand, I’m looking at the potential interplay of circumstances frustrating the likelihood of new national entrants). OTOH, the ability to bid on both D block and C Block may encourage bidders to be more aggressive in both blocks, and may create a larger pool of bidders for these blocks over all.

But what really worries me is the reserve prices. Why? See below . . . .

For some reason, people think it’s easy to assess how much spectrum licenses are “worth.” That’s why people come out with estimates all the time. Indeed, wildly optimistic or pessimistic estimates of auction revenues have become a staple of auction debates. While the Congressional Budget Office (CBO) uses fairly conservative methodology for the perfectly good reason that it should be conservative about revenues that don’t exist yet when planning the federal budget, the happy happy joy joy estimates of industry participants and others tend to set unreasonable expectations among members of Congress and the public that follow this.

But the plain fact is that even conservative estimates like those of CBO are wild ass guesses based on damn little information and subject to too much variability. Each spectrum auction represents an almost unique confluence of circumstances around existing technology, existing entities able to get financing, the adopted service rules, the perceived competition, and the unique physical characteristics of the spectrum auctioned. That’s a lot of variables. And the only relevant data are previous spectrum auctions with entirely different sets of characteristics, or auctions abroad, or revenue for existing services under existing competitive conditions, existing technology, and existing service offerings.

Let me demonstrate how complicated this is with recourse to a simple example. Suppose I had tried to estimate in 2005 what I could sell my house for on January 16, 2008. At the time, the values of neighboring houses kept rising significantly, and had for some years. Population growth in my neighborhood remained steady. While I might conservatively estimate that the rate of increase would level off, I doubt I would have predicted the current meltdown in the mortgage market and the crash of local property values. But predicting the value of my house three years ahead of time is significantly easier than trying to predict spectrum license values. I have a much better historical basis for comparison and lots of similar objects being sold under similar circumstances. I have many fewer variables to control for, and a much bigger pool of potential buyers.

So now that we appreciate the complexity of the problem, we can take a much more jaundiced view of our ability to set reserve prices based on spectrum auction estimates. In theory, reserve prices can be useful to ensure that bidders properly assess the value of the licenses and prevent licenses from going too cheaply. The problem is that the FCC is doing this because it wants to protect it’s regulatory patootie. It wants to make sure that the auction gets the minimum $10 billion CBO estimated back in 2005 it should make. So if it looks like the revenue is in trouble, the FCC will automatically reauction the licenses without conditions (C Block) and without the aggressive build out requirements (A and B Block). It is unclear what they will do to the D Block if the D Block doesn’t meet its reserve price.

The reserve prices are set with reference to the AWs auction. On the one hand, the actual pricing in the AWS auction was pretty low compared to previous auctions. Given the better qualities of this spectrum, using those numbers seems prudent on the surface. Until you consider how much the value of my house has crashed since even 2006.

The other problem is that bidders have an incentive to bid low, because they would prefer to have the licenses without conditions or without aggressive build out requirements. Given a relatively low pool of serious bidders, bidders may hang back and underbid in the first auction in a deliberate effort to trigger a reauction.

The FCC has tried to minimize this problem by limiting reauction to blocks that don’t make their reserve prices rather than requiring the entire auction to meet a minimum $10 billion. So if we have brisk bidding for the C Block and it goes for more than $4.6 billion in aggregate, but slow bidding for the A Block, then only the A Block will get reauctioned.

But now we get to the second problem. Traditionally, the FCC has set reserve prices on a license-by-license basis, or an aggregate for the total auction, and with good reason. A whole bunch of licenses are just dogs. They never sell. While dividing things up into smaller CMA and EA blocs is good for people trying to target specific markets, it means you get slices of territory that absolutely no one wants to cover because they are not expected to be economic at any price. For example, if you look at the Summary Page for the AWS auction, you will find that the FCC ended up holding 35 licenses that no one wanted. And there, the FCC set the total aggregate reserve price fairly low (about $2 billion for the total auction).

We have no good way to know what happens when you link reserve price with an automatic reauction, or when you set the reserve price by block. The reasons for doing so here do not rely on any kind of auction theory. Instead, the Commission appears to regard the reserve price as a form of insurance. But as real auction theorists (as opposed to yr hmbl obdnt blogger) explain, the knowledge of reserve prices and of the expected consequences of reserve prices impact the behavior of bidders in the game.

It doesn’t have to be a failure, I point out. Because the FCC intends to limit participation in a reauction to those who participated in the first auction, bidders cannot draw down the bidding pool by sitting out. That means they will need to maintain a minimum level of activity to stay in the game (the FCC auction rules penalize you for sitting on the sideline by reducing your available “bidding credits” if you fail to maintain minimum activity levels and don’t use a waiver). Even if prices start low, parties may swoop in to “bargain hunt” creating pressure on bidders to bid competitively rather than hold back and wait for reauction. The uncertainty over bidder identity will likewise push bidders to bid aggressively rather than risk potential rivals capturing the licenses cheaply (this affect should tapper off as the price of the license rises out of “bargain basement” into “real money” range). All I’m saying is I don’t know. And not knowing makes me nervous.

Stay tuned . . . .

Next (and last):
Part V: The “Property School” Takes It On The Chin.

Previous:
Part I: Putting Things In Context
Part II: C Block Conditions Matter
Part III: Anonymous Bidding