In the “bids and offers” auction procedure, bidders may submit either (non-withdrawable) bids or temporary offers on packages. By providing these alternatives to the bidders, the procedure mitigates many of the problems inherent in dynamic package-bidding systems.

A paper presenting the details of this procedure can be obtained in any of three ways. Click on a link below to make your choice.

A: Perhaps I was too concise. Copied below are the last sentences from the paragraphs defining bids and offers (page 4 of the pdf).

"A bid that is declared a currently-high bid cannot be withdrawn. It persists across rounds of the auction until a new, higher bid (or collection of bids) replaces it."

"At the end of a round, all offers that are not part of the current prospectively-winning collection are removed from the system."

Assume that a round begins with bids of 5 on A, 5 on B, and 20 on {A,B}. If a bid of 11 is submitted on A, then the bid of 5 is removed from the system, and the bid of 11 stays until another bid on A tops it. If, instead, an offer of 11 is submitted, the bid of 5 stays. If a new prospectively-winning collection is announced during the round, and the collection includes the singleton package {A}, it will use the offer instead of the bid. However, if the offer isn't in the round-ending prospectively-winning collection, it is removed from the system.

1. The way the document is currently written, the concept of an offer is first introduced too vaguely, and then introduced too technically. This caused me to have to read the document several times before I felt like I really knew what was going on. In order to convince the FCC (and bidders), you'll have to make it easier for them to get an intuitive feel for the idea.

2. The business about determining the correct level "t" is a bit confusing. I've been trying to figure out a simpler way of accomplishing the same thing. How about a system with a 3rd party who receives secret info on willingness to pay for small packages, and reveals this information selectively - i.e., to create a "fair" outcome when it is seen that two parties have enough demand to be willing to cooperate to displace a package bid? I suppose this type of system might not be renegotiation-proof, and suspect to all sorts of gaming, but I wanted to share the idea with you anyway...

3. Will people really be encouraged to make offers under your system? I wonder whether they will be afraid of revealing information that they don't really want to reveal, even if the offer is potentially temporary. I confess that I don't have a concrete example in mind, so I could be wrong - it's just a vague sense I have. If you and Ricky could demonstrate that bidders do have an incentive to submit straightfoward offers, I'd feel more confident. Anyway, the combination of (1) the complicatedness of determining the minimum acceptable offer, and (2) potential concerns about unnecessarily tipping off other bidders, makes me worry about the plan's feasibility.

4. I wonder if there's a simpler way to determine minimum bid increments. For example, instead of using a rule of "minimum increment of 10%", how about a rule that uses a fixed grid of acceptable bids? For example, one could say "all bids between $1M and $5M must be even multiples of $0.25M, all bids between $5M and $10M must be even multiples of $0.5M," and so on... I wonder if this will make the minimum required acceptable bid more clear.

Reply:

1. I hope the reply to the preceding question helps. And I certainly hope that the FCC will request a more extensive presentation if they are interested in one.

2. With the appropriate user interface, bidders don't need to see the computation of t: They only need see the minimum acceptable offer level for a package. The idea behind the calculation is that (quoting from the top of page 4, with emphasis added) "in order to beat the {A,B} bid, at least one of the two single-license bidders must be willing to bear at least half of the threshold burden." A bidder only willing to bear less than half the burden (say, on B) must wait until some other bid or offer on A lowers the minimum acceptable offer level on B. (In this way, "frivolous" offers are barred.)

3. Offers provide more flexibility than do bids of the same amount. Their intention is to draw matching offers that help solve the threshold problem, without locking a bidder down or forcing the bidder to "bid against itself". An alternative to killing offers at the end of a round is to permit bidders to put a time-limit on their offers. This didn't seem necessary, since a still-valid offer can always be resubmitted at the start of the next round.

4. The "fixed grid" approach is entirely compatible with the bids-and-offers procedure. I used a percentage rule only to simplify the examples.