136 S. Dasgupta and R.G. Hansen
4.10. The spectrum auctions and the role of debt in auctions
Beginning in 1994, the Federal Communications Commission in the United States auc-
tioned licenses for the use of radio spectrum in designated areas. The licenses were
auctioned using a novel auction format involving sequential rounds of sealed-bidding
on numerous licenses simultaneously. At the end of each round, complete information
on the level of bids for all licenses was revealed. The auction format was designed by
economists, and at least in regard to the vast sums of money raised, was a great suc-
cess. Numerous articles summarize all aspects of the auctions, including their design
and performance: see, for example,McAfee and McMillan (1996), Milgrom (2000)
andSalant (1997). For the empirical researcher, FCC auctions provide a wealth of in-
formation: for example, the FCC Web site (http://www.fcc.gov) lists all the bids in all
the auctions. Moreover, many of the participating companies are publicly traded, so that
company-specific information is also easily available. We focus here on one analysis
which studied the effect of debt on the FCC auctions.Clayton and Ravid (2002)con-
struct an auction model where bidders’ debt induces lower bids than would otherwise
be the case. In this model, bidders have outstanding debt that is large enough to induce
bankruptcy if the auction is not won. Lower bids decrease the probability of winning,
of course, but in this case guarantee some residual to the shareholders conditional on
winning. In effect, in this model, pre-existing debt holders are “third parties” who have
a prior claim of a part of the pie. Thus, pre-existing debt serves to reduce bidders’ values
and therefore reduces bids.^44 An empirical analysis of the FCC bidding data produces
a negative but generally insignificant effect of a bidder’s own debt on their bid but a
negative and significant effect on a firm’s bid of competitors’ debt levels.
Che and Gale (1998)were the first to explicitly study the role of debt in auctions.^45
They have a result similar to Clayton and Ravid, although the models rely on differ-
ent effects. In Che and Gale’s framework, a second-price auction yields lower expected
revenue than a first-price auction. To see how financial constraints affect revenue com-
parisons, suppose that due to budget constraints, bidders cannot bid more than a given
budget, which is observed only by the bidder. The private valuations and budgetary
endowments of each bidder are independently and identically distributed according to
some joint distribution function. In this context, since bidders in the second-price auc-
tion bid their value, but in the first-price auction they bidbelowtheir value, bidding
is more constrained in the second-price auction because of budget constraints, ceteris
paribus. As a consequence, the first-price auction generates higher expected revenue.
Che and Gale (1998)allow for financial constraints that are more general than we have
considered here: for example, these could take the form of a marginal cost of borrowing
that is increasing in the amount of the loan.
(^44) On the role of debt holders as “third parties” in the context of bilateral bargaining, seeDasgupta and
Sengupta (1993). On the role of “third party” shareholders in the context of bilateral bargaining, seeDasgupta
and Tao (1998, 2000).
(^45) For a recent contribution on the role of financing in auctions, seeRhodes-Kropf and Viswanathan (2005).