108 S. Dasgupta and R.G. Hansen
seller releases an informative variable that is affiliated with the other variables, then
the expected equilibrium price (for all auction forms) is at least as high as when the
information is not released.
3.4. Limitations of the common-value and general symmetric auctions
For corporate finance situations especially, issues of information and efficiency in auc-
tions should be important. Existing models do not allow for full consideration of some
of these issues.
In the independent private values auction, efficiency has only one dimension: whether
the item is sold to the bidder with the highest valuation. In the pure common value
model, there is no real allocation problem so that from an efficiency standpoint, one
might as well allocate the item randomly. While a random allocation may not pro-
vide optimal revenue for the seller, one should be suspicious of a model focused only
on wealth-transfer and not efficiency considerations. One can imagine a variety of
economic forces outside of the auction process itself that will tend to cause efficient
processes to develop (competition between auctioneers, or even the law). Models that
assume away any possibility of inefficiency may cause us to lose sight of the true eco-
nomic issues in comparing alternative selling mechanisms.
TheMilgrom and Weber (1982a, 1982b)model brings an allocation problem back
into the picture, in that bidders’ valuations differ, so there are efficiency implications
of the allocation. On another level, though, this relatively general model still fails to
permit a complete role for economic efficiency. As pointed out byHirshleifer (1971),
information can have both private and social value. For information to have social value,
it must have the capability to affect the allocation of resources. One would expect that
in an auction context, information would not only allow bidders to refine their estimates
of value, but since the bidders do have inherently different valuations, one would also
expect that information would possibly change relative valuations. That is, with one
information set, bidderimight have the highest expected value; but with a different
information set, bidderjmight have the highest expected value.
The Milgrom and Weber model does not permit this kind of role for information.
A simple example suffices to show this as well as to illustrate why it is important to
allow information to play an efficiency role. Consider the following two-bidder, two-
state model:
State
AB
Bidder 1 100 200
Bidder 2 200 100In State A, the asset is worth 100 to bidder 1 and 200 to bidder 2, with the valuations
reversing for State B. Recall that a major result from theMilgrom and Weber (1982a,
1982b)model is that the expected price increases upon the seller’s release of additional
information. In the example above this result does not hold. Consider an open auc-
tion, and let the information on state initially be diffuse, with each state believed to be