Methodological Developments and Issues in Experimental Auctions
(Rodolfo M. Nayga, University of Arkansas, Organizer)
ARE EXPERIMENTAL AUCTIONS DEMAND REVEALING
WHEN VALUES ARE AFFILIATED?
JAY R. CORRIGAN AND MATTHEW C. ROUSU
The theoretical properties of purely private
and purely common value auctions are well
understood (Krishna 2002), and have been
extensively tested in the lab (Kagel 1995).
Empirical researchers, however, have paid
much less attention to the arguably more realis-
tic scenario where a good’s value is “affiliated”
or determined by a combination of private and
common value components.
Milgrom and Weber (1982) show that when
a good’s value is affiliated, the second-price
sealed bid auction ( Vickrey 1961) is no longer
incentive compatible, because rational agents
must adjust their bidding strategies to take
into account that the winning bidder’s com-
mon value signal likely exceeds the good’s true
common value. Auction participants’ bids then
no longer reflect their best guess of a good’s
value, but instead are adjusted downward to
avoid the winner’s curse. In this environment,
second-price auctions may also lead to ineffi-
cient allocation if the bidder with the highest
private value receives a relatively low com-
mon value signal and, as a result, is outbid by a
competitor with a lower private value.
While these theoretical predictions are well
understood, economists have only recently
begun to test them empirically. Work in
this area has focused primarily on alloca-
tive efficiency and revenue maximization (e.g.,
Kirchkamp and Moldovanu 2004).These issues
Jay R. Corrigan is an associate professor, department of Eco-
nomics, Kenyon College, Gambier, Ohio. Matthew C. Rousu is an
associate professor, department of Economics, Susquehanna Uni-
versity, Selinsgrove, Pennsylvania. The authors thank Greg Colson
and Jayson Lusk for their helpful comments. Corrigan and Rousu
share lead authorship. This article was presented during an invited
paper session at the 2010 AAEA annual meeting in Denver, Col-
orado. The articles in these sessions are not subject to the journal’s
standard refereeing process.
are of primary importance, for example, to
policy-makers designing a radio spectrum auc-
tion where firms with heterogeneous costs
compete to buy one common value good. But
efficiency and revenue are less important in
the experimental auction valuation literature
(e.g., Lusk and Shogren 2007). What is impor-
tant here is the extent to which bids provide
an accurate and unbiased reflection of auction
participants’ underlying value signals.
Value affiliation could arise in an experimen-
tal auction environment in a number of ways.
For example, participants may be certain of the
value they place on the good up for auction, but
uncertain of the price the good sells for in the
field. Recognizing that auction bids should not
exceed the field price (Harrison, Harstad, and
Rutström 2004), censored values become affil-
iated if participants believe their competitors
possess additional information about the true
field price.
Alternatively, auction participants may
“mark down” the value of any good sold in
an experimental auction because of concerns
about whether the good will actually be
delivered at the end of the auction, or about
the quality of that good relative to field substi-
tutes. Marked down values become affiliated if
participants believe their competitors possess
additional information about the auction
market’s credibility.
Still another possibility is that while auc-
tion participants may be certain of the value
they would derive from a conventional good,
they may be uncertain of the value they
would derive from a similar good endowed
with some novel trait. One example would
be a comparison of conventional fresh pro-
duce with locally grown fresh produce. The
“locally grown” designation could take on
Amer. J. Agr. Econ. 93(2): 514–520; doi: 10.1093/ajae/aaq140
© The Author (2011). Published by Oxford University Press on behalf of the Agricultural and Applied Economics
Association. All rights reserved. For permissions,please e-mail: journals.permissions@oup.com