1 Introduction
Trust and the ability to make credible promises is central to well functioning mar-
kets. However, these preconditions cannot always be fulfilled. Badly functioning
enforcement system is a characteristic feature of many developing societies as well
as ones undergoing an institutional change. It is also typical for anonymous plat-
forms like the Internet.
By evidence and casual observations, the ascending English auction stands out
as a particularly useful mode of trading across different cultures and development
stages.1 Ascending auctions were in widespread use in the days of the Roman
Empire (the Latin root of the term "auction" means increasing). Yang (1950)
describes the ascending auctioning of clothes in Chinese monasteries in the 600
AC. Today the English auction is the far most frequently used auction format in
the Internet. For example, Lucking-Riley (2000) reports that almost 90% of the
Internet auction sites used the English auction as their selling mechanism.
It is not clear why the English auction is such a dominant auction form, and
why it is particularly so in circumstances of inadequate enforcement system. Pop-
ularity of the English auction is interesting since it is not, in general, the most
profitable auction mechanism (Myerson, 1981).2
This paper develops a theory that attempts to explain this empirical regular-
ity. The argument is not based on the profitability of the English auction but
rather its commitment properties. We claim that the English auction is the only
implementable mechanism if the market sides do not have any commitment power.
Lack of commitment power could be due to an inadequate enforcement system.
Full commitment is a standard assumption in the mechanism design theory.
In the auction set up it requires, on the one hand, that the seller (she) cannot
change the rules of the game in the middle of the play and, on the other, that
the buyers cannot leave the game once they participate it. By the commitment
assumption, it is legitimate to abstract from issues of bargaining under incomplete
information. However, it is not difficult to imagine that the seller, who has the
power to design the rules of the mechanism ex ante, may be able to redesign them
too ex post, once the information has been processed but the quo has not been
fulfilled. Familiar examples of auction manipulation include ex post bargaining
over the good, shill bidding, fees, reauctioning the good, etc..3 Redesign should
be especially difficult to prevent under inefficient occurrences.
To see what goes wrong in the absence of commitment, recall that incentive
compatibility and individual rationality are necessary for a (direct) mechanism to
work. If the seller holds the power to redesign the rules of the mechanism, she
wants to do so at least after any realization that leaves the buyers with positive
surplus. Since forward looking buyers anticipate that the seller will redesign the
1 Cassady (1967) is a standard reference on the history of auctions.
2 However, the English auction may be optimal in a restricted class of mechanisms, see e.g.
Milgrom and Weber (1982) or Lopomo (1998, 2001).
3 See McAdams and Schwarz (2006) for analysis and discussion on changing the auction rules
e.g. of the European 3G auctions. For another angle to shill bidding, see Izmalkov (2005).