Why do IPO auctions seem to fail?

Remember when Google (partially) bypassed the investment banks and held its own auction of its shares?  Many people thought this was the trend of the future.  Why let investment banks underprice the issue and take such a big cut?  Well it turns out that when IPO auctions are available they usually end up abandoned:

We document a somewhat surprising regularity: of the many countries
that have used IPO auctions, virtually all have abandoned them. The
common explanations given for the lack of popularity of the auction
method in the U.S., viz., issuer reluctance to try a new experimental
method, and underwriter pressure towards methods that lead to higher
fees, do not fit the evidence. We examine why auctions have failed and
verify, to the extent possible, that they are consistent with what
academic theory predicts. Both uniform price and discriminatory
auctions are plagued by unexpectedly large fluctuations in the number
of participants. The free rider problem and the winner’s curse hamper
price discovery and discourage investors from participating in
auctions. That may explain the inaccurate pricing and poor aftermarket
performance of IPOs using auctions.

Here is the link and paper, by Ravi Jagannathan and Ann Sherman.  Here are free copies of the paper.  The explanation is not complete, because the auctions often provide a superior price.  Nonetheless the risk is greater plus buyers are more likely to abandon the market in subsequent periods.  This article is long and not always fun to read, but it definitely falls into the "I learned something today" category.