X hits on this document

PDF document

Searching for Google’s Value: - page 6 / 44





6 / 44

market design allows us to build a forecast distribution of post-IPO prices, not just point estimates of the

expected post-IPO price. As a result, our markets show both the amount of uncertainty surrounding the

future post-IPO market price and the degree to which this uncertainty is resolved as the IPO process

evolves. This is an important contribution that is left unaddressed by the when-issued research. Third,

since our results are on the Google IPO with its unique auction design, we provide additional evidence on

what models are likely to explain underpricing.5

The rest of the paper is organized as follows. In Section 2, we outline the history and unique

features of the Google IPO. In Section 3, we describe the prediction markets we conducted to predict the

post-IPO Google value. In Section 4, we present our results and we conclude in Section 5.

  • 2.

    The Google IPO

    • A.


The Google initial public offering (IPO) was closely watched and unique. A search of

Lexis/Nexis for the words “Google” with “IPO” or “initial public offering” within 25 words, yields 769

hits between October 24, 2003 (when the potential for an IPO was first mentioned in the Wall Street

Journal) and August 19, 2004 (when trading in the stock began). Google’s use of an auction mechanism

to help set the IPO price is uncommon in the U.S., especially for an IPO of Google’s size. The stated goal

was to set an IPO price close to the ensuing market price. While there is debate over whether auction

mechanisms mitigate underpricing (see Sherman, 2004, for example), evidence from the French stock

market (Derrien and Womack, 2003) suggests an auction mechanism could have helped Google achieve

its goal. However, Google’s IPO price fell short of both its opening and its closing market prices on the

first day of trading by just over 15%, an amount close to the average initial underpricing of 15.3% for

U.S. IPOs reported by Jenkinson and Ljungqvist (2001, p. 27) and somewhat higher than the 11.9%


There are several other smaller distinctions as well. For example, we incorporate a variety of prediction market design features that encourage accurate price forecasts. The small size of our markets and evidence from other IEM markets lead us to believe that hedging and strategic manipulation are unlikely to bias prices in our prediction markets. Short selling is limited in when-issued markets. In prediction markets, synthetic short selling is constrained only by the budgets of traders.


Document info
Document views135
Page views135
Page last viewedTue Jan 17 02:54:28 UTC 2017