from Google’s prospectus and information released by investors after the IPO. This will provide evidence
about whether the prices in our prediction market could have been market clearing IPO prices for Google.
3. The Iowa Electronic Markets Google IPO Markets
Though other markets have a predictive component (e.g., futures markets), prediction markets
are designed specifically for forecasting purposes. Contracts in prediction markets have payoffs tied
directly to a future event of interest (in this paper, Google’s eventual market capitalization) and the
markets have design features that encourage revelation of true underlying expectations. Prices in
prediction markets provide forecasts about features of the associated event, for example its probability of
occurring or the consequences of its occurrence. The most well-known prediction markets are the Iowa
Electronic Markets (IEM for short, reviewed in Berg, Forsythe, Nelson and Rietz, 2003), which have been
used for more than 17 years to forecast election outcomes, other political and economic events, prices and
returns of stocks, corporate earnings and movie box office receipts. These real-money, small-scale
markets have proven remarkably accurate in the short run (Berg, Forsythe, Nelson and Rietz, 2003) and
the long run (Berg, Nelson and Rietz, 2003).
A. Description of the Google IPO Markets
The IEM conducted two markets associated with the Google IPO. Both markets traded contracts
with liquidation values based on the total market capitalization implied by the closing price of Google
stock at the end of the first day of trading. Contracts were based on total market capitalization rather than
share price so that the markets could open before initial price ranges and share quantities were announced.
IPOs had fees of 4%. So, while fees for smaller IPOs typically average 7%, the fee here seems in line after considering IPO size. The auction may have allowed Google to more accurately assess demand and avoid the costs associated with over-allotment options. However, Google’s over-allotment option was exactly 15%, the “typical” amount in the U.S. according to Brealey and Myers (2003, p. 413).