day before the final registration statement was approved, 143 contracts traded. There was no discernable
trend in prices. The lowest normalized closing price for the IPO_UP contract was $0.248 and the highest
was $0.375, implying a forecasted market capitalization of $24.8 to $37.5 billion. On August 18, the date
the prospectus was declared effective, trading volume was 228 contracts and the normalized closing price
was $0.267 implying a predicted market capitalization of $26.7 billion. While the capitalization
according to the August 18th IPO price was considerably below this (23.1 billion), Google’s market
capitalization at the open on August 19th was 27.1 billion. It closed at a market capitalization of $27.2
billion (resulting in contract payoffs of $0.272).
Trading in the Google WTA market was much heavier than in the linear market.24
From July 8
through August 17, 3,021 contracts traded. Figure 2 shows prices of the WTA contracts as an area chart.
Each band corresponds to one contract. The width of the band is the normalized price of the contract.
Each contract price is interpreted as the probability that Google’s market capitalization would be within
the associated range (in billions of dollars) after the first day of trading. The sum of normalized prices
(forecast probabilities) equals 1. The actual first-day, closing market capitalization of Google was $27.2
billion. Figure 2 shows that the median of the predicted distribution was in the range corresponding to the
actual market capitalization from August 8 through the end of the market on August 17.
As news came out, various IEM contracts changed in price. Late in the market (around August
10), IPO_25-30 and IPO_30-35 emerged as the most likely outcomes and the median of the distribution
fell in the 25-30 billion range (as shown in Figure 2). On August 18, the volume of trade on the IEM
G o o g l e W T A m a r k e t w a s 3 , 1 4 8 c o n t r a c t s . P r i c e s c o l l a p s e d t o l e s s t h a n $ 0 . 0 5 f o r a l l b u t t h e I P O _ 2 0 - 2 5
a n d I P O _ 2 5 - 3 0 c o n t r a c t s a n d m o s t q u e u e s w e r e c l e a r e d .
typically relies on higher volume markets with thick queues in the argument for efficiency (e.g., Berg, Forsythe and Rietz, 1996). However, experimental research suggests that even small double auction markets (e.g., with as few as four traders) can converge to efficient outcomes (e.g., Smith, Williams, Bratton and Vannoni, 1982). Further, IEM prediction markets are similar to those modeled theoretically by Milgrom and Stokey (1982). We should see no trade according to their theory if traders have concordant preferences and are risk averse (which would make holding only cash and unit portfolios a Pareto optimal distribution). In this case, shadow prices would, nevertheless, be efficient. This trading pattern also holds in our political markets, with much heavier trading in WTA contracts than in linear (vote share) contracts. See Berg, Nelson and Rietz (2003).