first day of trading according to the closing market price.^{18 }

ii. The Winner-Takes-All Market

# The Google Winner-Takes-All (WTA) market opened on June 29, 2004 with six “interval”

contracts.^{19 }Liquidation values of the initial contracts were determined as follows:

Contract IPO_0-20

IPO_20-25 IPO_25-30 IPO_30-35 IPO_35-40 IPO_gt40

Contract Liquidation Values $1 if market cap is less than or equal to $20 billion or if the IPO does not occur by March 31, 2005. $1 if market cap is greater than $20 billion but less than or equal to $25 billion. $1 if market cap is greater than $25 billion but less than or equal to $30 billion $1 if market cap is greater than $30 billion but less than or equal to $35 billion $1 if market cap is greater than $35 billion but less than or equal to $40 billion $1 if market cap is greater than $40 billion.

# On August 5, the IPO_gt40 contract was split into three contracts: IPO_40-45, IPO_45-50 and IPO_gt50

each with a $1 payoff in the associated capitalization range.^{20 }At the split, traders holding IPO_gt40

contracts received 1 share of each of the three new contracts in exchange for each IPO_gt40 contract they

held so that they incurred neither a gain nor loss in expected value from their previous portfolio position.

# Again, in the absence of hedging demand, prices should equal expected values in this market (see

footnote 17). Expected value pricing implies that the price of each contract should equal the probability

that the actual market capitalization will be in the associated capitalization range (E(value) =

p×$1+(1-p)×$0 = p, where p is the probability of being in the range). Thus, at each point in time prices

map out discrete parts of a forecast distribution for the future market capitalization. From this

distribution, we can estimate the expected post-IPO valuation of Google and obtain a direct measure of

the ex ante uncertainty surrounding this forecast.

18

19 20

Technically, we need two further assumptions to make this the forecasting relationship. We need to assume that the probability of no IPO before March 2005 is zero, which is consistent with Google’s stated strong intention to issue in the summer of 2004. We also need to assume that the probability of a market capitalization greater than $100 is effectively zero. Below, we will estimate a distribution of expected market capitalizations from the other IEM market we ran. This distribution is consistent with essentially zero likelihood of a market capitalization above $100 billion. The appendix contains the prospectus for this market. This was done because of sustained high prices for the IPO_gt40 contract. It was intended to expand the price ranges covered by contracts to more closely match the apparent range of potential outcomes forecast by our traders.

12