In groups of any size, students are bidders in a common value English auction. Each bidder knows only her own signal about the common value. Bidders can bid at any price that is higher than the current high bid. When the auction ends, the bidder with the highest bid wins and pays her bid.
This game teaches students why bidders are susceptible to the winner's curse in common value auctions. It can also illustrate how to mitigate overestimation by appropriately shading one's bid. Using this game, you can explore many real-world scenarios including bidding for oil field leases, spectrum licenses, or IPO offerings.
Shows why bidders are susceptible to the winner's curse.
Illustrate how to mitigate overestimation of a valuation.
Each bidder receives a signal of the value for the unit for sale. The true value of the unit is drawn from a uniform distribution with default endpoints Lowest Common Value = $10 and Highest Common Value = $100. The Signal Noise is the most by which the bidder's value signal can deviate from the true common value. The noise is drawn from a uniform distribution of [-Noise, Noise]. If the Noise puts a player's common value signal beyond the Lowest/Highest Common Value, the common value signal is truncated.
Each bidder draws a new signal each period if Periods > 1.
Bidders can bid at any time in the auction so long as their bid improves the previous bid by the Price Increment.
We first present a table summarizing bidding behavior and outcomes (Figure 1). We display the number of groups, average profit, percentage of bids greater than to common value, percentage of bids greater than the equilibrium bid, and the percentage of auctions won by the highest signal holder.
We next present a series of graphs summarizing participant bids by the common value signal they received (Figure 2). A line showing the equilibrium bid based on common value signal is also included for reference. It is common to see bids above the common value (somtimes considerably above), illustrating the Winner's Curse. With repeated iterations of this game, students learn to reduce this effect.
Finally, we present a graph showing aggregate bids by common value signal across groups, normalized by the common value of the auction item. You can toggle which group's results are shown on the graph. Click on the check box next to a group name in the graph legend to remove that group's bids.
Our robot (i.e., an automated player) strategy is the following:
The robot estimates its value by picking a random number from a normal distribution between their minimum and maximum value from their test drill.
If the current high bid is lower than their estimated value, the robot will increase the bid to a random number between the highest bid and a number halfway between their estimated value and the highest bid.
Suppose the robot's test drill said the value was $65 +/- $10. And the robot then picks $70 as it's value from a normal distribution that ranges from $55 to $75.
If the current bid is $50 and the robot's estimated value is $70, the robot will raise the bid to a random number between $51 and $60.