|
Abstract:
|
In competitive financial auctions (where lenders bid on interest rates for loans, for example), there are often multiple bidders tied for the lowest rate allowed by law. In these situations, the auction winner is chosen at random from amongst the tied bidders. Given a large list of auctions, bids, and winners, the distribution of wins across the bidders takes on a complex multinomial character, making traditional goodness-of-fit measures difficult to obtain. Using parallel simulation, we attempted to determine if the tie-breaking mechanism was consistent with random selection, whether the auction was "fair", how many bidders the auctioneer can expect to be treated "unfairly" by each series of auctions, and effective methods for the auctioneer to "cheat" without being caught by our analyses.
|
