Stochastic Deadlines

A Comparison of Parallel Multiple Auction Designs

Abstract

In online auction platforms, offers are listed side by side and may end at the same point in time. While theoretical studies predict efficient coordination across auctions, experimental and empirical studies observe efficiency losses, i.e. goods remain unsold. In order to mitigate this coordination failure, we contribute to the literature of auction design by introducing a stochastic deadline in parallel multiple auctions. In these parallel Candle Auctions, several auctions start at the same time but end (separately) due to a stochastic process. We think that the stochastic ending rule decreases the coordination failure because the threat of a sudden termination forces the bidders to coordinate across auctions early in the auction process. Indeed, we find that coordination is less pronounced in parallel Candle Auctions resulting in higher efficiency