Start Date
11-12-2016 12:00 AM
Description
Combinatorial exchanges are nowadays being used in day-ahead energy trading and other high-stakes markets. Linear and anonymous competitive equilibrium prices are desirable in multi-object auctions, but unfortunately such prices typically do not exist in combinatorial exchanges. In spite of this, day-ahead energy markets compute linear and anonymous prices at the expense of allocative efficiency. Anonymity and linearity of prices are also a requirement in other combinatorial exchanges, but such markets are not well understood. We discuss the market design for a large-scale combinatorial exchange for fishery access rights. The market allows for different allocation and payment rules. We analyze trade-offs of these rules with respect to efficiency loss incurred and computational hardness. Via analytical models and numerical simulations, we show that these losses can be up to 100\% in worst-case scenarios, but that they are small on average in larger markets.
Recommended Citation
Bichler, Martin; Fux, Vladimir; and Goeree, Jacob, "Linear payment rules for combinatorial exchanges" (2016). ICIS 2016 Proceedings. 1.
https://aisel.aisnet.org/icis2016/EBusiness/Presentations/1
Linear payment rules for combinatorial exchanges
Combinatorial exchanges are nowadays being used in day-ahead energy trading and other high-stakes markets. Linear and anonymous competitive equilibrium prices are desirable in multi-object auctions, but unfortunately such prices typically do not exist in combinatorial exchanges. In spite of this, day-ahead energy markets compute linear and anonymous prices at the expense of allocative efficiency. Anonymity and linearity of prices are also a requirement in other combinatorial exchanges, but such markets are not well understood. We discuss the market design for a large-scale combinatorial exchange for fishery access rights. The market allows for different allocation and payment rules. We analyze trade-offs of these rules with respect to efficiency loss incurred and computational hardness. Via analytical models and numerical simulations, we show that these losses can be up to 100\% in worst-case scenarios, but that they are small on average in larger markets.