Aarhus University Seal / Aarhus Universitets segl

Scientific Background

Focus Areas

CFEM focused on three scientific areas that were essential to the design and implementation of electronic markets:

  • Game Theory and Mechanism Design

  • Operations Research and Algorithmics

  • Cryptography

Results

In the research dimension, CFEM produced new research results building on methods in economic theory allowing existing and emerging electronic markets to be better analyzed and understood. Further, we designed new trading mechanisms that suited better the needs of existing and new forms of e-commerce. And finally, we developed new results in algorithmics, operations research and cryptography allowing such mechanisms to be implemented efficiently and securely.

From the application point of view, the center produced a number of concrete systems for doing electronic commerce in collaboration with our industrial partners. This ranged from prototypes to full scale productions systems, and included systems for auctions, procurement, market regulation and cost allocation.

The background

We saw a very rapid development in electronic commerce. Traditional forms of trading, such as auctions and procurement, moved to electronic implementation on the Internet, and something similar could be expected to happen for systems that regulate markets in cases where monopolies threaten normal competition, for instance in areas like water and power supply. Moreover, completely new forms of electronic markets emerged, such as Google's auction system allocating advertisement slots on web pages, electronic systems for allocating costs of network traffic, prediction and information markets, and advertisement markets in the setting of social networks.

Scientific disciplines that study the theory behind such systems are game theory and mechanism design. Very briefly, game theory provides a general theory about what will happen when rational agents interact while maximizing their own payoff. Mechanism design is based on this theory and provides the means for designing games (such as an auction) where the predicted outcome optimizes some objective, e.g., the total benefit for everyone or the monetary returns to the market organizer.

It is well known and supported by several examples in practice that mechanism design based on game theory, for systems like those mentioned above, can play a huge role in maximizing social welfare and/or individual profit. However, in the new highly complex settings described above, the games and mechanisms arising were often not studied in standard economic literature, and solving the resulting computational problems when the mechanisms are implemented require state-of-the-art operations research and algorithmics. A second issue was that the predictions of game theory were based on assumptions  about which information is available to the parties at any given time. These assumptions may not always be justified, and if they fail, the system may not behave as predicted. For instance, if a participant in an auction obtains information about another participant's bid and intention too early in the process, this can be misused to force a result that is unfair or incorrect.

Secure Multiparty Computation

In response to challenges such as the above, the center combined computational theory with microeconomic theory to be able to analyze systems involving many agents. We also used cryptography to structure the interaction and available information to ensure better alignment with the game theoretic models. E.g., we used modern cryptographic techniques (secure multiparty  computation) that allow controlling exactly what information is available to a player and when, while still allowing efficient processing of the confidential data.

Some of the modern research in game theory, operations research, algorithmics and cryptography had already proven its value in practice. For every search result page shown by Google, an auction is run based on cutting-edge game theory to allocate advertisements to slots on the page. Closer to home, our industrial partner TradeExtensions, a world leader on implementing electronic markets, is using state-of-the-art algorithms from operations research and game theory for analysis of combinatorial first-price auctions. Even closer to home is the SIMAP project, run by some of the partners behind this center. The project has implemented a Danish nation-wide electronic market where farmers trade production rights for sugar beets. An in-depth economic analysis was done on the design and consequences of using this so-called double auction, while the implementation was based on secure multiparty computation to keep bids in the auction confidential. Running this system has now been taken over by our industrial partner Partisia.

Summing up, while cutting-edge game theory and related science is being applied to electronic markets in a few isolated cases, this is primarily a witness to the potential that exists. In general, there is a gap between state-of-the-art of research on electronic markets on one side, and the methods currently applied in practice on the other. The mission of the our center will be to bridge this gap.