Information, Interaction and Mechanism Design

In this Focus Group, Hans Fischer Senior Fellow Prof. Dirk Bergemann (Yale University) collaborates with his hosts Prof. Martin Bichler and Prof. Felix Brandt (Decision Sciences and Systems, TUM).

Understanding and modeling markets has always played a central role in Economics, and it can probably be described as the central topic in Economics. Rather than modeling markets from a bird's-eye view, mechanism design was a very successful strand of Economic theory where incentives of individual decision makers are modeled explicitly. A number of recent Nobel Memorial Prizes in Economic Sciences document the success of this stream of research and this specific way of modeling markets.
In the recent years, the formal study of mechanism design has found concrete applications in the real world. It led to market designs in various fields such as spectrum auctions, procurement tenders, or kidney exchanges where multiple independent decision makers need to be coordinated such that the outcomes is economically efficient and the mechanisms are robust to various forms of manipulation. Mechanism design and game theory also had a fundamental impact on computer science and operations research, where problems were typically modeles from the point of view of a central decision maker. Solving complex planning problems in revenue management, scheduling, or transportation has led to a substantial amount of research at the intersection of Computer Science, Economics, and Management Science in the recent years.

While mechanism design theory was incredibly successful as a formal way to model markets, its scope is also restricted. The overall project tries to advance the theory of mechanism design in three imporant ways:

  • traditional auction theory is based on the assumption of a common prior distribution among all bidders. Such assumptions can rarely be found in the field, which is why equilibrium strategies derived for simple first price auctions are often a poor guideline for practitioners. We draw on recent advances on robust mechanism design by Bergemann and Morris (2013) to develop alternative models for first price auctions.
  • mechanism design has typically focused on static, one-time decisions such as isolated markets where all bidders are present at the same time. Dynamic mechanism design focuses on a sequence of interdependent decisions. The classic airline revenue management problem is a nice example, where an airline must decide how to price seats on a flight over a certain time frame in response to changing inventory and customer demand. Market participants arrive and depart over time and both, the maximization of efficiency and the maximization of seller revenue in such environments is still an open issue. We focus on a continuous time perspective in this research rather than the discrete-time models which have been developed so far.
  • the network structure of interactions among decision makers has often been ignored in the theory of mechanism design. An explicit consideration of network structures can help better understand the volatility of actions of agents in financial markets.

TUM-IAS funded doctoral candidate:
Florian Brandl, Decision Sciences and Systems