Applying Parallel Genetic Algorithms to Economic Problems: the Case of Agricultural Land Markets
By Alfons Balmann and Katrin Happe
ABSTRACT
This paper elaborates the use of distributed Genetic Algorithms (DGA) to study an artificial land rental market. The study is based on a spatial comparative-static model in which a number of spatially ordered agents (farms) compete in an auction for renting land. Each agent's behavior is determined by a genetic algorithm that is applied to an agent specific population of genomes representing particular bidding strategies. Agents interact directly through a migration mechanism that allows to spread renting strategies across the population of agents as well as indirectly over the rental market. Two market constellations are considered and different simulations with a variety of parameter constellations (migration rate, placement of farms, etc) are run: First, a situation of limited market access is defined. A series of simulation experiments shows that for this scenario the DGA generates results that fit comparative static equilibrium conditions like allocative efficiency and zero-profits. Second, in a limited market access scenario, only under very special conditions the DGA generates results that comply with oligopolistic behavior. The results of the two scenarios are analyzed and discussed as to the influence of the DGA procedure itself and a possible economic and game theoretic interpretation.
KEYWORDS: genetic algorithms, multi-agent systems, land market, game theory, market power
 View Innovations in Microeconomics: Microbehavior and Macroresults Session
View Full Paper (PDF file)
View Powerpoint Presentation (PDF file)
|
