Bargaining Theories Application
http://www.cis.upenn.edu/~mkearns/papers/behbargain.pdf
Many theories of bargaining splits and shares have been made in the fields of computer science and economics. In this research paper, a research team from the University of Pennsylvania explores the reality of these splits and whether these theories are applicable in real-life bargaining.
These experiments are among the largest on network effects. They involve 36 subjects in each of 54 different experiments. These 54 scenarios consisted of 3 trials of 18 different scenarios. Each has a high level of uniqueness from the others, but all consist of 36 nodes. The definition of a basic interaction is between two players, and how they decide to share some payment on an edge between them. The analysis of these decisions can come from many different angles, including social welfare, human differences, and more. These experiments attempted to answer the questions of relationships between deal limits and wealth, as well as the effects of network topology on performance.
Many theories suggest universal patterns that deals should follow. Many propose that an outcome must be stable, in that every player’s earning is more than its outside option. It should also be balanced so that both parties have equal additional benefit from this edge, measured by the amount a player exceeds its outside option. The paper also discusses the ideas of Kleinberg and Tardos of Cornell, who suggest that not every network can achieve a stable and balanced outcome.Many of these ideas follow from the Nash Bargaining Slution discussed in class.
The results of these experiments are summed up as follows. More than 1/3 of deals are equally shared, which means the majority of people are self-interested and avert towards inequality. Some deals even resulted in one node getting 0 of the edge’s value. However, this wasn’t common as people are not typically willing to get fully taken advantage of, as discussed in our Networks class. Many of the findings are related to existing theories.
Deeper findings indiciated that humans are usually effective at playing these games, but that some pay a large amount for refusing to make deals. Many make decisions that aren’t maximal or beneficial. After the event a node doesn’t make a deal or can’t agree on a split, the network is overall lowered in efficiency. Overall, the paper concludes that the idea of “rationality” in bargaining theories may need to be adjusted and further explored.