Rationality under different types of information
The research topic I want to study relates to how economic agents behave in a bargaining situation with asymmetric information. Over the next few months, I will be designing an experiment, similar to an auction, where players are randomly assigned values for a good, and then bargain for it as either seller or buyer. In particular, I will be looking at how far the outcomes of the game deviate from the game’s theoretical equilibrium, i.e. how rationally the players behave, and whether players act with different ‘degrees’ of rationality when possessing perfect or asymmetric information. Overall, the research question I aim to answer is: ‘How do economic agents act under different sets of information?’
There are a variety of economic theories to explain how agents behave in games when they have different sets of information. Slight changes in some of the existing models might provide a suitable basis for my own experiment. The book ‘Fun and Games: A Text On Game Theory’ by Kenneth Binmore gives a general overview of the theories behind games with asymmetric information. Binmore also explores the difference between games with complete and incomplete information but explains that equilibria can still be found for both types of games.
One issue with experiments like this one is the revelation problem, which is that, for the experiment to work, the design of a game must ensure that economic agents reveal their actual value for a good. This incentive-driven aspect is well explained by the book ‘Microeconomic Theory’ by Andreu Mas-Colell, who also evaluates remedies for the revelation problem. For my own experiment, one particular paper stands out as a choice for a basis: ‘Bargaining Under Asymmetric Information’ by William Samuelson. This paper establishes a model for an asymmetric information auction, which describes the equilibrium strategies for sellers and buyers when both parties are able to design the bargaining mechanism. This paper’s model is relatively easy to translate into an experiment but might need to be modified to find results that are more closely related to my research question. Overall, the existing literature provides a good basis for me to design my own experiment.
Experiments in this area of economics are applicable to a variety of real-life bargaining situations. Gosnell and Tavoni (2017) created a Nash bargaining game, where players had to choose costly strategies to prevent their wealth being affected by a simulated change of climate. The researchers assigned different starting points for endowments: in some iterations, everyone had the same initial wealth. In other iterations, some participants had larger endowments than others. They found that information, here complete but varying, does impact the result of a bargaining game.
Another example of how the degree of information affects the outcome of a game is from Berninghaus, et. al. (2012), who designed a coordination game experiment. He tested whether a player’s propensity to engage in corrupt behaviour changes depending on the level of information they receive beforehand. He found that more uncertainty (less information about the situation) reduces the likelihood of corruption. However, little experimental research has been conducted about the consequences of altering the type of information in a bargaining situation, rather than the degree of information.
This leads to my motivation for this research project. In my view, the assumptions made in many economic models that perfect, complete information exists in day-to-day situations is strongly questionable. Hence, I am excited to see how my research can enrich existing economic literature.