Investigating Herding Behavior in an Experimental Setup

An “information cascade” occurs when initial decisions coincide in a way that it is optimal for each of the subsequent individuals to ignore her or his private signals and follow the established pattern. Moscarini et. al (1996) developed a social learning model hitherto unexplored in an experimental setting. They considered the consequences on learning if the state of the world is changing stochastically over time. Such a changing world scenario has many parallels in the functioning of financial markets where agents with incomplete information pass on signals and there is always a chance that the underlying world may change between time periods. Using the experimental methodology developed by Anderson and Holt (1997), the theoretical propositions were put to test. The broad results from this experiment were not inconsistent with the theory. Further extensions for new research are also discussed