Activity

e3c

Emergence of herding in an experimental environment: A counterfactual to the rationality hypothesis

Join the seminar at https://loyola.webex.com/meet/rede3c

Herding emerges when people take sequential decisions and they imitate the behavior of their predecessors. Such behavior usually comes up because agents, when traying to take the best decision, rule out their private signals and tend to believe that the others have better information. In his seminal paper, Banerjee (1992) explained herd behavior by proposing the example of choosing a restaurant. In that paper, the baseline assumption was that restaurants were of different qualities. The main result is that people converge, by means of imitation, to the worst option in the market. However, what if the restaurants – or, generally speaking, assets – were of the same quality? In this paper we investigate the emergence of herding behavior. In this paper we investigate the emergence of herding behavior in an experiment in which subjects have to choose where to eat among two  restaurants. We study under what conditions an imitative effect emerges in the choice of the restaurant. Differently from the standard literature on herding, where the two restaurants have different quality, we hypothesize that they have the same quality (either high or low). We design two treatments, one with just a private signal, and one in which we introduce an additional public signal. As expected, both signals have a positive effect on the choice of the venue. Moreover, even the queue has a significant and positive impact on the choice of the restaurant in all scenarios. Interestingly enough, however, our results show that if both restaurants are of low quality, the subjects tend to underestimate the effect of the signals and overestimate the queue, thus generating a strong herding. This result resembles a very well-known historical phenomenon consisting in agents abandoning the path of «rationality» which during negative periods and following the masses.

Keywords: confirmation bias, Experiment, herding