An experiment to investigate the externalities of search*

This paper presents an experiment to investigate the impact of information externalities in a search model. The model implies that information spillovers create an initial delay in action by agents, followed by a flurry of activity once some public information is revealed. Conversely, this experiment reveals a shorter delay prior to the first subject moving than the model suggests. Furthermore, the subsequent increase in activity is greater than predicted. However, as the sessions progress, the subjects exhibit a degree of learning in each treatment. They prolong their search for higher payouts and, hence, their behaviour approaches that of the model's optimal strategy.