The stability of information cascades: how herd behavior breaks down

We extend the standard model of information cascades to situations where agents have to choose between switching to a new alternative or not. In particular, we add two new features to the standard model. First, agents continue to receive signals after they have chosen 'switching'; and second agents may revert their earlier decisions. The focus of the paper is to analyze how information is passed on within an information cascade, and the conditions under which breakdown occurs. We identify rules which describe the learning process and run simulations to estimate the properties of the information cascade.