Information cascades with endogenous signal precision

Nick Feltovich

Information cascades rely on individuals' own information being sufficiently imprecise that they ignore it, in favor of information inferred from others' observed behavior. This paper extends the standard information cascade model to allow individuals to choose how precise they want their information to be, when additional precision may be costly. There are two main results. First, free riding is rampant, due to the public-good aspect of information in this environment. Either none or nearly none of the individuals purchase additional information, even if it is almost costless. Second, reducing the cost of additional information may not result in more information obtained; indeed, it may lead to increased prevalence of information cascades and lower efficiency.

Paper (PDF)