The Power of Sunspots: An Experimental Analysis

Abstract

Abstract:We present an experiment in which extrinsic information (signals) may generate sunspot equilibria. The underlying coordination game has a unique symmetric non-sunspot equilibrium, which is also risk-dominant. Other equilibria can be ordered according to risk dominance. We introduce salient but extrinsic signals on which subjects may condition their actions. By varying the number of signals and the likelihood that different subjects receive the same signal, we measure how strong these signals affect behavior. Sunspot equilibria emerge naturally if there are salient public signals. Highly correlated private signals may also cause sunspot-driven behavior, even though this is no equilibrium. The higher the correlation of signals and the easier they can be aggregated, the more powerful they are in dragging behavior away from the risk-dominant to risk-dominated strategies. Sunspot-driven behavior may lead to welfare losses and exert negative externalities on agents, who do not receive extrinsic signals.

Believe it or not: Experimental evidence on sunspot equilibria with social networks

“…Marimon et al (1993), and Duffy and Fisher (2005) are two early examples: both of these papers provide convincing evidence that, at least in some contexts, sunspot shocks do matter. Heinemann et al (2012) also find evidence of sunspot equilibria in the presence of noisy sunspot signals. Agents are able to coordinate on an outcome when sunspot shocks occur in each of these games.…”