Strategic coordination in forecasting - An experimental study

While reputational herding has been shown to contribute to poor economic forecasting, the underlying behavioral mechanisms have not yet been empirically investigated. We run a forecasting experiment with contradictory incentives for accuracy and coordination, finding subjects' forecasts to be inaccurate and driven by the coordination motive. Coordination is achieved through the salient, risk-dominant equilibrium, i.e. merely forecasting the current values. Subjects succeeding in coordinating earn significantly more than those striving for accuracy. Our results emphasize that reputational herding should be considered as a driving force for persistently poor prediction accuracy and systematically biased forecasts towards consensus values.