Information rigidities in survey data: Evidence from dispersions in forecasts and forecast revisions

Predictable forecast errors in survey data documented in the existing literature suggest a deviation from the rational expectations hypothesis and are in favor of imperfect information models such as sticky and noisy information models. This article assesses the validity of the imperfect information models by establishing a linkage between dispersions in survey forecasts and survey forecast revisions. We find that the dynamics of dispersion in survey forecasts are consistent with the prediction of sticky information models, but at odds with that of conventional noisy information models as well as full information rational expectations models, both of which assume agents' continuous updating of their information sets.