While standard theory assumes rational, optimizing agents under full information, the latter is rarely found in reality. Information has to be acquired and processed—both involving costs. In rational-inattentiveness models agents update their information set only when the benefit outweighs the information cost. We test the rational-inattentiveness model in a controlled laboratory environment. Our design is a forecasting task with costly information and a clear cost–benefit structure. While we find numerous deviations from the model predictions on the individual level, the aggregate results are consistent with rational-inattentiveness and sticky information models rejecting simpler behavioral heuristics.
- Rational inattentiveness