The European Central Bank published Working Paper No 3054 by Bartosz Maćkowiak and Mirko Wiederholt that embeds an information provision experiment into a standard dynamic rational inattention framework to explain how survey respondents update beliefs when presented with information. The paper derives analytical predictions for how the estimated “treatment effect” in survey randomized control trials varies with the environment, attention costs and the perceived importance of being informed, and discusses what survey-based responses imply about behaviour outside the survey and for communication policy. In the benchmark case where individuals pay positive attention both in daily life and during the survey and the underlying state is persistent, the model predicts the treatment effect is strictly decreasing in the importance of being informed and in the variance of shocks to the state, while increasing in the marginal cost of attention outside the survey for a given relative reduction in attention costs during the survey. The paper also characterises corner cases: if individuals do not pay attention outside the survey but do during the survey (for example because information is unavailable or much more costly to process outside the survey), the treatment effect is strictly increasing in the importance of being informed; if post-survey attention drops to zero for a period, the relationship can be ambiguous. Using these results, the authors link smaller treatment effects in high-inflation settings to higher perceived stakes and/or volatility, and suggest that communications that restate already-public information may have less impact on those who already find it most important to be informed, whereas expanding the set of publicly available information can change that pattern.
European Central Bank 2025-05-12
European Central Bank publishes working paper modelling how rational inattention shapes information treatment effects in survey RCTs
The European Central Bank's Working Paper No 3054 integrates an information provision experiment into a dynamic rational inattention framework to analyze how survey respondents update beliefs. It predicts variations in the "treatment effect" based on attention costs, perceived importance of information, and environmental factors, with implications for communication policy. In high-inflation settings, smaller treatment effects may result from higher perceived stakes and volatility, and expanding publicly available information could alter these effects.