Abstract: Andre et al (2022, ReSTUD) demonstrated significant heterogeneity in how households think the economy (i.e., inflation and unemployment ) respond to various economic shocks. An implication is that heterogenous mental models are a meaningful source of cross-sectional heterogeneity in macroeconomic expectations. We build upon this work by using a large-scale, online experiment to document heterogeneity in the subjective mental models households use to anticipate the central bank’s policy decisions (i.e. the reaction function) and how those policy decisions transmit to the economy (i.e. the transmission function). We then test the ability of various types of central bank communication to coordinate each of these three types of subjective mental models on the true data generating process underlying the economy, on the central bank’s true reaction function, and on the true underlying transmission function.
Experimental design
We use an information intervention to study our ability to coordinate mental models of the economy.
Design details
Randomization method
Randomization into treatment conditions is handled within our software via random number generation in Python
Randomization unit
We randomize at the individual level.
Clusters
identical to number of observations below
Observations
Roughly 1200 based on the ability to detect small-to-moderate treatment effects at the 1% level with power = .8. We randomize subjects into 1 of 4 treatment conditions. Within each treatment condition, subjects will see either demand or supply shocks. Thus, we randomize subjects into 8 bins. This means roughly 150 subjects per bin.
Arms
300 per treatment arm. When considering demand and supply bins, 150.
Primary outcomes
To what extent do these forms of communication influence a subject's belief about the true underlying model of the economy? To what extent can we coordinate subjective mental models (i.e., beliefs about the true assessment, reaction, and transmission functions)? How does the efficacy of our interventions along these dimensions relate to various observable charecteristics (general intelligence, education, previous economic training, etc.). Do people's perceptions of whether or not they've changed their beliefs about how the economy functions get reflected in their choice data? If not, why? What is the longevity of observed treatment effects (i.e., what do we see a month after treatment, two months? etc. )
Cited by
No citation data available for this registration yet.