您的浏览器禁用了JavaScript(一种计算机语言,用以实现您与网页的交互),请解除该禁用,或者联系我们。[欧洲央行]:随机对照试验中的理性疏忽 - 发现报告

随机对照试验中的理性疏忽

随机对照试验中的理性疏忽

Working Paper Series Abstract We introduce an information provision experiment into a standard dynamic rational inat-tention model.We derive analytical results about how the treatment effect varies with char-acteristics of the environment and the individual.We use these results to discuss findings inthe empirical literature on information provision experiments that can be explained by rationalinattention of survey respondents and what this interpretation implies about behavior outsidethe survey. Keywords:rational inattention, information provision experiment, randomized control trial(JEL: D8, D9, E7). Non-technical summary In surveys with information provision experiments, one can observe how people change theirbeliefs, and sometimes also actions, after having been confronted with information. How can weexplain findings from information provision experiments? What do these findings tell us abouteconomic behavior outside the survey? What are the implications for communication policy? As a step toward answering these questions, this paper studies what the theory of rationalinattention predicts about information provision experiments. “Rational inattention” is the ideathat information is in principle available, but absorbing information requires paying attentionwhich is costly. We introduce an information provision experiment into a standard dynamicrational inattention model. We derive analytical results about how the treatment eƯect – howmuch people respond to information presented during a survey – depends on the characteristicsof the environment and the individual. We prove that the treatment eƯect during an information provision experiment is strictlydecreasingin the importance of being informed, so long as agents pay positive attention outsideand during the survey. When being informed becomes more important, two counteracting eƯectsarise: people want to pay more attention in daily life, which lowers the treatment eƯect; andpeople want to pay more attention during the survey, which raises the treatment eƯect. Theattention-in-daily-life eƯect dominates. Next, we consider the case when agents pay attention during the survey but not in daily life(because the cost of paying attention in daily life is much higher than during a survey, or therelevant information is simply unavailable outside the survey). In this case, the treatment eƯectis strictlyincreasingin the importance of being informed. The reason is that the first of the twoaforementioned eƯects, the attention-in-daily-life eƯect, is absent. Hence, the nature of theinformation – whether the information was available to respondents before the survey or not –aƯects the interpretation of the findings. We use our results to interpret findings from information provision experiments in the literature.As an example, the literature finds that the treatment eƯect is smaller when inflation is high. Ourinterpretation of this finding is that the importance of being informed about inflation and/or thesize of inflationary shocks increase with inflation, and such increases reduce the treatment eƯectin the model. The literature also finds that individuals who are less informed about monetarypolicy revise beliefs by more after an information treatment. Our model-based interpretation isthat these are individuals for whom monetary policy is relatively unimportant in daily life. Our analysis suggests that communication policy may have less eƯect among individuals whocare more about being informed and at times when being informed is more important, so long ascommunication involves merely restating information that has already been publicly available. Ifcommunication involves expanding the set of publicly available information, then it may be moreeƯective among individuals who care more about being informed and at times when beinginformed is more important. 1Introduction Surveys with information provision experiments, a form of randomized control trials (RCTs), havebecome popular in economics. Naturally, economists are interested in comparing the informationtreatment effects, on beliefs and on actions, measured in this literature with theories of behaviorunder incomplete information or bounded rationality, including rational inattention (RI) introducedby Sims (2003). This comparison is complicated because RI implies optimal attention choicebeforeandduringa survey. The goal of this paper is to make this comparison easier.1 As a motivation, consider a basic research question in information provision experiments: Doesone expect a larger or a smaller treatment effect when respondents care more about being informed?It is intuitive that when being informed is more important, people pay more attention in their dailylife, and consequently they enter a survey with a sharper prior, which reduces the treatment effect.But it is likewise intuitive that when being informed is more important, respondents have moreincentive to absorb information presented to them during the surv